Loading [MathJax]/jax/output/SVG/jax.js
Review

Exploring the challenges and opportunities of image processing and sensor fusion in autonomous vehicles: A comprehensive review

  • Autonomous vehicles are at the forefront of future transportation solutions, but their success hinges on reliable perception. This review paper surveys image processing and sensor fusion techniques vital for ensuring vehicle safety and efficiency. The paper focuses on object detection, recognition, tracking, and scene comprehension via computer vision and machine learning methodologies. In addition, the paper explores challenges within the field, such as robustness in adverse weather conditions, the demand for real-time processing, and the integration of complex sensor data. Furthermore, we examine localization techniques specific to autonomous vehicles. The results show that while substantial progress has been made in each subfield, there are persistent limitations. These include a shortage of comprehensive large-scale testing, the absence of diverse and robust datasets, and occasional inaccuracies in certain studies. These issues impede the seamless deployment of this technology in real-world scenarios. This comprehensive literature review contributes to a deeper understanding of the current state and future directions of image processing and sensor fusion in autonomous vehicles, aiding researchers and practitioners in advancing the development of reliable autonomous driving systems.

    Citation: Deven Nahata, Kareem Othman. Exploring the challenges and opportunities of image processing and sensor fusion in autonomous vehicles: A comprehensive review[J]. AIMS Electronics and Electrical Engineering, 2023, 7(4): 271-321. doi: 10.3934/electreng.2023016

    Related Papers:

    [1] A. D. Al Agha, A. M. Elaiw . Global dynamics of SARS-CoV-2/malaria model with antibody immune response. Mathematical Biosciences and Engineering, 2022, 19(8): 8380-8410. doi: 10.3934/mbe.2022390
    [2] A. M. Elaiw, Raghad S. Alsulami, A. D. Hobiny . Global dynamics of IAV/SARS-CoV-2 coinfection model with eclipse phase and antibody immunity. Mathematical Biosciences and Engineering, 2023, 20(2): 3873-3917. doi: 10.3934/mbe.2023182
    [3] Chentong Li, Jinhu Xu, Jiawei Liu, Yicang Zhou . The within-host viral kinetics of SARS-CoV-2. Mathematical Biosciences and Engineering, 2020, 17(4): 2853-2861. doi: 10.3934/mbe.2020159
    [4] Jinhu Xu, Yicang Zhou . Bifurcation analysis of HIV-1 infection model with cell-to-cell transmission and immune response delay. Mathematical Biosciences and Engineering, 2016, 13(2): 343-367. doi: 10.3934/mbe.2015006
    [5] Wenhan Guo, Yixin Xie, Alan E Lopez-Hernandez, Shengjie Sun, Lin Li . Electrostatic features for nucleocapsid proteins of SARS-CoV and SARS-CoV-2. Mathematical Biosciences and Engineering, 2021, 18(3): 2372-2383. doi: 10.3934/mbe.2021120
    [6] Yangyang Yu, Yuan Liu, Shi Zhao, Daihai He . A simple model to estimate the transmissibility of the Beta, Delta, and Omicron variants of SARS-COV-2 in South Africa. Mathematical Biosciences and Engineering, 2022, 19(10): 10361-10373. doi: 10.3934/mbe.2022485
    [7] Sumati Kumari Panda, Abdon Atangana, Juan J. Nieto . New insights on novel coronavirus 2019-nCoV/SARS-CoV-2 modelling in the aspect of fractional derivatives and fixed points. Mathematical Biosciences and Engineering, 2021, 18(6): 8683-8726. doi: 10.3934/mbe.2021430
    [8] Sarafa A. Iyaniwura, Rabiu Musa, Jude D. Kong . A generalized distributed delay model of COVID-19: An endemic model with immunity waning. Mathematical Biosciences and Engineering, 2023, 20(3): 5379-5412. doi: 10.3934/mbe.2023249
    [9] Tahir Khan, Roman Ullah, Gul Zaman, Jehad Alzabut . A mathematical model for the dynamics of SARS-CoV-2 virus using the Caputo-Fabrizio operator. Mathematical Biosciences and Engineering, 2021, 18(5): 6095-6116. doi: 10.3934/mbe.2021305
    [10] Matthew Hayden, Bryce Morrow, Wesley Yang, Jin Wang . Quantifying the role of airborne transmission in the spread of COVID-19. Mathematical Biosciences and Engineering, 2023, 20(1): 587-612. doi: 10.3934/mbe.2023027
  • Autonomous vehicles are at the forefront of future transportation solutions, but their success hinges on reliable perception. This review paper surveys image processing and sensor fusion techniques vital for ensuring vehicle safety and efficiency. The paper focuses on object detection, recognition, tracking, and scene comprehension via computer vision and machine learning methodologies. In addition, the paper explores challenges within the field, such as robustness in adverse weather conditions, the demand for real-time processing, and the integration of complex sensor data. Furthermore, we examine localization techniques specific to autonomous vehicles. The results show that while substantial progress has been made in each subfield, there are persistent limitations. These include a shortage of comprehensive large-scale testing, the absence of diverse and robust datasets, and occasional inaccuracies in certain studies. These issues impede the seamless deployment of this technology in real-world scenarios. This comprehensive literature review contributes to a deeper understanding of the current state and future directions of image processing and sensor fusion in autonomous vehicles, aiding researchers and practitioners in advancing the development of reliable autonomous driving systems.



    Over the past few years, the frequent occurrences of natural disasters in the world have caused serious damage to social construction and economic development, such as the flood disaster in India in 2013, the hail disaster in Yancheng in 2016, the 2019 hill fires in Australia and Hurricane Laura in the United States in 2020 [1,2,3]. In China, 69,227 people died in the "5.12" Wenchuan earthquake in 2008, with direct economic losses of $119.7 billion [4]. The "7.20" extremely heavy rain disaster in Zhengzhou in 2021 affected 14.786 million people, with direct economic losses reaching $17 billion [5]. Since most sudden natural disasters cannot be predicted in advance, strengthening the guarantee of emergency supplies after disasters has become the key to improving rescue efficiency. Therefore, in order to ensure the smooth development of rescue work, it is necessary to formulate a scientific and reasonable material emergency dispatching (MED) scheme, which can minimize the subsequent losses from the disaster.

    With the in-depth study of MED, the combinations of MED and actual scenes have become the main research direction. Therefore, an array of variant problems have been derived, such as the problem with time windows [6,7], the problem with multiple distribution centers [8,9], the multi-objective optimization problem [10,11] and the dynamic routing problem (DRP) [12]. Among them, the DRP has attracted high attention from researchers [13,14,15,16]. Psaraftis et al. [14] analyzed the impact of the new generation of information technology on DRP and summarized the research literature according to 11 classification standards. Reference [16] classified the DRP models into dynamic demand and dynamic road conditions and introduced the research progress of its route update strategies and solving algorithms. DRP is proposed relative to the static routing problem (SRP). In the SRP, relevant information such as distribution points, demand and road conditions are known and fixed before planning the distribution path, and the vehicle route will not be changed once it is determined. However, some of the information in the DRP may be changed, so its dynamics bring a great challenge in optimizing the material scheduling path. More specifically, due to the continuous impact of disasters and secondary disasters, the emergency situations that road blockage or destruction make impassable vehicles occur frequently in the disaster scenario, which makes it necessary to dynamically adjust the planned path according to real-time information during material scheduling. Therefore, it is more realistic to consider the "changing road conditions" in the study of emergency material dispatching. In addition, the essence of MED is a vehicle routing problem (VRP). As a classical combinatorial optimization problem, the VRP has been proved to be an NP-hard problem. Therefore, while paying attention to the construction strategy of the MED model, scholars have also conducted a lot of research on its solution algorithm. At present, there are mainly the exact algorithm [17,18,19] and the heuristic algorithm [20,21,22]. Among them, the exact algorithm establishes the corresponding mathematical model for a specific problem and uses mathematical methods to find the optimal solution. Although the exact algorithm can achieve better results in solving the problem, its computational time explosively increases with an increase of the problem size, which makes it only suitable for solving small-scale problems [23]. The heuristic algorithm can give a feasible solution for the combinatorial optimization problem within an acceptable range, which has better robustness and feasibility in dealing with large-scale problems compared to the exact algorithm. However, its solution performance is mostly limited by the hand-designed algorithm rules. Especially in the cases of single distribution centers, no time constraints and changes in road conditions, the stability of the heuristic algorithm is deficient [24,25].

    Although the research on "changing road conditions" has been widely carried out, finding high-performance algorithms to improve the execution efficiency, feasibility and reliability of the scheduling scheme is still an urgent practical demand problem in post-disaster material distribution. Therefore, considering the disadvantages of the traditional heuristic algorithm and the complexity of the material emergency dispatching with dynamic road conditions (MEDDRC) scheme, this paper proposes an improved whale swarm algorithm (IWSA) to solve the MEDDRC problem. The main contributions of this study are as follows:

    1) The improved scan Clarke-Wright (ISCW) method is designed to plan the initial driving route of vehicles, which can obtain the initial scheduling scheme with minimum cost.

    2) The group movement rule is proposed to update the individuals, which not only ensures that individuals in the population carry out a detailed local search of in different regions, but also enhances the global search capability of individuals.

    3) To avoid individuals prematurely congregating in a certain area during the iterative process, a different weights strategy is constructed to expand the individual search space, which is helpful to maintain the population diversity.

    4) The feasibility of the proposed algorithm in solving material emergency dispatch under dynamic changes in road conditions is verified through extensive experiments.

    The remaining sections of this paper are organized as follows. Section 2 provides a review of the relevant literature on MED in recent years. Section 3 develops a mathematical model for the MEDDRC problem. A detailed description of the proposed IWSA and the main strategies are shown in Section 4. Section 5 analyzes the experimental results of the algorithm. Our conclusions and future work are given in Section 6.

    As a key part of emergency management, the merits and demerits of the MED scheme will directly affect the smooth development of subsequent rescue work. For this reason, scholars have carried out a lot of research work around the MED. According to the different focuses of the research, the existing works mainly include two aspects of the model construction and solution.

    As the specific situations of disasters are different, the objectives and influencing factors considered in material dispatching are also somewhat different. Therefore, in order to meet the special needs in different disaster situations, researchers have constructed corresponding mathematical models for specific application scenarios. For example, in solving the single-objective problem, Sung and Lee [26] achieved maximizing the number of rescuers with limited resources by modeling patient priority as an ambulance scheduling problem. Balcik et al. [27] proposed a mixed integer programming model for studying the last mile distribution problem, which can reduce the transportation cost to maximize the benefit of the recipients. However, in the actual disaster relief operations, there is a lack of comprehensive consideration of multiple objectives. For this reason, Vitoriano et al. [28] constructed a multi-criteria optimization model for the MED problem by considering objective criteria such as time, cost, security and equity. Zhang et al. [29] constructed a multi-objective optimization model for multiple reserve points and multiple relief supplies, which can solve the difficult problem of the allocation-scheduling of emergency relief supplies. In addition, Chen et al. [30] presented a multi-objective optimization model with a utility–priority–economy allocation strategy, which can effectively solve the MED problem. However, due to the complexity of the situation in the disaster area and the uneven distribution of the locations of the affected points, it is difficult to meet the urgent time requirements of MED for a single rescue point. For this reason, Wex et al. [31] constructed a rescue unit allocation and scheduling model for multiple rescue points with the objective of minimizing the total completion time and obtained the optimal solution by using the heuristics algorithm. Meanwhile, Song et al. [32] introduced the emergency point satisfaction and materials lack loss coefficient and built an emergency material dispatching model with a multi-emergency point, a multi-rescue point and multi-stage. However, most of the above models only consider the influence of static factors, without considering the time-variant of road conditions and traffic environment, which will lead to the vehicle not being able to smoothly complete the material distribution tasks according to the existing scheduling scheme. Therefore, Qu et al. [33] took the timeliness and fairness of emergency material distribution as the goal and established a multi-objective optimization model for distribution path planning under multi-period dynamic conditions. In addition, the update strategies of the scheduling scheme such as a fixed time interval [34] and a key point [35] are often adopted. For example, Okhrin et al. [36] adopted the method of updating the status of the current road section at regular intervals to solve the problem of real-time vehicle routing; however, it may cause an inconsistency between the time of information update and the time of actual route change. Therefore, Li et al. [37] proposed a new method of updating the route at key points to solve the problem of dynamic network vehicle distribution, which was more effective compared to the traditional updating mechanism.

    In problem-solving, the quality of the scheduling scheme depends on the algorithm for solving the problem; a high-performance solving algorithm can significantly increase the possibility of obtaining an efficient and reasonable scheduling scheme. For this reason, researchers have conducted a lot of research on how to improve the performance of problem-solving algorithms. For example, Zhang et al. [38] used a local search technique and high-priority rule to design an algorithm based on linear programming to solve the optimal allocation of emergency resources, which can reduce the problem complexity by improving the linear programming solution after relaxation. Meanwhile, Queiroga et al. [39] proposed two branch cut and price (BCP) algorithms based on the backhaul vehicle scheduling problem, which incorporated advanced elements such as strong branching and route enumeration to reduce the computation time of the problem. Although the above work has achieved good results, the used algorithms have a high complexity in solving large-scale problems; therefore, most heuristic algorithms are used to solve such problems. For example, Xiang et al. [40] proposed the ant colony system (ACS) algorithm based on pairwise proximity learning to solve dynamic VRPs, which can predict the local visiting order of customers after the occurrence of changes according to the optimal routes found before the changes occur. Wang et al. [41] designed a multi-objective cellular genetic algorithm (GA) to optimize the post-disaster emergency resource allocation model, which achieved a balance between global and local optimization by introducing the auxiliary population and neighborhood structure in the cellular automata. In addition, Zhang et al. [42] proposed an improved particle swarm optimization (PSO) algorithm to solve the optimal scheduling problem with multiple objectives, which can increase the population diversity and expand the search range by adding perturbation operations to the update process of particle velocity. Although the heuristic algorithm has a high feasibility, it suffers from the disadvantage of falling into a local optimum prematurely, which makes it difficult to obtain an approximate optimal solution in the global scope. To this end, scholars adopted the whale swarm algorithm (WSA) [43] with an improved performance to solve the problem, and amended it on the basis of its superior iteration rules to enhance the global solution ability of the algorithm. For example, Dong and Ye [44] designed a new individual position movement strategy and a multi-neighborhood search strategy for WSA to solve the flow shop scheduling problem, which can balance the global and local search abilities of the algorithm to improve the solution performance. However, with the update and iteration of the population, the population diversity will rapidly decrease. For this reason, Zhang et al. [45] proposed a discrete WSA to solve the flow shop scheduling problem, and improved the individual movement rules of the WSA by constructing five mutation operators, which can maintain the diversity of the population and enhance the solving ability of the algorithm.

    In addition, the relevant literatures on model construction and solution algorithms for MED problems are listed in Table 1. Among them, the superior iteration rules of WSA make its performance better than the traditional heuristic algorithm. However, due to the discrete characteristics of the MEDDRC problem, the algorithm rule in the traditional WSA cannot well solve this problem. Moreover, during the iteration of the algorithm, the convergence of individual whale movements in the WSA makes the population individuals still gather quickly in a certain region of the solution space, which in turn limits the ability of individual whales to search for solutions in the global range. Therefore, in this paper, the algorithm rules of WSA will be improved in terms of the initial route generation strategy, the individual movement method and the search space expansion, which can adapt to the MEDDRC problem and enhance the effectiveness of the algorithm.

    Table 1.  Relevant literatures on the construction and solution for MED model.
    Research Objectives Solving Methods Application Scenarios Literature
    Maximizing the number of people rescued Column generation approach Medical resource allocation [26]
    Minimizing transportation costs; Maximizing benefits to aid recipients GAMS/Cplex Humanitarian relief [27]
    Maximizing transportation safety and distribution equity; Minimizing transportation time and costs GAMS/Cplex Humanitarian relief [28]
    Minimizing distribution time; Maximizing material satisfaction and reliability ACS、GA Emergency relief supplies distribution and dispatch [29]
    Maximizing total allocation utility; Minimizing distribution costs Goal programming、Fuzzy analytic hierarchy process Multi-depot and multiple affected area emergency
    logistics system
    [30]
    Minimizing the sum of completion times Heuristic algorithm Natural disaster relief [31]
    Minimizing total dispatching cost; Maximizing emergency point satisfaction PSO Nonlinear continuous consumption MED problem [32]
    Minimizing transportation time; Maximizing distribution equity GA Post-earthquake MED problem [33]
    Minimizing travel time GA Dynamic VRP [35]; [36];
    [37]
    Minimizing the rescue costs Linear programming Multiple-resource and multiple-depot MED problem [38]
    Minimizing distribution costs BCP VRP with backhauls [39]
    Minimizing total driving costs ACS Dynamic VRP [40]
    Minimize the disaster losses and the total risks of the selected routes in the rescue process GA MED for multi-period post-disaster [41]
    Minimizing distribution distance; Maximizing distribution equity PSO Emergency material allocation and vehicle routing plan generation [42]
    Minimizing completion costs WSA Flow shop scheduling problem [44]; [45]

     | Show Table
    DownLoad: CSV

    For the MEDDRC problem studied in this paper, due to the existence of a dynamic factor such as the change of road conditions, it is difficult for the fixed vehicle scheduling scheme to complete the distribution task of all disaster-affected points in the shortest time. Therefore, in order to describe the dynamic characteristics of MEDDRC and to consider the complexity of information received by the vehicle while driving, the concept of a "key point" [46] is introduced and divided into two levels of key points. The first-level key point is the location of the disaster-affected point, and the second-level key point is the intersection of two roads (that is, the intersections of paths to different affected points are considered the secondary key points). The details are shown in Figure 1.

    Figure 1.  Example of vehicle driving route.

    When a vehicle arrives at the first-level key point, the road condition information of the disaster area is updated, and the relevant information is used to judge whether the vehicle scheduling scheme needs to be redesigned to visit the disaster-affected points that has not been visited currently. Due to the different locations of the vehicles sent from the distribution center at the same time, it is necessary to determine the position of each vehicle when reformulating the vehicle scheduling plan. Assuming that at time t, vehicles k1 and k2 are located on the first-level key point and the path from disaster-affected point i to j, respectively, and there are n second-level key points between i and j, the departure time of the vehicle at each key point are ti, tj, tp(p is the p-th second-level key point, 1pn), respectively. If ti<ttp, the key point p is taken as the position of vehicle k, and if tp<ttj, the key point j is taken as the position of vehicle k. Figure 1 shows an example of two vehicles performing a distribution task. At time t, vehicle 1 arrives at key point 2, and vehicle 2 is located between key points p3 and p4. When updating the vehicle scheduling scheme, the starting position of vehicle 1 is the first-level key point 2, and the starting position of vehicle 2 is the second-level key point p4.

    There are many complex factors in the disaster area environment, which have brought many influences to the solution of the MEDDRC problem. Therefore, in order to reduce the complexity of the problem, the following constraints are given in this paper: 1) only one vehicle completes the material distribution task at each disaster-affected point; 2) the total demand of each affected point visited by each vehicle cannot exceed the maximum load of the vehicle; 3) the vehicles are fully loaded when they depart from the distribution center; and 4) the vehicle must leave after completing the material distribution task at the affected point. In addition, the objective function of the MEDDRC problem is to minimize the time cost spent by the vehicle, so the mathematical model is constructed as follows:

    mincdkSviSIjSIxijk×dij/xijk×dijvijvij (1)

    and subject to

    iRxijk=yjk,j,k (2)
    jRxijk=yik,i,k (3)
    kKyik×qiC,i (4)
    t0jk=x0jk×C,j,k (5)
    iR{0}xihkjR{0}xhjk=0,hR,kK (6)

    The notations in the above formula are defined as shown in Table 2.

    Table 2.  Notations for various parameters of the mathematical model.
    Sets Definition
    SI Set of key points
    R Set of disaster-affected points
    K Set of vehicles
    Parameters
    0 Number of distribution center
    C Maximum load of vehicle
    k The k-th vehicle
    ijh Node variable
    qi Material demand of node i
    cd The cost per unit time of vehicle travel
    dij Distance between nodes i and j
    vij Running speed of the vehicle between nodes i and j
    tijk Load of vehicle k when it reaches node j from node i
    Variables
    xijk Path decision variable, xijk=1 when vehicle k goes from i to j, and xijk=0 otherwise
    yik Vehicle decision variable, yik = 1 when disaster-affected point i is delivered by vehicle k, and yik = 0 otherwise

     | Show Table
    DownLoad: CSV

    In this paper, an IWSA is proposed to solve the MEDDRC problem. The framework of the IWSA is shown in Figure 2. Firstly, the ISCW algorithm is used to plan the initial driving route of the vehicle for obtaining a high-quality initial solution. Then, the group movement rule is used to update the individuals to ensure the detailed local search of population individuals in different regions. In addition, the individual search space is extended by constructing the different weights strategy to maintain the population diversity. The details of the algorithm are shown in Sections 4.2 and 4.3.

    Figure 2.  Flow chart of IWSA for solving MEDDRC problem.

    At the initial time, it is necessary to formulate the optimal path for the vehicle based on the current known location of the disaster-affected point, demand and road information. Assuming that the vehicle is running at a normal speed, the Clarke-Wright (C-W) algorithm [47] can generally be used to generate the initial path of the vehicle. Although the C-W algorithm has good results when used alone, its solution speed will decrease as the VRP scale increases. Therefore, in order to improve the performance of the algorithm to obtain a better initial path, this paper proposes an ISCW algorithm, which takes the angles between the lines connecting the disaster-affected points and the origin as the division basis, scans all disaster-affected points counterclockwise, and uses the improved C-W algorithm to plan the access path of the disaster-affected point. The basic idea is as follows:

    1) Calculate the angle of the disaster-affected point in polar coordinates. Taking the distribution center as the origin, the coordinates of the disaster-affected point are converted from a rectangular coordinate system to a polar coordinate system, and rays are made from the origin to all the disaster-affected points, respectively. The line between the origin and a randomly selected disaster-affected point is taken as the polar axis. The remaining disaster-affected points are scanned counterclockwise starting from the polar axis, and the angle between the polar axis and the line connecting the disaster-affected point and the origin is calculated.

    2) Determine the adjacent disaster-affected points with the largest included angle. Starting from the polar axis, the included angle between the lines formed by connecting the adjacent disaster-affected points to the origin is calculated in turn. The disaster-affected points corresponding to the two rays with the largest angle are selected, and their relative positions are judged (that is, the counterclockwise angle between the polar axis and the line connecting the origin and the disaster-affected point).

    3) Determine the starting point of scanning and group the disaster-affected points. According to the two disaster-affected points with the largest angle determined in 2), the disaster-affected point with the largest angle with the polar axis is selected as the scanning starting point of counterclockwise, and the remaining disaster-affected points are grouped according to the maximum load constraint of the vehicle. That is, if adding a new scanned disaster-affected point makes the total demand in the current group more than the maximum load of the vehicle, the disaster-affected point is added to a new group, and the remaining disaster-affected points are continue scanned.

    4) Determine the visiting order of the disaster-affected points. According to the grouping of the disaster-affected points, the improved C-W algorithm is used to determine the service order of the disaster-affected points in each group, that is, the visiting order is determined sequentially by the saving value. Specifically, for a group of disaster-affected points that do not violate the vehicle load constraint, when the two selected disaster-affected points with the largest savings value are not related to the planned disaster-affected points, the improved C-W algorithm directly adds them to the set of planned disaster-affected points instead of selecting new vehicles to visit.

    When the road conditions change, the main works of IWSA proposed in this paper include: 1) the initial population is generated based on the shortest path mutation to adapt to the dynamic characteristics of the problem; 2) when the population is iteratively updated, the guide individual and the specific movement mode of each individual in the population are determined by the group movement strategy; and 3) in the process of population iteration, when the population diversity is lower than a certain threshold, the method of moving individuals based on different weights is used to screen the individuals and generate new individuals.

    Due to the update of road condition information, the start node is not the distribution center when re-planning the unvisited disaster-affected points, and the vehicles are not necessarily fully loaded at this time. Therefore, in order to adapt to the specificity of the problem, this paper uses the hybrid operator based on the shortest path mutation to initialize the population. The shortest path algorithm is used to generate the initial solution of the subproblem, and the initial solution is expanded by using the 2-opt operator [48] and the split mutation operator [49] to obtain the initial population. The specific method is shown in Figure 3.

    Figure 3.  Principle of 2-opt operator and split mutation operator.

    In Figure 3, the 2-opt operator flips the order of nodes between two randomly selected positions l1 and l2 to generate new individuals. The split mutation operator randomly selects a split point l, keeps the node order before and after the split point l unchanged, and uses the improved C-W algorithm to replan the remaining nodes for generating two new individuals.

    The individual movement method in WSA is mainly suitable for solving continuous problems; however, MEDDRC is a discrete problem. Therefore, in order to adapt to its discreteness and improve the quality of the problems solution, this paper constructs a group movement strategy for updating individuals. The main ideas are as follows:

    1) Grouping of individuals in the population. Hierarchical clustering is used to divide the individuals in the current population equally according to their differences, and to minimize the total individual differences within the group. The clustering objective is as in Eq (7), and the individual with the best quality in each group is recorded as Gb. Eq (8) is used to calculate the difference between individuals i and j, and Eq (9) is used to measure the quality of individuals in the population P.

    argminci=1xSixμi2 (7)
    dij=1rijn+(mi+mj)/2 (8)
    fq(i)=1/disti (9)

    In Eq (7), c is the class number, x is the sample vector, μi is the mean vector of the i-th class samples, and Si is the set of the i-th class samples. In Eq (8), rij represents the same path length traveled by vehicles in individual i and j, n is the total number of disaster-affected points, and mi and mj are the number of vehicles used in individual i and j, respectively. In Eq (9), disti is the total distance traveled by the vehicles to finish the material distribution task for all disaster-affected points represented by individual i, and fq(i) represents the quality of individual i.

    2) Update the optimal individual in the long term memory unit (LTMU). If the current population is the initial population, the highest quality individual Wb and the higher quality individual with the largest difference from Wb are selected from it to join the LTMU; otherwise, the highest quality individual pb in the current population is found and compared with Wb. If individual pb is better than Wb, the Wb is replaced by pb.

    3) Update other individuals in the LTMU. The high-quality individual with the largest difference from Wb is selected from the current population to be added to LTMU, and the remaining individuals except in the Wb original LTMU are deleted.

    4) Individual movement. Within the group, all individuals in the group except the optimal individual Gb are moved with Gb as the guide. Between groups, the individual from the LTMU, which is closest to the optimal individual Gb of each group and better quality, is selected as its guide individual and moved. For the optimal individual pb in the population, if it is the same as the optimal individual Wb in LTMU, the 2-opt operator is used to generate its offspring individuals.

    In order to better illustrate the idea of the group movement strategy, Figure 4 shows an algorithm diagram with 12 disaster-affected points and LTMU size of 4.

    Figure 4.  Diagram of the group movement strategy.

    According to their differences and similarities, when the individual moves to its guide individual, this paper adopts the following movement rules. First, the same path segment in individual i and its guide individual j is found, such as abc, and the path is regarded as a whole node A, whose initial visit position is node a, and its end visit position is node c. Then, the improved C-W algorithm is used to plan the path between the node A and other nodes to obtain the offspring individual k. Figure 5 shows a specific example.

    Figure 5.  Diagram of individual movement.

    In Figure 5, the same path segments of individual i and guide individual j are 1→2 and 6→7, so they are regarded as two whole nodes, and the path planning is carried out together with the remaining nodes, so as to obtain the new offspring individual k.

    In the MEDDRC problem, the solution space of the problem grows explosively with the increase of the number of nodes. The movement of individual whale is a local small-range search based on the original individual and guided by the guide individual. Therefore, by iteratively updating of the population, the individuals gradually gather in a certain area, which makes it difficult for the superposition of the movable space of all individuals in the population to cover the entire solution space of the problem when the individual move toward the guide individual. If the individuals in the population gather in a certain area of the solution space prematurely, the population diversity will be lost, which may cause the algorithm to fall into a local optimum, making it difficult to find the global optimal solution of the problem. Therefore, in order to maintain a high population diversity and to ensure that individuals can be distributed in a large area of the solution space, this paper uses the strategy of different weight to expand the individual search space. When the population diversity is below a certain threshold, half of the individuals with a poor quality and a smaller total difference value in the current population are discarded. Then the search space of the remaining individuals in the population is expanded using different weight strategies to obtain new offspring individuals.

    Population diversity can be used to represent the differences among individuals in a population. When the individuals in the population gather in a certain area, the differences among individuals are small. The population diversity is currently low, which makes the total search space of the individual during moving is also smaller. On the contrary, when the individuals in the population are distributed in a large area of the solution space, the differences among individuals are larger. Currently, the population diversity is higher, which makes the total search space of the individual during moving is also larger. Therefore, the population diversity can be used to reflect the size of the total search space of the individuals in the population during moving, and the population diversity can be represented by inter-individual differences. In this paper, Eq (10) is constructed based on Eq (9) to measure population diversity.

    D=(ijdij)/(ijdij)s2s2 (10)

    In Eq (10), dij is the difference between individual i and j calculated by Eq (9), where s is the population size.

    Everyone in the population can reflect the number of vehicles required for material distribution to all the disaster-affected points and the order of vehicle visits to the disaster-affected sites. In addition, it can be known from Eq (8) that the individual quality is inversely proportional to the total distance traveled by the vehicle represented by the individual, that is, the longer the total distance traveled by the vehicle represented by the individual, the worse its quality. In the same individual, each vehicle departs from the distribution center and returns after completing the distribution task. Due to the different locations of the disaster-affected points, the distance traveled by each vehicle is different, so the distance traveled by each vehicle in the individual has a different degree of influence on its quality. In the same vehicle depot (the vehicle depot refers to the complete travel path of a vehicle, including the disaster location and distribution sequence of the vehicle), the relative position of the disaster-affected point influences the total distance traveled by the vehicle. This paper considers the influence of different parts of the individual when expanding the individual search space, and proposes a method for moving individuals based on different weights. The specific idea is as follows:

    1) For individuals in the population, Eq (11) is used to calculate the proportion of the running distance dist of each vehicle in the individual in the total running distance Dist, thus the weight of each vehicle depot is obtained, and the vehicle depot lc with the largest weight is selected.

    w=dist/Dist (11)

    In Eq (11), if dist is the driving distance of a certain vehicle depot, then Dist is the total driving distance. If dist is the running distance required by vehicles to visit a certain node, then Dist is the running distance of the vehicle depot where the node is located.

    2) Equation (11) is used to calculate the proportion of the total distance dist between the remaining nodes in lc except the distribution center node and its anteroposterior nodes in the total running distance Dist of lc, so as to obtain the weight of each node in lc. Then, the node a with the largest weight is deleted, and the available load q of the vehicles in the current vehicle depot is calculated.

    3) All the nodes except the distribution center in the remaining vehicle depots are traversed in turn, and node b, which can minimize the total distance of lc by inserting it at node a, is selected. Meanwhile, it is satisfied that the demand of node b does not exceed the available load of vehicles in lc, and the demand of node a does not exceed the available load of the vehicle depot where node b is located. The positions of nodes a and b are exchanged to generate new individuals.

    The suddenness and destructiveness of natural disasters has brought severe challenges to governments around the world. Moreover, for a populous country with a complex geographical environment like China, a timely and effective post-disaster response is particularly important. Therefore, in order to provide decision support for managers to formulate effective disaster relief strategies, this paper constructs a MEDDRC model according to the rescue characteristics of natural disaster events, and designs the ISWA solving algorithm. In the following sections, the effectiveness of IWSA in solving the MEDDRC model will be verified through experiments.

    Our experiments were coded in Python 3.8, compiled on PyCharm, and run on an Intel (R) Core (TM) i5-8500T CPU @ 2.10GHz, Windows 10 operating system. In this paper, the experiments are conducted using the randomly generated dataset, and the change of road condition is represented by randomly modifying the distance between the affected points. The number of nodes in the data set is 10, 20, 50 and 100 respectively, and the size of the data set is 50. The node positions are randomly selected in [0,100]2, and the node demand is randomly generated in [1,9], and the vehicle loads corresponding to different node numbers are 20, 30, 40 and 50, respectively.

    The parameters in this paper mainly include the population size, the size of the long-term memory unit, the grouping number of individuals in the population and the threshold of population diversity, in which the relationship among the grouping number g_num, the population size pop_num and the number of hierarchical clustering n can be shown in Eq (12).

    g_num=pop_num/2n (12)

    A larger number of groups means that the division of individuals in the population is more detailed, and the search space is relatively small when the individuals in the group move. A smaller number of groups means that the division of individuals in the population is wider, and the search space is relatively larger when the individuals in the group move. In order to achieve better performance of the algorithm, this paper used a pre-experimental method to determine some parameters, that is, combining the existing works and numerous tests to obtain a set of parameters with relatively good results. The pre-experiments were carried out with a node size of 20, a data set size of 50 and a vehicle load of 30. The final parameters are set as follows: the population size is 30, the long-term memory unit size is 6, the number of groups is 2, and the population diversity threshold is 0.4.

    In order to verify the performance of the ISCW in the model, the ISCW, scan algorithm and C-W algorithm are used to generate the initial vehicle distribution routes of the MEDDRC problem respectively, and the results are shown in Table 3.

    Table 3.  Performance comparison of initial paths generated by different algorithms.
    Scales Batches ISCW scan algorithm Average cost deviation C-W algorithm Average cost deviation
    Average distance (km) Average number of vehicles Average time (s) Average distance (km) Average number of vehicles Average time (s) Average distance (km) Average number of vehicles Average time (s)
    10 1 540.810 3.33 3.758 × 10−3 561.839 3.3 2.299 × 10−4 3.743% 540.948 3.77 6.386 × 10−3 0.026%
    2 522.060 3.26 6.899 × 10−4 536.310 3.25 9.995 × 10−5 2.657% 525.022 3.74 2.098 × 10−3 0.564%
    3 534.432 3.09 7.496 × 10−4 545.404 3.09 5.995 × 10−5 2.012% 544.666 3.7 2.059 × 10−3 1.879%
    20 1 776.420 3.97 1.669 × 10−3 840.303 4.01 9.993 × 10−5 7.602% 865.543 5.88 8.405 × 10−3 10.297%
    2 780.678 4.02 1.529 × 10−3 829.346 4.04 2.299 × 10−4 5.868% 859.095 5.83 8.454 × 10−3 9.128%
    3 778.583 4.12 1.739 × 10−3 826.095 4.07 1.199 × 10−4 5.751% 855.665 5.83 9.054 × 10−3 9.008%
    50 1 1469.121 7.11 5.427 × 10−3 1725.696 7.11 3.899 × 10−4 14.868% 1707.334 12.25 7.152 × 10−2 13.952%
    2 1490.557 7.05 4.837 × 10−3 1743.931 7.1 3.000 × 10−4 14.529% 1750.497 12.46 6.466 × 10−2 14.849%
    3 1508.282 7.05 4.967 × 10−3 1780.993 7.04 3.600 × 10−4 15.312% 1760.988 12.22 6.835 × 10−2 14.350%
    100 1 2465.197 11 1.166 × 10−2 3097.789 10.95 6.296 × 10−4 20.421% 3021.683 22.14 3.222 × 10−1 18.416%
    2 2479.745 11.12 1.195 × 10−2 3156.978 11.1 6.696 × 10−4 21.452% 3054.719 22.04 3.132 × 10−1 18.822%
    3 2465.101 10.99 1.151 × 10−2 3082.964 10.97 6.296 × 10−4 20.041% 3035.305 22.39 3.079 × 10−1 18.786%

     | Show Table
    DownLoad: CSV

    As can be seen from Table 3, under different problem scales, the average distance solved by ISCW is smaller than that of both the scan algorithm and the C-W algorithm for the same set of samples. When the number of nodes is 10, 20, 50 and 100, the average distance calculated by ISCW is reduced by 2.804 and 0.823%, 6.407 and 9.478%, 14.903 and 14.384%, 20.638 and 18.675%, respectively, on average compared with scan algorithm and C-W algorithm. The reduction degree of the average distance solved by the ISCW increases greatly with the increase of problem scale, which is attributed to the improvement of the scan algorithm. In the scan algorithm, a disaster-affected point is randomly selected, and other disaster-affected points are scanned in a certain direction according to the angle between the lines joining the disaster-affected points and the distribution center. By default, the loop formed between the disaster-affected points in similar directions is short; however, the influence of the size of the angle between the lines connecting the disaster-affected points and the distribution center is ignored. In the ISCW, one of the two disaster points with the largest included angle is selected as the starting point to scan, which can avoid dividing the two nodes with the largest included angle in the same vehicle and reduce the distance of the solution to a certain extent.

    In addition, Table 3 also shows that the average number of vehicles calculated by the ISCW is roughly the same as that of the scan algorithm for the same set of samples, but less than that of the C-W algorithm. This is because in the C-W algorithm, when the two disaster-affected points to be selected are not related to the disaster-affected points in the current planning set, a vehicle will be reassigned to access, which may cause unnecessary use of the vehicle or waste of used vehicle capacity. However, in the ISCW, since each group of disaster-affected points is obtained according to the improved scan algorithm, the total demand of the disaster-affected points must be less than or equal to the maximum load of the vehicle. Therefore, if the C-W algorithm is used to optimize the path for each group of disaster-affected points, the remaining disaster-affected points can be directly added to the planned disaster-affected points set when there are new vehicles to be added, which can avoid the waste of vehicle capacity. Additionally, in terms of solution time, the calculation time of ISCW is between that of scan algorithm and C-W algorithm. The ISCW takes more time to calculate than that of the scan algorithm. This is because the ISCW also uses the improved C-W algorithm to optimize the path of the disaster-affected point except for the scan algorithm, so it takes more time than just using the scan algorithm. Compared with the C-W algorithm, the ISCW takes less time because the C-W algorithm needs to select the appropriate nodes from all the disaster-affected points; in turn, according to the saving value, the set of disaster-affected points is relatively large. The ISCW firstly uses the scanning algorithm to group the disaster-affected points, which reduces the set of disaster-affected points compared with directly using C-W algorithm, and can shorten the time spent in selecting nodes to a certain extent. Therefore, compared with using the C-W algorithm, the solution time of the ISCW is shorter.

    To verify the performance of the IWSA in solving the MEDDRC problem, the IWSA, the ACS [50] and the adaptive chaos genetic algorithm (ACGA) [51] are used to solve the problem in this paper respectively, and the calculation results are shown in Table 4 and Figures 6 and 7.

    Table 4.  Calculation results of three algorithms on five groups of data with different numbers of nodes.
    Number of nodes Dataset Best solution (km) Deviation
    IWSA ACS ACGA IWSA/ACS IWSA/ACGA
    10 1 1425 1682 1537 15.28% 7.29%
    2 1974 2020 1706 2.27% −15.70%
    3 1613 1791 1684 9.97% 4.26%
    4 1690 1404 1755 −20.35% 3.71%
    5 1311 1724 2140 23.96% 38.72%
    20 1 2828 3222 3515 12.22% 19.54%
    2 2439 2816 2127 13.39% −14.67%
    3 1991 2978 3132 33.13% 36.41%
    4 1862 2305 2820 19.19% 33.96%
    5 2793 2751 3239 −1.53% 13.77%
    50 1 4171 7457 8764 44.07% 52.41%
    2 3982 8799 7861 54.74% 49.34%
    3 3540 7145 6663 50.46% 46.88%
    4 4079 8299 7623 50.85% 46.49%
    5 4178 6743 7165 38.04% 41.69%
    100 1 6279 17650 15137 64.42% 58.52%
    2 6880 15287 15973 55.00% 56.93%
    3 6505 17772 15919 63.40% 59.13%
    4 6606 16373 15208 59.65% 56.56%
    5 5587 15432 16756 63.80% 66.66%

     | Show Table
    DownLoad: CSV
    Figure 6.  Calculation results of IWSA and ACS with different number of nodes. Note:mean represents the mean of the results for the same set of sample data.
    Figure 7.  Calculation results of IWSA and ACGA with different number of nodes.

    It can be seen from Table 4 that when the three algorithms are used to solve the five groups of sample data under different node numbers, the IWSA has better solution performance than the ACS and the ACGA with the increase of number of nodes. Among the five groups of samples, when the number of nodes is 10, the best solutions of groups 1, 2, 3 and 5 samples solved by the IWSA are better than that of the ACS, and the best solutions of groups 1, 3, 4 and 5 samples solved by the IWSA are better than that of the ACGA. When the number of nodes is 20, the best solutions of groups 1, 2, 3 and 4 samples solved by the IWSA are better than that of the ACS, and the best solutions of groups 1, 3, 4 and 5 samples solved by the IWSA are better than that of the ACGA. This is because, ideally, the search area of individuals in the population is positively correlated with the size of the problem solution space. When the number of nodes is small, the solution space of the problem is relatively small, and the area that the individuals in the population can explore also becomes smaller when solving the problem. With the iterative update of the population, the probability of individuals in the population exploring the best solution with different algorithm rules is higher, so there will be cases in some sample results where the best solutions obtained by both the ACS and the ACGA are better than that of the IWSA. When the number of nodes is 50 and 100, the best solutions for the five groups of samples obtained by the IWSA are better than that of the ACS, and as the number of nodes increases the deviation of the best solutions between the IWSA and the ACS gradually increases, with the average deviation of 47.63 and 61.25%, respectively. Similarly, for the ACGA, with the increasing number of nodes, the deviation of the best solutions between the IWSA and the ACGA also increases gradually, and the average deviation reaches 47.36 and 59.56%, respectively.

    Figure 6 shows that the best solutions obtained by the IWSA and the ACS have different degrees of superiority and inferiority under different number of nodes. Among them, when the number of nodes is 10, there are 71 groups of results calculated by the IWSA better than that of the ACS, that is, relative to ACS, the solving rate of the best solution of the IWSA (refers to the percentage of sample groups of one algorithm better than other algorithm in the experimental results) is 71%. When the number of nodes is 20, the solving rate of the IWSA is 98%. When the number of nodes is 50 and 100, the results solved by the IWSA are better than that of the ACS in 100 groups of samples, respectively, that is, the solving rate of the best solution of the IWSA is 100%. In addition, Figure 7 shows that when the number of nodes is 10, 87% of the best solution obtained by the IWSA is better than that of the ACGA in 100 groups of samples. When the number of nodes is 20, the best solution obtained by the IWSA is better than that of the ACGA in 93% of samples. When the number of nodes is 50 and 100, the best solution curves calculated by the IWSA are all below the average curve, while the best solution curves obtained by the ACGA are all above the average curve, which indicates that the best solutions obtained by the IWSA are better than that of the ACGA.

    Combining Table 4 and Figures 6 and 7, the IWSA, the ACS and the ACGA have different performance in solving the same problem, which is related to the scale of the problem. With the increase of the number of nodes, the solving rate of the IWSA gradually increases, and it is better than that of the ACS and the ACGA. As we know, the optimization algorithm is prone to fall into the local extreme value prematurely, which makes individuals quickly cluster in a small area during the search process. Moreover, we also found that the population diversity indeed decreased rapidly through the calculation of Eq (10) during the experiments. Therefore, by abandoning some individuals in our improvement, the reduction speed of population diversity is effectively delayed, which allows the algorithm has enough time to find the global optimal solution in the search space. In addition, due to the irregular distribution of initial positions of individuals in the solution space caused by the randomness of the heuristic algorithm itself, the moving positions of individuals are also unstable in the process of iterative updating. Therefore, the individual can only move in the direction that may be better under the corresponding rules, and its final position is also related to the specific movement rules and parameters of the algorithm. For small-scale problems, the solution space is relatively small, and the area that the algorithm needs to explore in the optimization process is also smaller. At this time, when three algorithms are used to solve the problem, the best solutions obtained by the ACS and the ACGA are better than that of the IWSA. However, due to the advantages of individual movement rules designed in the IWSA, individuals can search more carefully in the solution space, so that the solving rate of the best solution of the IWSA is still higher than that of the ACS and the ACGA. The performance of the IWSA is much better than that of the ACS and the ACGA in solving large-scale problems. This may be because the grouping mechanism of IWSA allows individuals to conduct more detailed searches within the smaller area of the solution space. Moreover, when the individual search space is extended based on different weights, it can search for optimization multiple times in a larger search area based on the better individual. When the diversity of the population is below a certain threshold, the expansion of the search space can not only prevent individuals from congregating in a small area to avoid the algorithm falling into the local optimum prematurely, but also improve the probability of individuals exploring the best solution to a certain extent.

    Although the advanced individual optimization method enables the IWSA to obtain the best solution with higher quality in solving, its grouping mechanism and the detailed exploration of individuals in the population among the solution space make its time cost higher than that of the ACS and the ACGA. In order to better analyze the time cost of the IWSA, 100 groups of data under different node numbers are divided into 10 groups, and the IWSA, the ACS and the ACGA are used to calculate. The average time cost of each group of samples is shown in Table 5.

    Table 5.  Calculation time of the three algorithms.
    Groups Calculation time of IWSA (s) Calculation time of ACS (s) Calculation time of ACGA (s)
    10 20 50 100 10 20 50 100 10 20 50 100
    1 2.788 7.824 33.550 115.154 0.240 0.968 10.456 56.289 0.081 0.114 0.252 0.599
    2 2.613 7.365 31.796 123.056 0.243 1.286 9.755 59.531 0.085 0.119 0.278 0.555
    3 2.472 8.083 32.471 117.133 0.214 0.997 10.300 58.496 0.085 0.128 0.269 0.624
    4 1.973 8.137 31.961 114.464 0.242 0.977 10.636 55.421 0.082 0.115 0.269 0.624
    5 2.581 7.061 34.702 113.447 0.214 0.962 10.539 56.041 0.107 0.123 0.271 0.609
    6 2.862 7.213 32.138 114.730 0.247 0.995 9.707 55.431 0.080 0.127 0.283 0.553
    7 2.102 8.021 32.105 114.785 0.230 0.936 9.811 56.834 0.089 0.119 0.249 0.607
    8 2.121 7.749 32.648 114.843 0.238 1.027 9.566 61.343 0.090 0.116 0.267 0.610
    9 2.201 7.377 31.975 118.408 0.220 0.954 8.880 60.475 0.081 0.122 0.273 0.589
    10 1.974 7.760 31.797 114.001 0.257 0.974 9.464 53.117 0.087 0.117 0.261 0.576
    Average calculation time (s) 2.369 7.659 32.514 116.002 0.234 1.008 9.911 57.298 0.087 0.120 0.267 0.595

     | Show Table
    DownLoad: CSV

    Table 5 shows that the solution time cost of the three algorithms increases as the number of nodes increases, and the time cost required by the IWSA to solve the problem is the highest. The average time cost required by both the IWSA and the ACGA increases with the number of nodes. The relationship between them increases by multiple, reaching 195 times when the number of nodes is 100. On the contrary, the relationship between the average time cost required by both the IWSA and the ACS decreases by multiple. When the number of nodes is 10, 20, 50 and 100, the relationship between them is roughly a multiple of 10, 7, 3 and 2.

    The multiplicative relationships between the time cost consumed by different algorithms show different trends with the change of the number of nodes. This is because the specific solution rules designed by different algorithms are different during the solution process, such as population initialization method, individual update operations and so on. As a result, the solution times required by the three algorithms are quite different for the same number of iterations and population size, that is, different algorithms have different time complexity due to different specific solution rules. Compared with the ACS and the ACGA, the update operation of population individuals in the IWSA is more complicated. After the hierarchical clustering algorithm is used to group individuals in the population, the individuals move to their guide individuals according to the movement rules of the group. When the population diversity is lower than the threshold, the search area needs to be expanded based on the current individuals, which makes everyone carry out a more detailed search in its neighboring area. Although it improves the probability of individuals searching for the best solution, it also increases the time cost.

    Moreover, in the IWSA, the number of execution steps of the individual update operation is determined by the dimension of the individual (the number of nodes that are not currently accessed). Therefore, in the best case, the maximum number of execution steps of its update operation is number=COUNT×MAXInter×pop_num×n (COUNT and MAXInter are the maximum number of iterations, n is the number of nodes (problem size)), and the best time complexity is O(n). In the worst case, the maximum number of execution steps of its update operation is number=COUNT×MAXInter×pop_num×n×pop_num×n, and the worst time complexity is O(n2). However, due to the different movement rules when updating population individuals, the time complexities of ACS and ACGA are O(n) and O(1), respectively. The time costs of the three algorithms are different in solving problem of the same size due to the difference in the time complexity caused by the different solution rules within the algorithms, which makes the time difference more apparent in numerical terms with the increase of the problem size.

    Based on the dynamic change characteristics of road conditions in disaster areas, this paper proposes an IWSA for updating the material scheduling scheme to solve the shortcomings of the current MEDDRC problem such as the poor timeliness of updating the scheduling scheme and the existing algorithms easy to fall into the local extreme value prematurely in solving. The ISCW is constructed to formulate the initial vehicle routing, and the group movement strategy and different weights strategy are proposed to optimize the WSA from two aspects of individual movement mode and expanding the search space of individuals in the population respectively. The IWSA realizes the purpose of updating the vehicle scheduling scheme in real-time under the change of road conditions, which can provide support for decision-makers to formulate effective material emergency scheduling strategies in major natural disaster events.

    At the same time, when the number of nodes is 10, 20, 50 and 100, the simulation calculation is carried out on four randomly generated data sets containing 100 groups of samples. Through the analysis of the experimental data, the following conclusions are obtained:

    1) When generating the initial driving path of the vehicle, the best solution obtained by the ISCW is better than that of both the scan algorithm and the C-W algorithm. Moreover, with the increase of the problem scale, the reduction degree of the average distance solved by the ISCW increases greatly.

    2) In the case of road conditions changing, the IWSA is used to plan the vehicle scheduling scheme. When the number of nodes is 10 and 20, the proportion of the best solution obtained by the IWSA is better than that of both the ACS and the ACGA is 71, 87% and 98, 93%, respectively. When the number of nodes is 50 and 100, the best solutions obtained by IWSA are better than that of both the ACS and the ACGA. And with the increase of the number of nodes, the average distance of the best solution obtained by the IWSA is much better than that of both the ACS and the ACGA.

    3) In terms of algorithm complexity, as the individuals in the IWSA move within the solution space, everyone performs detailed searches in their neighboring area, so a large amount of time is spent on searching for the optimal solution. Therefore, compared with the ACS and the ACGA, the IWSA spends a higher time cost.

    4) Although the proposed algorithm can solve the MEDDRC problem and has good results, it still has the disadvantage of high time complexity, which is related to its special solution rules. Besides, the values of the parameters can also have a large impact on the performance of the algorithm. Therefore, a suitable set of parameters is also particularly important for optimization algorithms. It is our next priority that how to find an optimal set of algorithm parameters to improve the model solving performance while reducing the computation time.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    This research was supported by Natural Science Innovative Fund of Henan University of Technology (2021ZKCJ18), Joint Fund of Science and Technology R & D Program of Henan Province (222103810078) and Science and Technology Research Project of Henan Province (232102211010).

    The authors declare there is no conflict of interest.

    Table .  The abbreviations used in this article and their full names.
    Abbreviations Full Name
    MED Material Emergency Dispatching
    DRP Dynamic Routing Problem
    SRP Static Routing Problem
    VRP Vehicle Routing Problem
    MEDDRC Material Emergency Dispatching with Dynamic Road Conditions
    WSA Whale Swarm Algorithm
    IWSA Improved Whale Swarm Algorithm
    C-W Clarke-Wright
    ISCW Improved Scan Clarke-Wright
    BCP Branch Cut and Price
    GA Genetic Algorithm
    PSO Particle Swarm Optimization
    LTMU Long Term Memory Unit
    ACS Ant Colony System
    ACGA Adaptive Chaos Genetic Algorithm

     | Show Table
    DownLoad: CSV


    [1] World Health Organization (2018) Global Status Report on Road Safety. WHO: Geneva, Switzerland.
    [2] Othman K (2021) Public acceptance and perception of autonomous vehicles: a comprehensive review. AI and Ethics 1: 355-387. https://doi.org/10.1007/s43681-021-00041-8 doi: 10.1007/s43681-021-00041-8
    [3] Autonomous Vehicle Market to Garner Growth 63.5%. Available from: https://www.precedenceresearch.com/autonomous-vehicle-market
    [4] Glon, R, Edelstein, S (2020) The History of Self-Driving Cars. Available from: https://www.digitaltrends.com/cars/history-of-self-driving-cars-milestones/
    [5] Wiggers K (2020) Waymo's Autonomous Cars Have Driven 20 Million Miles on Public Roads. Available from: https://venturebeat.com/2020/01/06/waymos-autonomous-cars-have-driven-20-million-miles-on-public-roads/
    [6] Othman K (2022) Exploring the implications of autonomous vehicles: A comprehensive review. Innovative Infrastructure Solutions 7: 165. https://doi.org/10.1007/s41062-022-00763-6 doi: 10.1007/s41062-022-00763-6
    [7] Shuttleworth J (2019) SAE Standard News: J3016 Automated-Driving Graphic Update. Available from: https://www.sae.org/news/2019/01/sae-updates-j3016-automated-driving-graphic
    [8] Autopilot. Available from: https://www.tesla.com/en_IE/autopilot
    [9] Othman K (2021) Impact of autonomous vehicles on the physical infrastructure: Changes and challenges. Designs 5: 40. https://doi.org/10.3390/designs5030040 doi: 10.3390/designs5030040
    [10] Othman K (2023) Exploring the evolution of public acceptance towards autonomous vehicles with the level of knowledge. Innovative Infrastructure Solutions 8: 208. https://doi.org/10.1007/s41062-023-01180-z doi: 10.1007/s41062-023-01180-z
    [11] Othman K (2022) Multidimension analysis of autonomous vehicles: the future of mobility. Civil Engineering Journal 7: 71-93. https://doi.org/10.28991/CEJ-SP2021-07-06 doi: 10.28991/CEJ-SP2021-07-06
    [12] Mozaffari S, Al-Jarrah OY, Dianati M, Jennings P, Mouzakitis A (2020) Deep Learning-Based Vehicle Behavior Prediction for Autonomous Driving Applications: A Review. IEEE Trans Intell Transp Syst, 1-15.
    [13] Mehra A, Mandal M, Narang P, Chamola V (2020) ReViewNet: A Fast and Resource Optimized Network for Enabling Safe Autonomous Driving in Hazy Weather Conditions. IEEE Trans Intell Transp Syst, 1-11. https://doi.org/10.1109/TITS.2020.3013099 doi: 10.1109/TITS.2020.3013099
    [14] Othman K (2023) Public attitude towards autonomous vehicles before and after crashes: A detailed analysis based on the demographic characteristics. Cogent Engineering 10: 2156063. https://doi.org/10.1109/TITS.2020.3013099 doi: 10.1109/TITS.2020.3013099
    [15] Velasco-Hernandez G, Yeong DJ, Barry J, Walsh J (2020) Autonomous Driving Architectures, Perception and Data Fusion: A Review. Proceedings of the 2020 IEEE 16th International Conference on Intelligent Computer Communication and Processing (ICCP 2020), Cluj-Napoca, Romania, 3-5. https://doi.org/10.1109/ICCP51029.2020.9266268
    [16] Giacalone J, Bourgeois L, Ancora A (2019) Challenges in aggregation of heterogeneous sensors of Autonomous Driving Systems. Proceedings of the 2019 IEEE Sensors Applications Symposium (SAS), Sophia Antipolis, France, 11-13. https://doi.org/10.1109/SAS.2019.8706005
    [17] Liu X, Baiocchi O (2016) A comparison of the definitions for smart sensors, smart objects and Things in IoT. Proceedings of the 2016 IEEE 7th Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON), Vancouver, BC, Canada, 13-15.
    [18] Wojciechowicz T, Smart Sensor vs Base Sensor—What's the Difference? Symmetry Blog. Available from: https://www.semiconductorstore.com/blog/2018/Smart-Sensor-vs-Base-Sensor-Whats-the-Difference-Symmetry-Blog/3538/#: ~: text = By%20using%20a%20smart%20sensor, achieve%20on%20a%20base%20sensor
    [19] Fayyad J, Jaradat MA, Gruyer D, Najjaran H (2020) Deep Learning Sensor Fusion for Autonomous Vehicle Perception and Localization: A Review. Sensors 20: 4220. https://doi.org/10.3390/s20154220 doi: 10.3390/s20154220
    [20] What Are Convolutional Neural Networks? IBM. 2020. Available from: https://www.ibm.com/cloud/learn/convolutional-neural-networks
    [21] Saha S (2018) A Comprehensive Guide to Convolutional Neural Networks — the ELI5 way. Data Science and ML. Saturn Cloud. Available from: https://saturncloud.io/blog/a-comprehensive-guide-to-convolutional-neural-networks-the-eli5-way/
    [22] Brownlee J (2019) A Gentle Introduction to the Rectified Linear Unit (ReLU). In Deep Learning Performance. Machine Learning Mastery. Available from: https://machinelearningmastery.com/rectified-linear-activation-function-for-deep-learning-neural-networks/
    [23] What is LIDAR? Learn How Lidar Works. Velodyne Lidar. 2022. Available from: https://velodynelidar.com/what-is-lidar/
    [24] Wang P (2021) Research on comparison of LIDAR and camera in autonomous driving. Journal of Physics: Conference Series 2093: 012032. https://doi.org/10.1088/1742-6596/2093/1/012032 doi: 10.1088/1742-6596/2093/1/012032
    [25] ScienceDirect (2018) Inertial measurement. Inertial Measurement - an overview. Available from: https://www.sciencedirect.com/topics/engineering/inertial-measurement
    [26] Camera, radar and LIDAR: A comparison of the three types of sensors and their limitations. 2021. Available from: https://autocrypt.io/camera-radar-lidar-comparison-three-types-of-sensors/
    [27] The use of radar technology in Autonomous Vehicles. 2022. Cadence. Available from: https://resources.system-analysis.cadence.com/blog/msa2022-the-use-of-radar-technology-in-autonomous-vehicles
    [28] Dobler S, Kondel V (2023) LiDAR and Radar Battle For Autonomous Vehicle Turf. Determining the future of autonomous driving system. Available from: https://www.oliverwyman.com/our-expertise/insights/2023/jul/lidar-radar-future-of-autonomous-driving-systems.html
    [29] Minaee S, Boykov Y, Forikli F, Plaza A, Kehtarnavaz N, Terzopoulos D (2021) Image Segmentation Using Deep Learning: A Survey. IEEE Transactions on Pattern Analysis and Machine Intelligence 44: 3523-3542. https://doi.org/10.1109/tpami.2021.3059968 doi: 10.1109/tpami.2021.3059968
    [30] Sensor fusion. Sensor Fusion - an overview. ScienceDirect Topics. 2017. Available from: https://www.sciencedirect.com/topics/engineering/sensor-fusion
    [31] Nabati R, Qi H (2019) RRPN: Radar Region Proposal Network for Object Detection in Autonomous Vehicles. 2019 IEEE International Conference on Image Processing (ICIP), 3093-3097. https://doi.org/10.1109/ICIP.2019.8803392
    [32] Lewis G (2016) Object Detection for Autonomous Vehicles.
    [33] Satilmis Y, Tufan F, Şara M, Karslı M, Eken S, Sayar A (2019) CNN Based Traffic Sign Recognition for Mini Autonomous Vehicles. Information Systems Architecture and Technology: Proceedings of 39th International Conference on Information Systems Architecture and Technology-ISAT 2018: Part II, 85-94. https://doi.org/10.1007/978-3-319-99996-8_8
    [34] Shen X, Batkovic I, Govindarajan V, Falcone P, Darrell T, Borrelli F (2020) ParkPredict: Motion and Intent Prediction of Vehicles in Parking Lots. 2020 IEEE Intelligent Vehicles Symposium (IV), Las Vegas, NV, USA, 1170-1175. https://doi.org/10.1109/IV47402.2020.9304795
    [35] Gao H, Cheng B, Wang J, Li K, Zhao J, Li D (2018) Object Classification using CNN-Based Fusion of Vision and LIDAR in Autonomous Vehicle Environment. IEEE T Ind Inform 14: 4224-4231. https://doi.org/10.1109/TII.2018.2822828 doi: 10.1109/TII.2018.2822828
    [36] Saez A, Bergasa L, Romeral E, Guillén M, Barea R, Sanz R (2018) CNN-based Fisheye Image Real-Time Semantic Segmentation. 2018 IEEE Intelligent Vehicles Symposium (IV), 1039-1044. https://doi.org/10.1109/IVS.2018.8500456
    [37] Hofesmann E (2020) IoU a better detection evaluation metric. Towards Data Science. Available from: https://towardsdatascience.com/iou-a-better-detection-evaluation-metric-45a511185be1
    [38] Farag W, Saleh Z (2018) Behavior Cloning for Autonomous Driving using Convolutional Neural Networks. 2018 International Conference on Innovation and Intelligence for Informatics, Computing, and Technologies (3ICT), 1-7. https://doi.org/10.1109/3ICT.2018.8855753
    [39] Iftikhar S, Asim M, Zhang Z, El-Latif AAA (2022) Advance generalization technique through 3D CNN to overcome the false positives pedestrian in autonomous vehicles. Telecommun Syst 80: 545-557.
    [40] Gao Y, Tian F, Li J, Fang Z, Al-Rubaye S, Song W, et al. (2022) Joint optimization of depth and ego-motion for intelligent autonomous vehicles. IEEE T Intell Transp Syst. https://doi.org/10.1109/TITS.2022.3159275
    [41] Liang T, Bao H, Pan W, Pan F (2022) Traffic sign detection via improved sparse R-CNN for autonomous vehicles. J Adv Transport 2022: 1-16. https://doi.org/10.1155/2022/3825532 doi: 10.1155/2022/3825532
    [42] Zhu C, Mehrabi A, Xiao Y, Wen Y (2019) CrowdParking: Crowdsourcing Based Parking Navigation in Autonomous Driving Era. 2019 International Conference on Electromagnetics in Advanced Applications (ICEAA), 1401-1405. https://doi.org/10.1109/ICEAA.2019.8879201
    [43] Park M, Kim H, Park S (2021) A Convolutional Neural Network-Based End-to-End Self-Driving Using LiDAR and Camera Fusion: Analysis Perspectives in a Real-World Environment. Electronics 10: 2608. https://doi.org/10.3390/electronics10212608 doi: 10.3390/electronics10212608
    [44] Shen X, Lacayo M, Guggilla N, Borrelli F (2022) ParkPredict+: Multimodal Intent and Motion Prediction for Vehicles in Parking Lots with CNN and Transformer. 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC), 3999-4004. https://doi.org/10.1109/ITSC55140.2022.9922162
    [45] Heinen MR, Osorio FS, Heinen FJ, Kelber C (2006) SEVA3D: Using Arti cial Neural Networks to Autonomous Vehicle Parking Control. 2006 IEEE International Joint Conference on Neural Network Proceedings, 4704-4711. https://doi.org/10.1109/IJCNN.2006.247124
    [46] Wang Y, Ren B (2020) Quadrotor-Enabled Autonomous Parking Occupancy Detection. 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Las Vegas, NV, USA, 8287-8292. https://doi.org/10.1109/IROS45743.2020.9341081
    [47] Min C, Xu J, Xiao L, Zhao D, Nie Y, Dai B (2021) Attentional Graph Neural Network for Parking-slot Detection. IEEE Robotic Autom Lett 6: 3445-3450. https://doi.org/10.1109/LRA.2021.3064270 doi: 10.1109/LRA.2021.3064270
    [48] Bernuth AV, Volk G, Bringmann O (2019) Simulating Photo-realistic Snow and Fog on Existing Images for Enhanced CNN Training and Evaluation. 2019 IEEE Intelligent Transportation Systems Conference (ITSC), 41-46. https://doi.org/10.1109/ITSC.2019.8917367
    [49] Lei Y, Emaru T, Ravankar AA, Kobayashi Y, Wang S (2020) Semantic Image Segmentation on Snow Driving Scenarios. 2020 IEEE International Conference on Mechatronics and Automation (ICMA), 1094-1100. https://doi.org/10.1109/ICMA49215.2020.9233538
    [50] Bijelic M, Gruber T, Mannan F, Kraus F, Ritter W, Dietmayer K, et al. (2020) Seeing Through Fog Without Seeing Fog: Deep Multimodal Sensor Fusion in Unseen Adverse Weather. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 11682-11692. https://doi.org/10.1109/CVPR42600.2020.01170
    [51] Cai Y, Sun X, Wang H, Chen L, Jiang H (2016) Night-Time Vehicle Detection Algorithm Based on Visual Saliency and Deep Learning. Journal of Sensors 2016: 1-7. https://doi.org/10.1155/2016/8046529 doi: 10.1155/2016/8046529
    [52] Liu Q, Li X, Yuan S, Li Z (2021) Decision-Making Technology for Autonomous Vehicles: Learning-Based Methods, Applications and Future Outlook. 2021 IEEE International Intelligent Transportation Systems Conference (ITSC), 30-37. https://doi.org/10.1109/ITSC48978.2021.9564580
    [53] Jiménez F, Clavijo M, Cerrato A (2022) Perception, Positioning and Decision-Making Algorithms Adaptation for an Autonomous Valet Parking System Based on Infrastructure Reference Points Using One Single LiDAR. Sensors 22: 979. https://doi.org/10.3390/s22030979 doi: 10.3390/s22030979
    [54] Ferguson D, Baker C, Likhachev M, Dolan J (2008) A reasoning framework for autonomous urban driving. 2008 IEEE Intelligent Vehicles Symposium, 775-780. https://doi.org/10.1109/IVS.2008.4621247
    [55] Babu M, Oza Y, Singh AK, Krishna KM, Medasani S (2018) Model Predictive Control for Autonomous Driving Based on Time Scaled Collision Cone. 2018 European Control Conference (ECC), 641-648. https://doi.org/10.23919/ECC.2018.8550510
    [56] Zhang X, Liniger A, Sakai A, Borrelli F (2018) Autonomous Parking Using Optimization-Based Collision Avoidance. 2018 IEEE Conference on Decision and Control (CDC), 4327-4332. https://doi.org/10.1109/CDC.2018.8619433
    [57] Gindullina E, Mortag S, Dudin M, Badia L (2021) Multi-Agent Navigation of a Multi-Storey Parking Garage via Game Theory. 2021 IEEE 22nd International Symposium on a World of Wireless, Mobile and Multimedia Networks (WoWMoM), Pisa, Italy, 280-285. https://doi.org/10.1109/WoWMoM51794.2021.00052
    [58] Sheng W, Li B, Zhong X (2021) Autonomous Parking Trajectory Planning With Tiny Passages: A Combination of Multistage Hybrid A-Star Algorithm and Numerical Optimal Control. IEEE Access 9: 102801-102810. https://doi.org/10.1109/ACCESS.2021.3098676 doi: 10.1109/ACCESS.2021.3098676
    [59] Hongbo G, Guotao X, Xinyu Z, Bo C (2017) Autonomous parking control for intelligent vehicles based on a novel algorithm. The Journal of China Universities of Posts and Telecommunications 24: 51-56. https://doi.org/10.1016/S1005-8885(17)60223-1 doi: 10.1016/S1005-8885(17)60223-1
    [60] Li Q, Li R, Ji K, Dai W (2015) Kalman filter and its application. In 2015 8th International Conference on Intelligent Networks and Intelligent Systems (ICINIS), 74-77. IEEE.
    [61] Kato S, Tokunaga S, Maruyama Y, Maeda S, Hirabayashi M, Kitsukawa Y, et al. (2018) Autoware on Board: Enabling Autonomous Vehicles with Embedded Systems. 2018 ACM/IEEE 9th International Conference on Cyber-Physical Systems (ICCPS), 287-296. https://doi.org/10.1109/ICCPS.2018.00035
    [62] Li Q, Queralta JP, Gia TN, Zou Z, Westerlund T (2020) Multi Sensor Fusion for Navigation and Mapping in Autonomous Vehicles: Accurate Localization in Urban Environments. Unmanned Systems 8: 229-237.
    [63] Realpe M, Vintimilla B, Vlacic L (2016) MULTI-SENSOR FUSION MODULE IN A FAULT TOLERANT PERCEPTION SYSTEM FOR AUTONOMOUS VEHICLES. Journal of Automation and Control Engineering 4: 460-466. https://doi.org/10.18178/joace.4.6.460-466 doi: 10.18178/joace.4.6.460-466
    [64] Saxena S, Isukapati IK, Smith SF, Dolan JM (2019) Multiagent Sensor Fusion for Connected & Autonomous Vehicles to Enhance Navigation Safety. 2019 IEEE Intelligent Transportation Systems Conference (ITSC), 2490-2495. https://doi.org/10.1109/ITSC.2019.8917298
    [65] Nabati R, Qi H (2020) Radar-Camera Sensor Fusion for Joint Object Detection and Distance Estimation in Autonomous Vehicles. arXiv, abs/2009.08428.
    [66] Farag W (2020) Kalman-filter-based sensor fusion applied to road-objects detection and tracking for autonomous vehicles. Proceedings of the Institution of Mechanical Engineers, Part I: Journal of Systems and Control Engineering 235: 1125-1138. https://doi.org/10.1177/0959651820975523 doi: 10.1177/0959651820975523
    [67] Liu Y, Fan X, Lv C, Wu J, Li L, Ding D (2017) An innovative information fusion method with adaptive Kalman filter for integrated INS/GPS navigation of autonomous vehicles. Mech Syst Signal Process 100: 605-616. https://doi.org/10.1016/j.ymssp.2017.07.051 doi: 10.1016/j.ymssp.2017.07.051
    [68] Ouyang Z, Cui J, Dong X, Li Y, Niu J (2021) SaccadeFork: A lightweight multi-sensor fusion-based target detector. Informa Fusion 77: 172-183. https://doi.org/10.1016/j.inffus.2021.07.004 doi: 10.1016/j.inffus.2021.07.004
    [69] Aldibaja M, Kuramoto A, Yanase R, Kim TH, Yonada K, Suganuma N (2018) Lateral Road-mark Reconstruction Using Neural Network for Safe Autonomous Driving in Snow-wet Environments. 2018 IEEE International Conference on Intelligence and Safety for Robotics (ISR), 486-493. https://doi.org/10.1109/IISR.2018.8535758
    [70] Convolutional Neural Network (CNN). Developers Breach. Available from: https://developersbreach.com/convolution-neural-network-deep-learning/
    [71] Jocher G, Keita Z (2022) YOLO Object Detection Explained: A Beginner's Guide. DataCamp. Available from: https://www.datacamp.com/blog/yolo-object-detection-explained
    [72] Chablani M (2017) YOLO — You only look once, real time object detection explained. Towards Data Science. Available from: https://towardsdatascience.com/yolo-you-only-look-once-real-time-object-detection-explained-492dc9230006
    [73] Scanbot SDK (2022) YOLO object detection and its applications in computer vision. Available from: https://www.linkedin.com/pulse/yolo-object-detection-its-applications-computer-vision-scanbotsdk/
    [74] Redmon J, Divvala S, Girshick R, Farhadi A (2016) You only look once: Unified, real-time object detection. Proceedings of the IEEE conference on computer vision and pattern recognition, 779-788.
    [75] Gandhi R (2018) R-CNN, Fast R-CNN, Faster R-CNN, YOLO — Object Detection Algorithms. Towards Data Science. Available from: https://towardsdatascience.com/r-cnn-fast-r-cnn-faster-r-cnn-yolo-object-detection-algorithms-36d53571365e
    [76] Ananth S (2019) Faster R-CNN for object detection. Towards Data Science. Available from: https://towardsdatascience.com/faster-r-cnn-for-object-detection-a-technical-summary-474c5b857b46
    [77] Pujara A (2020) Concept of AlexNet: - Convolutional Neural Network. Analytics Vidhya. Available from: https://medium.com/analytics-vidhya/concept-of-alexnet-convolutional-neural-network-6e73b4f9ee30
    [78] Ertan H (2021) CNN-LSTM based Models for Multiple Parallel Input and Multi-Step Forecast. Towards Data Science. Available from: https://towardsdatascience.com/cnn-lstm-based-models-for-multiple-parallel-input-and-multi-step-forecast-6fe2172f7668
    [79] Romera E, Álvarez JM, Bergasa LM, Arroyo R (2017) ERFNet: Efficient Residual Factorized ConvNet for Real-Time Semantic Segmentation. IEEE T Intell Transp Syst 19: 263-272. https://doi.org/10.1109/TITS.2017.2750080 doi: 10.1109/TITS.2017.2750080
    [80] Sanchez-Lengeling B, Reif E, Pearce A, Wiltschko AB (2021) A Gentle Introduction to Graph Neural Networks. Distill.pub. Available from: https://distill.pub/2021/gnn-intro/
    [81] Wood T, Transformer Neural Network Definition. DeepAI. Available from: https://deepai.org/machine-learning-glossary-and-terms/transformer-neural-network
    [82] Rjoub G, Wahab OA, Bentahar J, Bataineh AS (2021) Improving autonomous vehicles safety in snow weather using federated YOLO CNN learning. In Mobile Web and Intelligent Information Systems: 17th International Conference, MobiWIS 2021, Virtual Event, 121-134. Cham: Springer International Publishing. https://doi.org/10.1007/978-3-030-83164-6_10
    [83] Zaghari N, Fathy M, Jameii SM, Shahverdy M (2021) The improvement in obstacle detection in autonomous vehicles using YOLO non-maximum suppression fuzzy algorithm. The Journal of Supercomputing 77: 13421-13446. https://doi.org/10.1007/s11227-021-03813-5 doi: 10.1007/s11227-021-03813-5
    [84] Kavitha R, Nivetha S (2021) Pothole and object detection for an autonomous vehicle using yolo. In 2021 5th International Conference on Intelligent Computing and Control Systems (ICICCS), 1585-1589. IEEE.
    [85] Pandey R, Malik A (2021) Object detection and movement prediction for autonomous vehicle: a review. In 2021 2nd International Conference on Secure Cyber Computing and Communications (ICSCCC), 60-65. IEEE. https://doi.org/10.1109/ICSCCC51823.2021.9478167
    [86] Mseddi WS, Sedrine MA, Attia R (2021) YOLOv5 based visual localization for autonomous vehicles. In 2021 29th European Signal Processing Conference (EUSIPCO), 746-750. IEEE.
    [87] Liang S, Wu H, Zhen L, Hua Q, Garg S, Kaddoum G, et al. (2022) Edge YOLO: Real-time intelligent object detection system based on edge-cloud cooperation in autonomous vehicles. IEEE T Intell Transp Syst 23: 25345-25360. https://doi.org/10.1109/TITS.2022.3158253 doi: 10.1109/TITS.2022.3158253
    [88] Mohanapriya S, Natesan P, Indhumathi P, Mohanapriya STP, Monisha R (2021) Object and lane detection for autonomous vehicle using YOLO V3 algorithm. In AIP Conference Proceedings 2387: 140009. AIP Publishing LLC. https://doi.org/10.1063/5.0068836
    [89] Dewi C, Chen RC, Jiang X, Yu H (2022) Deep convolutional neural network for enhancing traffic sign recognition developed on Yolo V4. Multimed Tools Appl 81: 37821-37845. https://doi.org/10.1007/s11042-022-12962-5 doi: 10.1007/s11042-022-12962-5
    [90] Benjumea A, Teeti I, Cuzzolin F, Bradley A (2021) YOLO-Z: Improving small object detection in YOLOv5 for autonomous vehicles. arXiv preprint arXiv: 2112.11798.
    [91] Kosuru VSR, Venkitaraman AK (2022) Preventing the False Negatives of Vehicle Object Detection in Autonomous Driving Control Using Clear Object Filter Technique. In 2022 Third International Conference on Smart Technologies in Computing, Electrical and Electronics (ICSTCEE), 1-6. IEEE. https://doi.org/10.1109/ICSTCEE56972.2022.10100170
    [92] Fanthony IV, Husin Z, Hikmarika H, Dwijayanti S, Suprapto BY (2021) YOLO Algorithm-Based Surrounding Object Identification on Autonomous Electric Vehicle. In 2021 8th International Conference on Electrical Engineering, Computer Science and Informatics (EECSI), 151-156. IEEE. https://doi.org/10.23919/EECSI53397.2021.9624275
    [93] Motwani NP, Soumya S, Singh U (2022) Object Detection and Tracking for Autonomous Vehicles using Deep Learning Technique-YOLO. In 2022 International Conference on Smart Generation Computing, Communication and Networking (SMART GENCON), 1-6. IEEE. https://doi.org/10.1109/SMARTGENCON56628.2022.10083703
    [94] Valeja Y, Pathare S, Patel D, Pawar M (2021) Traffic Sign Detection using Clara and Yolo in Python. In 2021 7th International Conference on Advanced Computing and Communication Systems (ICACCS) 1: 367-371. IEEE. https://doi.org/10.1109/ICACCS51430.2021.9442065
    [95] Prakash M, Janarthanan M, Devi D (2023) Multiple Objects Identification for Autonomous Car using YOLO and CNN. 2023 7th International Conference on Intelligent Computing and Control Systems (ICICCS), 597-601. IEEE. https://doi.org/10.1109/ICICCS56967.2023.10142751
    [96] Unlu E, Zenou E, Riviere N, Dupouy PE (2019) An autonomous drone surveillance and tracking architecture. In 2019 Autonomous Vehicles and Machines Conference, AVM 2019 31: 35-1 - 35-7. https://doi.org/10.2352/ISSN.2470-1173.2019.15.AVM-035
    [97] Iftikhar S, Asim M, Zhang Z, El-Latif AAA (2022) Advance generalization technique through 3D CNN to overcome the false positives pedestrian in autonomous vehicles. Telecommun Syst 80: 545-557. https://doi.org/10.1007/s11235-022-00930-1 doi: 10.1007/s11235-022-00930-1
    [98] Dazlee NMAA, Khalil SA, Abdul-Rahman S, Mutalib S (2022) Object detection for autonomous vehicles with sensor-based technology using yolo. International Journal of Intelligent Systems and Applications in Engineering, 10: 129-134. https://doi.org/10.18201/ijisae.2022.276 doi: 10.18201/ijisae.2022.276
    [99] Masmoudi M, Ghazzai H, Frikha M, Massoud Y (2019) Object detection learning techniques for autonomous vehicle applications. In 2019 IEEE International Conference on Vehicular Electronics and Safety (ICVES), 1-5. IEEE. https://doi.org/10.1109/ICVES.2019.8906437
    [100] Farrukh FUD, Zhang C, Jiang Y, Zhang Z, Wang Z, Wang Z, et al. (2020) Power efficient tiny yolo cnn using reduced hardware resources based on booth multiplier and wallace tree adders. IEEE Open Journal of Circuits and Systems 1: 76-87. https://doi.org/10.1109/OJCAS.2020.3007334 doi: 10.1109/OJCAS.2020.3007334
    [101] Wang G, Guo J, Chen Y, Li Y, Xu Q (2019) A PSO and BFO-based learning strategy applied to faster R-CNN for object detection in autonomous driving. IEEE Access 7: 18840-18859. https://doi.org/10.1109/ACCESS.2019.2897283 doi: 10.1109/ACCESS.2019.2897283
    [102] Li X, Xie Z, Deng X, Wu Y, Pi Y (2022) Traffic sign detection based on improved faster R-CNN for autonomous driving. The Journal of Supercomputing, 1-21. https://doi.org/10.1007/s11227-021-04230-4
    [103] Bin Issa R, Das M, Rahman MS, Barua M, Rhaman MK, Ripon KSN, et al. (2021) Double deep Q-learning and faster R-Cnn-based autonomous vehicle navigation and obstacle avoidance in dynamic environment. Sensors 21: 1468. https://doi.org/10.3390/s21041468 doi: 10.3390/s21041468
    [104] Li P, Chen X, Shen S (2019) Stereo r-cnn based 3d object detection for autonomous driving. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7644-7652. https://doi.org/10.1109/CVPR.2019.00783
    [105] Chen ST, Cornelius C, Martin J, Chau DH (2019) Shapeshifter: Robust physical adversarial attack on faster r-cnn object detector. In Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2018, Proceedings, Part I 18, 52-68. Springer International Publishing. https://doi.org/10.1007/978-3-030-10925-7_4
    [106] Mostafa T, Chowdhury SJ, Rhaman MK, Alam MGR (2022) Occluded Object Detection for Autonomous Vehicles Employing YOLOv5, YOLOX and Faster R-CNN. In 2022 IEEE 13th Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON), 0405-0410. IEEE. https://doi.org/10.1109/IEMCON56893.2022.9946565
    [107] Yang W, Li Z, Wang C, Li J (2020) A multi-task Faster R-CNN method for 3D vehicle detection based on a single image. Appl Soft Comput 95: 106533. https://doi.org/10.1016/j.asoc.2020.106533 doi: 10.1016/j.asoc.2020.106533
    [108] Liang T, Bao H, Pan W, Pan F (2022) Traffic sign detection via improved sparse R-CNN for autonomous vehicles. J Adv Transport 2022: 1-16. https://doi.org/10.1155/2022/3825532 doi: 10.1155/2022/3825532
    [109] Kukreja R, Rinchen S, Vaidya B, Mouftah HT (2020) Evaluating traffic signs detection using faster r-cnn for autonomous driving. In 2020 IEEE 25th International Workshop on Computer Aided Modeling and Design of Communication Links and Networks (CAMAD), 1-6. IEEE. https://doi.org/10.1109/CAMAD50429.2020.9209289
    [110] Amin S, Galasso F (2017) Geometric proposals for faster R-CNN. In 2017 14th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS), 1-6. IEEE. https://doi.org/10.1109/AVSS.2017.8078518
    [111] Chan PH, Huggett A, Souvalioti G, Jennings P, Donzella V (2022) Influence of AVC and HEVC compression on detection of vehicles through Faster R-CNN. IEEE T Intell Transp Syst. https://doi.org/10.36227/techrxiv.19808566.v1
    [112] Kortmann F, Talits K, Fassmeyer P, Warnecke A, Meier N, Heger J, et al. (2020) Detecting various road damage types in global countries utilizing faster r-cnn. In 2020 IEEE International Conference on Big Data (Big Data), 5563-5571. IEEE. https://doi.org/10.1109/BigData50022.2020.9378245
    [113] Qian R, Liu Q, Yue Y, Coenen F, Zhang B (2016) Road surface traffic sign detection with hybrid region proposal and fast R-CNN. In 2016 12th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery (ICNC-FSKD), 555-559. IEEE. https://doi.org/10.1109/FSKD.2016.7603233
    [114] Nabati R, Qi H (2019) Rrpn: Radar region proposal network for object detection in autonomous vehicles. In 2019 IEEE International Conference on Image Processing (ICIP), 3093-3097. IEEE. https://doi.org/10.1109/ICIP.2019.8803392
    [115] Cheng P, Liu W, Zhang Y, Ma H (2018) LOCO: local context based faster R-CNN for small traffic sign detection. In MultiMedia Modeling: 24th International Conference, MMM 2018, Bangkok, Thailand, February 5-7, 2018, Proceedings, Part I 24, 329-341. Springer International Publishing. https://doi.org/10.1007/978-3-319-73603-7_27
    [116] Bi R, Xiong J, Tian Y, Li Q, Choo KKR (2022) Achieving lightweight and privacy-preserving object detection for connected autonomous vehicles. IEEE Internet Things 10: 2314-2329.
    [117] Fan Q, Brown L, Smith J (2016) A closer look at Faster R-CNN for vehicle detection. In 2016 IEEE intelligent vehicles symposium (IV), 124-129. IEEE. https://doi.org/10.1109/IVS.2016.7535375
    [118] Chen L, Lin S, Lu X, Cao D, Wu H, Guo C, et al. (2021) Deep neural network based vehicle and pedestrian detection for autonomous driving: A survey. IEEE T Intell Transp Syst 22: 3234-3246. https://doi.org/10.1109/TITS.2020.2993926 doi: 10.1109/TITS.2020.2993926
    [119] Saleh K, Hossny M, Hossny A, Nahavandi S (2017) Cyclist detection in lidar scans using faster r-cnn and synthetic depth images. In 2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC), 1-6. IEEE. https://doi.org/10.1109/ITSC.2017.8317599
    [120] Carranza-García M, Torres-Mateo J, Lara-Benítez P, García-Gutiérrez J (2020) On the performance of one-stage and two-stage object detectors in autonomous vehicles using camera data. Remote Sensing 13: 89. https://doi.org/10.3390/rs13010089 doi: 10.3390/rs13010089
    [121] Adam K, Mohd II, Younis YM (2021) The impact of the soft errors in convolutional neural network on GPUs: Alexnet as case study. Procedia Computer Science 182: 89-94. https://doi.org/10.1016/j.procs.2021.02.012 doi: 10.1016/j.procs.2021.02.012
    [122] Tan L, Yu K, Lin L, Cheng X, Srivastava G, Lin JCW, et al. (2021) Speech emotion recognition enhanced traffic efficiency solution for autonomous vehicles in a 5G-enabled space-air-ground integrated intelligent transportation system. IEEE T Intell Transp Syst 23: 2830-2842. https://doi.org/10.1109/TITS.2021.3119921 doi: 10.1109/TITS.2021.3119921
    [123] Szymak P, Gasiorowski M (2020) Using pretrained alexnet deep learning neural network for recognition of underwater objects. NAŠE MORE: znanstveni časopis za more I pomorstvo 67: 9-13. https://doi.org/10.17818/NM/2020/1.2 doi: 10.17818/NM/2020/1.2
    [124] Gao H, Cheng B, Wang J, Li K, Zhao J, Li D (2018) Object classification using CNN-based fusion of vision and LIDAR in autonomous vehicle environment. IEEE T Ind Inform 14: 4224-4231. https://doi.org/10.1109/TII.2018.2822828 doi: 10.1109/TII.2018.2822828
    [125] Zhu Z, Hu Z, Dai W, Chen H, Lv Z (2022) Deep learning for autonomous vehicle and pedestrian interaction safety. Safety Sci 145: 105479. https://doi.org/10.1016/j.ssci.2021.105479 doi: 10.1016/j.ssci.2021.105479
    [126] Kocić J, Jovičić N, Drndarević V (2019) An end-to-end deep neural network for autonomous driving designed for embedded automotive platforms. Sensors 19: 2064. https://doi.org/10.3390/s19092064 doi: 10.3390/s19092064
    [127] Kumaar S, Mannar S, Omkar SN (2018) Juncnet: A deep neural network for road junction disambiguation for autonomous vehicles. arXiv preprint arXiv: 1809.01011.
    [128] Magee A (2019) Place-based navigation for autonomous vehicles with deep learning neural networks. Doctoral dissertation, Monterey, CA; Naval Postgraduate School.
    [129] Kaymak Ç, Uçar A (2019) Semantic image segmentation for autonomous driving using fully convolutional networks. In 2019 International Artificial Intelligence and Data Processing Symposium (IDAP), 1-8. IEEE. https://doi.org/10.1109/IDAP.2019.8875923
    [130] Xie G, Shangguan A, Fei R, Ji W, Ma W, Hei X (2020) Motion trajectory prediction based on a CNN-LSTM sequential model. Sci China Inform Sci 63: 1-21. https://doi.org/10.1007/s11432-019-2761-y doi: 10.1007/s11432-019-2761-y
    [131] Kortli Y, Gabsi S, Voon LFLY, Jridi M, Merzougui M, Atri M (2022) Deep embedded hybrid CNN-LSTM network for lane detection on NVIDIA Jetson Xavier NX. Knowl-based syst 240: 107941. https://doi.org/10.1016/j.knosys.2021.107941 doi: 10.1016/j.knosys.2021.107941
    [132] Li P, Abdel-Aty M, Yuan J (2020) Real-time crash risk prediction on arterials based on LSTM-CNN. Accident Anal Prev 135: 105371. https://doi.org/10.1016/j.aap.2019.105371 doi: 10.1016/j.aap.2019.105371
    [133] Dong B, Liu H, Bai Y, Lin J, Xu Z, Xu X, Kong Q (2021) Multi-modal trajectory prediction for autonomous driving with semantic map and dynamic graph attention network. arXiv preprint arXiv: 2103.16273.
    [134] Zhao M, Li Y, Asif S, Zhu Y, Tang F (2022) C-LSTM: CNN and LSTM Based Offloading Prediction Model in Mobile Edge Computing (MEC). In 2022 IEEE 23rd International Conference on High Performance Switching and Routing (HPSR), 245-251. IEEE. https://doi.org/10.1109/HPSR54439.2022.9831405
    [135] Li X, Ying X, Chuah MC (2019) Grip++: Enhanced graph-based interaction-aware trajectory prediction for autonomous driving. arXiv preprint arXiv: 1907.07792.
    [136] Zhi Z, Liu D, Liu L (2022) A performance compensation method for GPS/INS integrated navigation system based on CNN-LSTM during GPS outages. Measurement 188: 110516. https://doi.org/10.1016/j.measurement.2021.110516 doi: 10.1016/j.measurement.2021.110516
    [137] Anbalagan S, Raja G, Gurumoorthy S, Suresh RD, Dev K (2023) IIDS: Intelligent Intrusion Detection System for Sustainable Development in Autonomous Vehicles. IEEE T Intell Transp Syst. https://doi.org/10.1109/TITS.2023.3271768
    [138] Ziya TAN, KARAKOSE M (2020) Comparative study for deep reinforcement learning with CNN, RNN, and LSTM in autonomous navigation. In 2020 International Conference on Data Analytics for Business and Industry: Way Towards a Sustainable Economy (ICDABI), 1-5. IEEE.
    [139] Poibrenski A, Klusch M, Vozniak I, Müller C (2021) Multimodal multi-pedestrian path prediction for autonomous cars. ACM SIGAPP Applied Computing Review 20: 5-17. https://doi.org/10.1145/3447332.3447333 doi: 10.1145/3447332.3447333
    [140] Sáez Á, Bergasa LM, López-Guillén E, Romera E, Tradacete M, Gómez-Huélamo C, et al. (2019) Real-time semantic segmentation for fisheye urban driving images based on ERFNet. Sensors 19: 503. https://doi.org/10.3390/s19030503 doi: 10.3390/s19030503
    [141] Breitenstein J, Löhdefink J, Fingscheidt T (2022) Joint Prediction of Amodal and Visible Semantic Segmentation for Automated Driving. In European Conference on Computer Vision, 633-645. Cham: Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-25056-9_40
    [142] Deng L, Cao H, Dong Q, Jiang Y (2023) Semi-supervised lane detection for continuous traffic scenes. Traffic Inj Prev, 1-6. https://doi.org/10.1080/15389588.2023.2219794
    [143] Yao S, Lan F, Chen J (2022) Visual Odometry Integrated Semantic Constraints towards Autonomous Driving (No. 2022-01-7095). SAE Technical Paper. https://doi.org/10.4271/2022-01-7095
    [144] Divakarla U, Bhat R, Madagaonkar SB, Pranav DV, Shyam C, Chandrashekar K (2023) Semantic Segmentation for Autonomous Driving. In Information and Communication Technology for Competitive Strategies (ICTCS 2022) Intelligent Strategies for ICT, 683-694. Singapore: Springer Nature Singapore. https://doi.org/10.1007/978-981-19-9304-6_61
    [145] Kachhoria R, Jaiswal S, Lokhande M, Rodge J (2023) Lane detection and path prediction in autonomous vehicle using deep learning. In Intelligent Edge Computing for Cyber Physical Applications, 111-127. Academic Press. https://doi.org/10.1016/B978-0-323-99412-5.00012-5
    [146] Chen T, Chen A (2022) Road Sign Recognition Method Based on Segmentation and Attention Mechanism. Mob Inform Syst 2022. https://doi.org/10.1155/2022/6389580
    [147] Song C, Tan SJ, Khor A, Cao P, Zhao Y, Li G (2022) Method of Vehicle Behavior Analysis for Real-Time Video Streaming Based on Mobilenet-YOLOV4 and ERFNET. In 2022 IEEE 7th International Conference on Intelligent Transportation Engineering (ICITE), 473-480. IEEE. https://doi.org/10.1109/ICITE56321.2022.10101430
    [148] Ye D, Han R (2022) Image semantic segmentation method based on improved ERFNet model. The Journal of Engineering 2022: 180-190. https://doi.org/10.1049/tje2.12104 doi: 10.1049/tje2.12104
    [149] Zhang L, Jiang F, Yang J, Kong B, Hussain A (2023) A real‐time lane detection network using two‐directional separation attention. Comput‐Aided Civ Inf. https://doi.org/10.1111/mice.13051
    [150] Fan J, Wang F, Chu H, Hu X, Cheng Y, Gao B (2022) Mlfnet: Multi-level fusion network for real-time semantic segmentation of autonomous driving. IEEE Transactions on Intelligent Vehicles 8: 756-767. https://doi.org/10.1109/TIV.2022.3176860 doi: 10.1109/TIV.2022.3176860
    [151] Mullick K, Jain H, Gupta S, Kale AA (2023) Domain Adaptation of Synthetic Driving Datasets for Real-World Autonomous Driving. arXiv preprint arXiv: 2302.04149.
    [152] Zhang L, Jiang F, Yang J, Kong B, Hussain A, Gogate M, et al. (2022) DNet-CNet: A novel cascaded deep network for real-time lane detection and classification. J Amb Intel Hum Comput 14: 10745-10760. https://doi.org/10.1007/s12652-022-04346-2 doi: 10.1007/s12652-022-04346-2
    [153] Florea H, Petrovai A, Giosan I, Oniga F, Varga R, Nedevschi S (2022) Enhanced perception for autonomous driving using semantic and geometric data fusion. Sensors 22: 5061. https://doi.org/10.3390/s22135061 doi: 10.3390/s22135061
    [154] Bouzidi W, Bouaafia S, Hajjaji MA, Bergasa LM, Enhanced U-Net Approach: Semantic Segmentation for Self-Driving Cars Applications.
    [155] PETROVAI A (2022) Deep Learning-based Visual Perception for Autonomous Driving. Doctoral dissertation, Technical University of Cluj-Napoca.
    [156] Breitenstein J, Fingscheidt T (2022) Amodal Cityscapes: A New Dataset, its Generation, and an Amodal Semantic Segmentation Challenge Baseline. In 2022 IEEE Intelligent Vehicles Symposium (IV), 1018-1025. IEEE. https://doi.org/10.1109/IV51971.2022.9827342
    [157] An TH, Kang J, Min KW (2023) Network adaptation for color image semantic segmentation. IET Image Process.
    [158] Karine A, Napoléon T, Jridi M (2022) Semantic Images Segmentation for Autonomous Driving Using Self-Attention Knowledge Distillation. In 2022 16th International Conference on Signal-Image Technology & Internet-Based Systems (SITIS), 198-202. IEEE. https://doi.org/10.1109/SITIS57111.2022.00044
    [159] Yang X, Yu Y, Zhang Z, Huang Y, Liu Z, Niu Z, et al. (2023) Lightweight lane marking detection CNNs by self soft label attention. Multimedia Tools and Applications 82: 5607-5626. https://doi.org/10.1007/s11042-022-13442-6 doi: 10.1007/s11042-022-13442-6
    [160] Chniti H, Mahfoudh M (2022) Designing a Model of Driving Scenarios for Autonomous Vehicles. In Knowledge Science, Engineering and Management: 15th International Conference, KSEM 2022, Singapore, August 6-8, 2022, Proceedings, Part II, 396-405. Cham: Springer International Publishing. https://doi.org/10.1007/978-3-031-10986-7_32
    [161] Zhuang Y, Pu Z, Yang H, Wang Y (2022) Edge-Artificial Intelligence-Powered Parking Surveillance With Quantized Neural Networks. IEEE Intel Transp Syst Mag 14: 107-121. https://doi.org/10.1109/MITS.2022.3182358 doi: 10.1109/MITS.2022.3182358
    [162] Serras B, Gonçalves C, Dias T, Osório AL (2022) Extending the Synoptics of Things (SoT) framework to manage iSoS technology landscapes. In 2022 International Young Engineers Forum (YEF-ECE), 80-85. IEEE. https://doi.org/10.1109/YEF-ECE55092.2022.9849899
    [163] Atar S, Singh S, Agrawal S, Chaurasia R, Sule S, Gadamsetty S, et al. (2022) LCPP: Low Computational Processing Pipeline for Delivery Robots. ICAART (3), 130-138. https://doi.org/10.5220/0010786300003116
    [164] Geldhauser C, Matt AD, Stussak C (2022) I AM AI Gradient Descent-an Open-Source Digital Game for Inquiry-Based CLIL Learning. In Proceedings of the AAAI Conference on Artificial Intelligence 36: 12751-12757). https://doi.org/10.1609/aaai.v36i11.21553
    [165] Li J, Xu W, Deng L, Xiao Y, Han Z, Zheng H (2023) Deep learning for visual recognition and detection of aquatic animals: A review. Rev Aquacult 15: 409-433. https://doi.org/10.1111/raq.12726 doi: 10.1111/raq.12726
    [166] Ramalingam S, A Study and Review of Classical, Machine Learning and Deep Learning Methods of Software Reliability Estimation for Safety-Critical Systems.
    [167] Gamal O, Imran M, Roth H, Wahrburg J (2020) Assistive parking systems knowledge transfer to end-to-end deep learning for autonomous parking. In 2020 6th International conference on mechatronics and robotics engineering (ICMRE), 216-221. IEEE. https://doi.org/10.1109/ICMRE49073.2020.9065014
    [168] Kashyap A, Iqbal M, Pattabiraman K, Seltzer M (2021) ReLUSyn: Synthesizing Stealthy Attacks for Deep Neural Network Based Cyber-Physical Systems. arXiv preprint arXiv: 2105.10393.
    [169] Heinen, MR, Osório FS, Heinen FJ, Kelber C (2006) Seva3d: Using arti cial neural networks to autonomous vehicle parking control. In The 2006 IEEE International Joint Conference on Neural Network Proceedings, 4704-4711. IEEE. https://doi.org/10.1109/IJCNN.2006.247124
    [170] Heinen MR, Osório FS, Heinen FJ, Kelber C (2006) Autonomous vehicle parking and pull out using artificial neural networks. In Proceedings of the I Workshop on Computational Intelligence (WCI).
    [171] Min C, Xu J, Xiao L, Zhao D, Nie Y, Dai B (2021) Attentional graph neural network for parking-slot detection. IEEE Robot Autom Lett 6: 3445-3450. https://doi.org/10.1109/LRA.2021.3064270 doi: 10.1109/LRA.2021.3064270
    [172] Zhang W, Liu H, Liu Y, Zhou J, Xiong H (2020) Semi-supervised hierarchical recurrent graph neural network for city-wide parking availability prediction. In Proceedings of the AAAI Conference on Artificial Intelligence 34: 1186-1193. https://doi.org/10.1609/aaai.v34i01.5471
    [173] Park J, Chun J, Kim SH, Kim Y, Park J (2021) Learning to schedule job-shop problems: representation and policy learning using graph neural network and reinforcement learning. Int J Prod Res 59: 3360-3377. https://doi.org/10.1080/00207543.2020.1870013 doi: 10.1080/00207543.2020.1870013
    [174] Lee H, Lee S, Kim J, Jung H, Yoon KJ, Gandla S, et al. (2023) Stretchable array electromyography sensor with graph neural network for static and dynamic gestures recognition system. npj Flex Electron 7: 20. https://doi.org/10.1038/s41528-023-00246-3 doi: 10.1038/s41528-023-00246-3
    [175] Meyer E, Brenner M, Zhang B, Schickert M, Musani B, Althoff M (2023) Geometric deep learning for autonomous driving: Unlocking the power of graph neural networks with CommonRoad-Geometric. 2023 IEEE Intelligent Vehicles Symposium (IV), 1-8. https://doi.org/10.1109/IV55152.2023.10186741
    [176] Singh D, Srivastava R (2022) Graph Neural Network with RNNs based trajectory prediction of dynamic agents for autonomous vehicle. Appl Intell 52: 12801-12816. https://doi.org/10.1007/s10489-021-03120-9 doi: 10.1007/s10489-021-03120-9
    [177] Singh D, Srivastava R (2022) Multi-scale graph-transformer network for trajectory prediction of the autonomous vehicles. Intel Serv Robot 15: 307-320. https://doi.org/10.1007/s11370-022-00422-w doi: 10.1007/s11370-022-00422-w
    [178] Klimke M, Völz B, Buchholz M (2022) Cooperative Behavior Planning for Automated Driving using Graph Neural Networks. In 2022 IEEE Intelligent Vehicles Symposium (IV), 167-174. IEEE. https://doi.org/10.1109/IV51971.2022.9827230
    [179] Lee D, Gu Y, Hoang J, Marchetti-Bowick M (2019) Joint interaction and trajectory prediction for autonomous driving using graph neural networks. arXiv preprint arXiv: 1912.07882.
    [180] Jin K, Wang H, Liu C, Zhai Y, Tang L (2022) Graph neural network based relation learning for abnormal perception information detection in self-driving scenarios. In 2022 International Conference on Robotics and Automation (ICRA), 8943-8949. IEEE. https://doi.org/10.1109/ICRA46639.2022.9812411
    [181] Yang F, Li X, Liu Q, Li Z, Gao X (2022) Generalized single-vehicle-based graph reinforcement learning for decision-making in autonomous driving. Sensors 22: 4935. https://doi.org/10.3390/s22134935 doi: 10.3390/s22134935
    [182] Cao D, Li J, Ma H, Tomizuka M (2021) Spectral temporal graph neural network for trajectory prediction. In 2021 IEEE International Conference on Robotics and Automation (ICRA), 1839-1845. IEEE. https://doi.org/10.1109/ICRA48506.2021.9561461
    [183] Diehl F, Brunner T, Le MT, Knoll A (2019) Graph neural networks for modelling traffic participant interaction. In 2019 IEEE Intelligent Vehicles Symposium (IV), 695-701. IEEE. https://doi.org/10.1109/IVS.2019.8814066
    [184] Ma C, Li Y, Yang F, Zhang Z, Zhuang Y, Jia H, et al. (2019) Deep association: End-to-end graph-based learning for multiple object tracking with conv-graph neural network. In Proceedings of the 2019 on International Conference on Multimedia Retrieval, 253-261.
    [185] Wang L, Zhang X, Zeng W, Liu W, Yang L, Li J, et al. (2022) Global perception-based robust parking space detection using a low-cost camera. IEEE Transactions on Intelligent Vehicles 8: 1439-1448. https://doi.org/10.1109/TIV.2022.3186035 doi: 10.1109/TIV.2022.3186035
    [186] Shi R, Yang S, Chen Y, Wang R, Zhang M, Lu J, et al. (2023) CNN‐Transformer for visual‐tactile fusion applied in road recognition of autonomous vehicles. Pattern Recogn Lett 166: 200-208. https://doi.org/10.1016/j.patrec.2022.11.023 doi: 10.1016/j.patrec.2022.11.023
    [187] Singh D, Srivastava R (2022) Multi-scale graph-transformer network for trajectory prediction of the autonomous vehicles. Intel Serv Robot 15: 307-320. https://doi.org/10.1007/s11370-022-00422-w doi: 10.1007/s11370-022-00422-w
    [188] Zhang H, Yang Z, Xiong H, Zhu T, Long Z, Wu W (2023) Transformer Aided Adaptive Extended Kalman Filter for Autonomous Vehicle Mass Estimation. Processes 11: 887. https://doi.org/10.3390/pr11030887 doi: 10.3390/pr11030887
    [189] Li G, Qiu Y, Yang Y, Li Z, Li S, Chu W, et al. (2022) Lane change strategies for autonomous vehicles: a deep reinforcement learning approach based on transformer. IEEE Transactions on Intelligent Vehicles.
    [190] Rafiq G, Rafiq M, Choi, GS (2023) Spectral representation learning and fusion for autonomous vehicles trip description exploiting recurrent transformer. IEEE Access. https://doi.org/10.1109/ACCESS.2023.3287783
    [191] Shao H, Wang L, Chen R, Li H, Liu Y (2023) Safety-enhanced autonomous driving using interpretable sensor fusion transformer. In Conference on Robot Learning, 726-737. PMLR.
    [192] Tseng CH, Zhang J, Sun MT, Sakai K, Ku WS (2022) Multi-modal Transformer Path Prediction for Autonomous Vehicle. arXiv preprint arXiv: 2208.07256.
    [193] Hu H, Wang Q, Zhang Z, Li Z, Gao Z (2023) Holistic transformer: A joint neural network for trajectory prediction and decision-making of autonomous vehicles. Pattern Recogn 141: 109592. https://doi.org/10.1016/j.patcog.2023.109592 doi: 10.1016/j.patcog.2023.109592
    [194] Mozaffari S, Koufos K, Dianati M (2023) Multimodal Manoeuvre and Trajectory Prediction for Autonomous Vehicles Using Transformer Networks. IEEE Robot Autom Lett 8: 6123-6130. https://doi.org/10.1109/LRA.2023.3301720 doi: 10.1109/LRA.2023.3301720
    [195] Tian Y, Wang J, Wang Y, Zhao C, Yao F, Wang X (2022) Federated vehicular transformers and their federations: Privacy-preserving computing and cooperation for autonomous driving. IEEE Transactions on Intelligent Vehicles. https://doi.org/10.1109/TIV.2022.3197815
    [196] Xu R, Xiang H, Tu Z, Xia X, Yang MH, Ma J (2022) V2X-ViT: Vehicle-to-everything cooperative perception with vision transformer. In Computer Vision-ECCV 2022: 17th European Conference, 107-124. Cham: Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-19842-7_7
    [197] Chen W, Wang F, Sun H (2021) S2tnet: Spatio-temporal transformer networks for trajectory prediction in autonomous driving. In Asian Conference on Machine Learning 454-469. PMLR.
    [198] Postnikov A, Gamayunov A, Ferrer G (2021) Transformer based trajectory prediction. arXiv preprint arXiv: 2112.04350.
    [199] Ngiam J, Caine B, Vasudevan V, Zhang Z, Chiang HT, Ling J, et al. (2021) Scene Transformer: A unified architecture for predicting multiple agent trajectories. arXiv preprint arXiv: 2106.08417.
    [200] Khosyi'in M, Budisusila EN, Prasetyowati SAD, Suprapto BY, Nawawi Z (2021) Design of Autonomous Vehicle Navigation Using GNSS Based on Pixhawk 2.1. In 2021 8th International Conference on Electrical Engineering, Computer Science and Informatics (EECSI), 175-180. IEEE. https://doi.org/10.23919/EECSI53397.2021.9624244
    [201] Schütz A, Sánchez-Morales DE, Pany T (2020) Precise positioning through a loosely-coupled sensor fusion of GNSS-RTK, INS and LiDAR for autonomous driving. In 2020 IEEE/ION Position, Location and Navigation Symposium (PLANS), 219-225. IEEE. https://doi.org/10.1109/PLANS46316.2020.9109934
    [202] Swaminathan HB, Sommer A, Becker A, Atzmueller M (2022) Performance Evaluation of GNSS Position Augmentation Methods for Autonomous Vehicles in Urban Environments. Sensors 22: 8419. https://doi.org/10.3390/s22218419 doi: 10.3390/s22218419
    [203] Elsayed H, El-Mowafy A, Wang K (2023) Bounding of correlated double-differenced GNSS observation errors using NRTK for precise positioning of autonomous vehicles. Measurement 206: 112303. https://doi.org/10.1016/j.measurement.2022.112303 doi: 10.1016/j.measurement.2022.112303
    [204] Jianghui GE, Hua CH, Jiang GU, Guangcai LI, Na WE (2020) Three multi-frequency and multi-system GNSS high-precision point positioning methods and their performance in complex urban environment. Acta Geodaetica et Cartographica Sinica 49: 1.
    [205] Lee W, Geneva P, Yang Y, Huang G (2022) Tightly-coupled GNSS-aided Visual-Inertial Localization. In 2022 International Conference on Robotics and Automation (ICRA), 9484-9491. IEEE. https://doi.org/10.1109/ICRA46639.2022.9811362
    [206] Wen W, Hsu LT (2021) 3D LiDAR aided GNSS real-time kinematic positioning. In Proceedings of the 34th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2021), 2212-2220. https://doi.org/10.33012/2021.18072
    [207] Li T, Zhang H, Gao Z, Chen Q, Niu X (2018) High-accuracy positioning in urban environments using single-frequency multi-GNSS RTK/MEMS-IMU integration. Remote sensing 10: 205. https://doi.org/10.3390/rs10020205 doi: 10.3390/rs10020205
    [208] Jia M, Lee H, Khalife J, Kassas ZM, Seo J (2021) Ground vehicle navigation integrity monitoring for multi-constellation GNSS fused with cellular signals of opportunity. In 2021 IEEE International Intelligent Transportation Systems Conference (ITSC), 3978-3983. IEEE. https://doi.org/10.1109/ITSC48978.2021.9564686
    [209] Sadli R, Afkir M, Hadid A, Rivenq A, Taleb-Ahmed A (2022) Map-Matching-Based Localization Using Camera and Low-Cost GPS for Lane-Level Accuracy. Sensors 22: 2434. https://doi.org/10.3390/s22072434 doi: 10.3390/s22072434
    [210] Somogyi H, Soumelidis A (2020) Comparison of High-Precision GNSS systems for development of an autonomous localization system. In 2020 23rd International Symposium on Measurement and Control in Robotics (ISMCR), 1-6. IEEE. https://doi.org/10.1109/ISMCR51255.2020.9263762
    [211] Geng J, Guo J, Chang H, Li X (2019) Toward global instantaneous decimeter-level positioning using tightly coupled multi-constellation and multi-frequency GNSS. J Geodesy 93: 977-991. https://doi.org/10.1007/s00190-018-1219-y doi: 10.1007/s00190-018-1219-y
    [212] Liu S (2020) Engineering autonomous vehicles and robots: the dragonfly modular-based approach. John Wiley & Sons. https://doi.org/10.1002/9781119570516
    [213] Meng Q, Hsu LT (2021) Integrity for autonomous vehicles and towards a novel alert limit determination method. Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering 235: 996-1006. https://doi.org/10.1177/0954407020965760 doi: 10.1177/0954407020965760
    [214] Abosekeen A, Noureldin A, Korenberg MJ (2019) Improving the RISS/GNSS land-vehicles integrated navigation system using magnetic azimuth updates. IEEE T Intell Transp Syst 21: 1250-1263. https://doi.org/10.1109/TITS.2019.2905871 doi: 10.1109/TITS.2019.2905871
    [215] Rodriguez-Solano C, Nick T, Gleb Z, Xiaoming C, Ken D, Lorenz G (2021) Protection level of the trimble RTX positioning engine for autonomous applications. In Proceedings of the 34th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2021), 1577-1595. https://doi.org/10.33012/2021.17889
    [216] Bressler J, Reisdorf P, Obst M, Wanielik G (2016) GNSS positioning in non-line-of-sight context—A survey. In 2016 IEEE 19th international conference on intelligent transportation systems (ITSC), 1147-1154. IEEE. https://doi.org/10.1109/ITSC.2016.7795701
    [217] Patel RH, Härri J, Bonnet C (2017) Impact of localization errors on automated vehicle control strategies. In 2017 IEEE Vehicular Networking Conference (VNC), 61-68. IEEE. https://doi.org/10.1109/VNC.2017.8275649
    [218] Tao Z, Bonnifait P (2016) Sequential data fusion of GNSS pseudoranges and Dopplers with map-based vision systems. IEEE Transactions on Intelligent Vehicles 1: 254-265. https://doi.org/10.1109/TIV.2017.2658185 doi: 10.1109/TIV.2017.2658185
    [219] Kato S, Tokunaga S, Maruyama Y, Maeda S, Hirabayashi M, Kitsukawa Y, et al. (2018) Autoware on board: Enabling autonomous vehicles with embedded systems. In 2018 ACM/IEEE 9th International Conference on Cyber-Physical Systems (ICCPS), 287-296. IEEE. https://doi.org/10.1109/ICCPS.2018.00035
    [220] Kato S, Takeuchi E, Ishiguro Y, Ninomiya Y, Takeda K, Hamada T (2015) An open approach to autonomous vehicles. IEEE Micro 35: 60-68. https://doi.org/10.1109/MM.2015.133 doi: 10.1109/MM.2015.133
    [221] Raju VM, Gupta V, Lomate S (2019) Performance of open autonomous vehicle platforms: Autoware and Apollo. In 2019 IEEE 5th International Conference for Convergence in Technology (I2CT), 1-5. IEEE. https://doi.org/10.1109/I2CT45611.2019.9033734
    [222] Tsukada M, Oi T, Ito A, Hirata M, Esaki H (2020) AutoC2X: Open-source software to realize V2X cooperative perception among autonomous vehicles. In 2020 IEEE 92nd Vehicular Technology Conference (VTC2020-Fall), 1-6. IEEE. https://doi.org/10.1109/VTC2020-Fall49728.2020.9348525
    [223] Kawabata N, Kuwabara Y, Kawasaki T (2021) Self-Localization of Autonomous Car Using Autoware. IEICE Technical Report 120: 103-108.
    [224] Carballo A, Wong D, Ninomiya Y, Kato S, Takeda K (2019) Training engineers in autonomous driving technologies using autoware. In 2019 IEEE Intelligent Transportation Systems Conference (ITSC), 3347-3354. IEEE. https://doi.org/10.1109/ITSC.2019.8917152
    [225] Dhakal S, Qu D, Carrillo D, Yang Q, Fu S (2021) Oasd: An open approach to self-driving vehicle. In 2021 Fourth International Conference on Connected and Autonomous Driving (MetroCAD), 54-61. IEEE. https://doi.org/10.1109/MetroCAD51599.2021.00017
    [226] Tun WN, Kim S, Lee JW, Darweesh H (2019) Open-source tool of vector map for path planning in autoware autonomous driving software. 2019 IEEE International Conference on Big Data and Smart Computing (BigComp), 1-3. IEEE. https://doi.org/10.1109/BIGCOMP.2019.8679340
    [227] Rong G, Shin BH, Tabatabaee H, Lu Q, Lemke S, Možeiko M, et al. (2020) Lgsvl simulator: A high fidelity simulator for autonomous driving. In 2020 IEEE 23rd International conference on intelligent transportation systems (ITSC), 1-6. IEEE. https://doi.org/10.1109/ITSC45102.2020.9294422
    [228] Akai N, Morales LY, Yamaguchi T, Takeuchi E, Yoshihara Y, Okuda H, et al. (2017) Autonomous driving based on accurate localization using multilayer LiDAR and dead reckoning. In 2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC), 1-6. IEEE. https://doi.org/10.1109/ITSC.2017.8317797
    [229] Garcia J, Feng Y, Shen J, Almanee S, Xia Y, Chen AQA (2020) A comprehensive study of autonomous vehicle bugs. In Proceedings of the ACM/IEEE 42nd international conference on software engineering, 385-396. https://doi.org/10.1145/3377811.3380397
    [230] Tsukada M, Oi T, Kitazawa M, Esaki H (2020) Networked roadside perception units for autonomous driving. Sensors 20: 5320. https://doi.org/10.3390/s20185320 doi: 10.3390/s20185320
    [231] Chishiro H, Suito K, Ito T, Maeda S, Azumi T, Funaoka K, et al. (2019) Towards heterogeneous computing platforms for autonomous driving. In 2019 IEEE International Conference on Embedded Software and Systems (ICESS), 1-8. IEEE. https://doi.org/10.1109/ICESS.2019.8782446
    [232] Pang S, Kent D, Cai X, Al-Qassab H, Morris D, Radha H (2018) 3d scan registration based localization for autonomous vehicles-a comparison of ndt and icp under realistic conditions. In 2018 IEEE 88th vehicular technology conference (VTC-Fall), 1-5. IEEE. https://doi.org/10.1109/VTCFall.2018.8690819
    [233] Munir F, Azam S, Sheri AM, Ko Y, Jeon M (2019) Where Am I: Localization and 3D Maps for Autonomous Vehicles. In VEHITS, 452-457. https://doi.org/10.5220/0007718400002179
    [234] Wen W, Hsu LT, Zhang G (2018) Performance analysis of NDT-based graph SLAM for autonomous vehicle in diverse typical driving scenarios of Hong Kong. Sensors 18: 3928. https://doi.org/10.3390/s18113928 doi: 10.3390/s18113928
    [235] Lin X, Wang F, Yang B, Zhang W (2021). Autonomous vehicle localization with prior visual point cloud map constraints in GNSS-challenged environments. Remote Sensing 13: 506. https://doi.org/10.3390/rs13030506 doi: 10.3390/rs13030506
    [236] Akai N, Morales LY, Takeuchi E, Yoshihara Y, Ninomiya Y (2017) Robust localization using 3D NDT scan matching with experimentally determined uncertainty and road marker matching. In 2017 IEEE Intelligent Vehicles Symposium (IV), 1356-1363. IEEE. https://doi.org/10.1109/IVS.2017.7995900
    [237] Akai N, Morales LY, Yamaguchi T, Takeuchi E, Yoshihara Y, Okuda H, et al. (2017) Autonomous driving based on accurate localization using multilayer LiDAR and dead reckoning. In 2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC), 1-6. IEEE. https://doi.org/10.1109/ITSC.2017.8317797
    [238] Li Q, Queralta JP, Gia TN, Zou Z, Westerlund T (2020) Multi-sensor fusion for navigation and mapping in autonomous vehicles: Accurate localization in urban environments. Unmanned Systems 8: 229-237. https://doi.org/10.1142/S2301385020500168 doi: 10.1142/S2301385020500168
    [239] Saarinen J, Andreasson H, Stoyanov T, Lilienthal, AJ (2013) Normal distributions transform Monte-Carlo localization (NDT-MCL). In 2013 IEEE/RSJ international conference on intelligent robots and systems, 382-389. IEEE. https://doi.org/10.1109/IROS.2013.6696380
    [240] Ahmed SZ, Saputra VB, Verma S, Zhang K, Adiwahono AH (2019) Sparse-3D lidar outdoor map-based autonomous vehicle localization. In 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 1614-1619. IEEE. https://doi.org/10.1109/IROS40897.2019.8967596
    [241] Elhousni M, Huang X (2020) A survey on 3d lidar localization for autonomous vehicles. In 2020 IEEE Intelligent Vehicles Symposium (IV), 1879-1884. IEEE. https://doi.org/10.1109/IV47402.2020.9304812
    [242] Srinara S, Lee CM, Tsai S, Tsai GJ, Chiang KW (2021) Performance analysis of 3D NDT scan matching for autonomous vehicles using INS/GNSS/3D LiDAR-SLAM integration scheme. In 2021 IEEE International Symposium on Inertial Sensors and Systems (INERTIAL), 1-4. IEEE. https://doi.org/10.1109/INERTIAL51137.2021.9430476
    [243] Poulose A, Baek M, Han DS (2022) Point Cloud Map Generation and Localization for Autonomous Vehicles Using 3D Lidar Scans. In 2022 27th Asia Pacific Conference on Communications (APCC), 336-341. IEEE. https://doi.org/10.1109/APCC55198.2022.9943630
    [244] Javanmardi E, Javanmardi M, Gu Y, Kamijo S (2020) Pre-estimating self-localization error of NDT-based map-matching from map only. IEEE T Intell Transp Syst 22: 7652-7666. https://doi.org/10.1109/TITS.2020.3006854 doi: 10.1109/TITS.2020.3006854
    [245] Javanmardi E, Javanmardi M, Gu Y, Kamijo S (2018) November. Adaptive resolution refinement of NDT map based on localization error modeled by map factors. 2018 21st International Conference on Intelligent Transportation Systems (ITSC), 2237-2243. IEEE. https://doi.org/10.1109/ITSC.2018.8569236
    [246] Jang KW, Jeong WJ, Kang Y (2022) Development of a GPU-Accelerated NDT Localization Algorithm for GNSS-Denied Urban Areas. Sensors 22: 1913. https://doi.org/10.3390/s22051913 doi: 10.3390/s22051913
    [247] Javanmardi E, Javanmardi M, Gu Y, Kamijo S (2017) Autonomous vehicle self-localization based on multilayer 2D vector map and multi-channel LiDAR. 2017 IEEE Intelligent Vehicles Symposium (IV), 437-442. IEEE. https://doi.org/10.1109/IVS.2017.7995757
    [248] Wen W, Zhan W, Hsu LT (2019) Robust Localization Using 3D NDT Matching and Beam Model for Autonomous Vehicles in an Urban Scenario with Dynamic Obstacles. Proceedings of Mobile Mapping Technology, Shenzhen, China.
    [249] Javanmardi E, Gu Y, Javanmardi M, Kamijo S (2019) Autonomous vehicle self-localization based on abstract map and multi-channel LiDAR in urban area. IATSS research 43: 1-13. https://doi.org/10.1016/j.iatssr.2018.05.001 doi: 10.1016/j.iatssr.2018.05.001
    [250] Kan YC, Hsu LT, Chung E (2021) Performance evaluation on map-based NDT scan matching localization using simulated occlusion datasets. IEEE Sensors Letters 5: 1-4. https://doi.org/10.1109/LSENS.2021.3060097 doi: 10.1109/LSENS.2021.3060097
    [251] Fayyad J, Jaradat MA, Gruyer D, Najjaran H (2020) Deep learning sensor fusion for autonomous vehicle perception and localization: A review. Sensors 20: 4220. https://doi.org/10.3390/s20154220 doi: 10.3390/s20154220
    [252] Laconte J, Kasmi A, Aufrère R, Vaidis M, Chapuis R (2021) A survey of localization methods for autonomous vehicles in highway scenarios. Sensors 22: 247. https://doi.org/10.3390/s22010247 doi: 10.3390/s22010247
    [253] Spangenberg R, Goehring D, Rojas R (2016) Pole-based localization for autonomous vehicles in urban scenarios. 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2161-2166. IEEE. https://doi.org/10.1109/IROS.2016.7759339
    [254] Reid TG, Houts SE, Cammarata R, Mills G, Agarwal S, Vora A, et al. (2019) Localization requirements for autonomous vehicles. SAE Intl J CAV 2: 173-190. https://doi.org/10.4271/12-02-03-0012 doi: 10.4271/12-02-03-0012
    [255] Elhousni M, Huang X (2020) October. A survey on 3d lidar localization for autonomous vehicles. 2020 IEEE Intelligent Vehicles Symposium (IV), 1879-1884. IEEE. https://doi.org/10.1109/IV47402.2020.9304812
    [256] de Miguel MÁ, García F, Armingol JM (2020) Improved LiDAR probabilistic localization for autonomous vehicles using GNSS. Sensors 20: 3145. https://doi.org/10.3390/s20113145 doi: 10.3390/s20113145
    [257] Wang L, Zhang Y, Wang J (2017) Map-based localization method for autonomous vehicles using 3D-LIDAR. IFAC-PapersOnLine 50: 276-281. https://doi.org/10.1016/j.ifacol.2017.08.046 doi: 10.1016/j.ifacol.2017.08.046
    [258] Meng X, Wang H, Liu B (2017) A robust vehicle localization approach based on gnss/imu/dmi/lidar sensor fusion for autonomous vehicles. Sensors 17: 2140. https://doi.org/10.3390/s17092140 doi: 10.3390/s17092140
    [259] Kamijo S, Gu Y, Hsu L (2015) Autonomous vehicle technologies: Localization and mapping. Fundam Rev 9: 131-141, 2015. https://doi.org/10.1587/essfr.9.2_131
    [260] Lin X, Wang F, Yang B, Zhang W (2021) Autonomous vehicle localization with prior visual point cloud map constraints in GNSS-challenged environments. Remote Sensing 13: 506. https://doi.org/10.3390/rs13030506 doi: 10.3390/rs13030506
    [261] Werries A, Dolan J (2016) Adaptive Kalman filtering methods for low-cost GPS/INS localization for autonomous vehicles (No. CMU-RI-TR-16-18). Carnegie-Mellon University.
    [262] Jalal F, Nasir F (2021) Underwater navigation, localization and path planning for autonomous vehicles: A review. 2021 International Bhurban Conference on Applied Sciences and Technologies (IBCAST), 817-828. IEEE. https://doi.org/10.1109/IBCAST51254.2021.9393315
    [263] Luo Q, Cao Y, Liu J, Benslimane A (2019) Localization and navigation in autonomous driving: Threats and countermeasures. IEEE Wirel Commun 26: 38-45. https://doi.org/10.1109/MWC.2019.1800533 doi: 10.1109/MWC.2019.1800533
    [264] Wang H, Xue C, Zhou Y, Wen F, Zhang H (2021) Visual semantic localization based on hd map for autonomous vehicles in urban scenarios. 2021 IEEE International Conference on Robotics and Automation (ICRA), 11255-11261. IEEE. https://doi.org/10.1109/ICRA48506.2021.9561459
    [265] Park M, Kang Y (2021) Experimental verification of a drift controller for autonomous vehicle tracking: A circular trajectory using LQR method. Int J Control Autom Syst 19: 404-416. https://doi.org/10.1007/s12555-019-0757-2 doi: 10.1007/s12555-019-0757-2
    [266] Pang H, Liu N, Hu C, Xu Z (2022) A practical trajectory tracking control of autonomous vehicles using linear time-varying MPC method. Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering 236: 709-723. https://doi.org/10.1177/09544070211022904 doi: 10.1177/09544070211022904
    [267] Borrelli F, Morari M (2007) Offset free model predictive control. 2007 46th IEEE conference on decision and control, 1245-1250. IEEE. https://doi.org/10.1109/CDC.2007.4434770
    [268] Cheng S, Li L, Chen X, Wu J (2020) Model-predictive-control-based path tracking controller of autonomous vehicle considering parametric uncertainties and velocity-varying. IEEE T Ind Electron 68: 8698-8707. https://doi.org/10.1109/TIE.2020.3009585 doi: 10.1109/TIE.2020.3009585
    [269] Williams G, Drews P, Goldfain B, Rehg JM, Theodorou EA (2018) Information-theoretic model predictive control: Theory and applications to autonomous driving. IEEE T Robot 34: 1603-1622. https://doi.org/10.1109/TRO.2018.2865891 doi: 10.1109/TRO.2018.2865891
    [270] Petrovskaya A, Thrun S (2008) Model based vehicle tracking for autonomous driving in urban environments. Proceedings of robotics: science and systems IV, Zurich, Switzerland, 34. https://doi.org/10.15607/RSS.2008.IV.023
    [271] Galceran E, Olson E, Eustice RM (2015) Augmented vehicle tracking under occlusions for decision-making in autonomous driving. 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 3559-3565. IEEE. https://doi.org/10.1109/IROS.2015.7353874
    [272] Wang H, Wang B, Liu B, Meng X, Yang G (2017) Pedestrian recognition and tracking using 3D LiDAR for autonomous vehicle. Robot Auton Syst 88: 71-78. https://doi.org/10.1016/j.robot.2016.11.014 doi: 10.1016/j.robot.2016.11.014
    [273] Kothare MV, Balakrishnan V, Morari M (1996) Robust constrained model predictive control using linear matrix inequalities. Automatica 32: 1361-1379. https://doi.org/10.1016/0005-1098(96)00063-5 doi: 10.1016/0005-1098(96)00063-5
    [274] Falcone P, Borrelli F, Tseng HE, Asgari J, Hrovat D (2008) Linear time‐varying model predictive control and its application to active steering systems: Stability analysis and experimental validation. International Journal of Robust and Nonlinear Control: IFAC‐Affiliated Journal 18: 862-875. https://doi.org/10.1002/rnc.1245 doi: 10.1002/rnc.1245
    [275] Wang Y, Shao Q, Zhou J, Zheng H, Chen H (2020). Longitudinal and lateral control of autonomous vehicles in multi-vehicle driving environments. IET Intell Transp Syst 14: 924-935. https://doi.org/10.1049/iet-its.2019.0846 doi: 10.1049/iet-its.2019.0846
    [276] Cui J, Liew LS, Sabaliauskaite G, Zhou F (2019) A Review on Safety Failures, Security Attacks, and Available Countermeasures for Autonomous Vehicles. Ad Hoc Networks 90: 101823. https://doi.org/10.1016/j.adhoc.2018.12.006 doi: 10.1016/j.adhoc.2018.12.006
    [277] Ferdowsi A, Challita U, Saad W, Mandayam NB (2018) Robust Deep Reinforcement Learning for Security and Safety in Autonomous Vehicle Systems. 2018 21st International Conference on Intelligent Transportation Systems (ITSC), 307-312. https://doi.org/10.1109/ITSC.2018.8569635
    [278] Xu W, Yan C, Jia W, Ji X, Liu J (2018) Analyzing and Enhancing the Security of Ultrasonic Sensors for Autonomous Vehicles. IEEE Internet Things 5: 5015-5029. https://doi.org/10.1109/JIOT.2018.2867917 doi: 10.1109/JIOT.2018.2867917
  • Reader Comments
  • © 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(5793) PDF downloads(334) Cited by(2)

Figures and Tables

Tables(4)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog