Operation Of Autonomous Vehicles Research Articles

Cost-Effectiveness of Introducing Autonomous Trucks: From the Perspective of the Total Cost of Operation in Logistics

This study explored the economic influence of introducing fully autonomous trucks (ATs) on the total cost of operation (TCO) from the viewpoint of freight transport operators. We utilized the cost–benefit analysis and scenario planning frameworks using a case study of a middle-mile route between a port and a large logistics distribution center in South Korea, where an autonomous truck pilot project has been running for several years. We employed a data source obtained from Hanjin Transportation Co., LTD., which enables us to reflect the actual expenses of operating freight transportation in the context of the study route. The findings of this study indicate that ATs would attain substantial operation cost savings for freight transport operators across all five scenarios, ranging from the most pessimistic to the most optimistic. Specifically, results of the base scenario reveal that monthly TCO decreases by 56.25, 47.04, 42.97, and 41.73 percent for 1-, 5-, 12-, and 25-ton trucks, respectively. The cost reduction per month was USD 2399 for a 1-ton truck, USD 2891 for a 5-ton truck, and USD 3438 for a 12-ton truck. Even in the most pessimistic scenario, the operation cost savings for operating with ATs in the route are USD 1415, USD 1642, USD 1985, and USD 2222 per month for 1-, 5-, 12-, and 25-ton trucks, respectively. When factoring in changes in cost, the wage reduction for vehicle drivers was substantial (USD 2430, USD 2835, USD 3240, and USD 3645 for 1-, 5-, 12-, and 25-ton trucks, respectively). However, remote operation was a new cost factor that increased TCO, which increased by USD 243,384,357 in the base case. This study’s findings have multiple implications, including (1) informing economic efficiency and productivity of operating autonomous vehicles (AVs), (2) providing insights on indicative cost estimates for AV operations, (3) providing valuable information for making informed investment decisions, and (4) supporting the notion that freight transport is an especially attractive market for AVs.

Decision Support System for autonomous vehicle maintenance management

Autonomous vehicles are complex objects equipped with the tools for monitoring the technical condition and transmitting data for diagnostics and prediction. The design of new models of autonomous and robotic vehicles is inextricably linked with the development of a maintenance system. The efficient functioning of the maintenance system is ensured by the use of intelligent technologies and digital twins. The article proposes an approach to designing a decision support system for the maintenance and repair of a fleet of autonomous vehicles. The decision support system is focused on robotic agricultural vehicles. The proposed structure of the system is of a general nature and can be used in transport and logistics enterprises in various sectors of the economy. The general architecture of the maintenance system is presented, including the neuro-digital twin of the vehicle, the analytics unit, simulation models of operation processes, and service centers. The components of the neuro-digital twin are digital twins of car units, a knowledge base, domain ontology, artificial neural networks, and a team of experts. The proposed approach is based on a combination of intelligent technologies and simulation modeling in collaboration with a team of experts. A set of original simulation models for the operation of a fleet of autonomous vehicles has been developed. New models based on stochastic timed colored Petri nets are proposed to analyze the processes during the autonomous vehicle operation. The developed decision support system can be used both at the stage of virtual commissioning and in the real operation of autonomous vehicles.

A Comparative Study of Object Detection using YOLO and SSD Algorithms

One of the most significant and difficult areas of computer vision is object detection, which is used in many aspects of daily life, including security monitoring, autonomous vehicle operation, and so on, to find instances of semantic objects belonging to a particular class. In this survey, we first look at the current approaches of typical detection models and discuss the benchmark datasets in order to comprehensively and deeply comprehend the major development status of the object detection pipeline. This project explains how convolutional neural network-based deep learning techniques are used for object detection. In this study, deep learning methods for cutting-edge object identification systems are evaluated. Any kind of visual medium is represented by a computer as a collection of numerical numbers. In order to inspect the contents of images, they need image processing techniques. In order to determine which is the quickest and most effective object detection algorithm, this research analyzes two popular algorithms: Single Shot Detection (SSD) and You Only Look Once (YOLO). The COCO (Common Object in Context) dataset is used in this comparative research to assess the effectiveness of the two algorithms and to analyze their strengths and weaknesses in terms of factors like accuracy, precision, and speed. Based on the findings analysis, it can be said that YOLO-v3 outperforms SSD in the same testing scenario. Keywords- Object Detection, Deep learning, CNN, COCO, YOLO, SSD

MMH-STA: A Macro-Micro-Hierarchical Spatio-Temporal Attention Method for Multi-Agent Trajectory Prediction in Unsignalized Roundabouts

Simultaneously predicting the future trajectories of heterogeneous multiple agents in a neighborhood is crucial for ensuring the safe and efficient operations of autonomous vehicles in intelligent transportation systems. The unsignalized roundabout is a typical traffic scenario wherein the multivariate and complex information interactions make the prediction task challenging. To address the aforementioned challenges, a macro-micro-hierarchical spatio-temporal attention (MMH-STA) architecture, which can effectively extract the temporal and spatial features of multiple agents based on the interaction mechanism, is presented in this paper. This work makes three contributions: 1) A novel hierarchical framework, which considers the heterogeneity of different types of agents, is proposed for trajectory prediction in the roundabout environment. Similarly, the macro-state for interaction with the roundabout structure and the micro-state for interaction with other agents are introduced for an agent. 2) A heterogeneous graph is devised to represent the spatial interactions of a multi-agent, which is reflected in connections between different types of agents and the properties of nodes and edges in the graph. 3) A novel heterogeneous graph attention network with a multi-order neighborhood is designed to describe the spatial feature interactions in the neighborhood. Finally, a characterized decoder forecasts the future trajectories of multiple agents concomitantly. The experimental results reveal that the proposed model can effectively implement multi-agent trajectory prediction in roundabout scenarios with high accuracy and state-of-the-art performance compared to the baseline.

Use of Emerging Internet of Things Technologies for Developing Infrastructure-to-Vehicle Communication Systems

Infrastructure-to-vehicle (I2V) communication is an essential part of an intelligent transportation system. Bluetooth low energy (BLE) is one of the emerging Internet of Things (IoT) technologies that can enable the exchange of data at low-energy consumption. This study evaluated the application of IoT technologies, in particular BLE beacons—I2V communication devices to support autonomous vehicle (AV) operation in a real-world setting. An AV development platform built on a Lexus RX450h car was used to perform the evaluation. The testing program included two controlled-environment experiments and two real-world validation experiments. The controlled-environment experiments were used to select the BLE beacon settings to be considered. The validation experiments were performed in two different settings: a city street and an interstate highway in the City of Cincinnati in Ohio. To this end, beacons were attached to 10 different traffic signs in a city street as well as to a side barrier of an interstate highway. The ability of the AV to detect the traffic signs and the beacons attached to the side barrier was evaluated at different speeds. The study revealed promising results and indicated that the BLE beacon technology could potentially be a reliable solution to develop I2V communication systems that can be easily deployed to modify existing transportation infrastructure.

The Project of an Autonomous Microboat with a Laser Device for Estimation of Water Area Pollution by Microplastic

Water area pollution by microplastic particles smaller than 5 mm is a serious environmental problem. Usually, studies of water pollution are carried out by taking water samples with their further analysis in the laboratory. Therefore, study of the water areas for the presence of plastic with the identification of extensive pollution sites in real time is actual.The project of an autonomous unmanned research vehicle operating in water area and transferring the research data to the information processing center is discussed in the article. The device produces laser probing of the aquatic environment to detect polyethylene microparticles and to build maps of pollution sites in the studied water area.To implement the proposed project, the following tasks are solved:detection of plastic microparticles in the studied volume of the water environment where particle sizes are several orders of magnitude smaller than the studied water volume;identification of microparticles in presence of particles of other types of substances in the investigated water environment volume;estimation of the number particles’ in the studied water environmentA new approach which consists in creating an autonomous microboat that explores the water area along a route set from a control center located on the shore is proposed. The study involves the detection of microparticles in a certain volume of liquid flowing through a research tube installed in the bottom of the boat. There is a Schauberger tube in the central part of which particles are concentrated inside the tube. It makes possible to search for particles by laser probing with a wavelength about 3.5 μm, on which the peak absorption of light by polyethylene many times exceeds the water absorption.To ensure the energy efficiency of the microboat it is proposed to use a scheme for laser pulses combining using a fiber-optic delay line. This makes it possible to reduce the power consumption of the laser by almost an order of magnitude. The proposed project of the boat will make it possible to conduct studies of large water areas with the detection of plastic microparticles’ pollution sites. Examples of the used optical fiber, photodetector and laser source for the problem being solved are derived in the article.

Sensor work: enabling the interoperation of autonomous vehicles

Abstract This article examines the “sensor work” carried out in the development of autonomous vehicles which, without sensor data, would not and arguably still do not, have the capacity to decide on where, and how, to drive. I begin by discussing three aspects of sensor technologies considered to be the foundation for sensor work being carried out in autonomous vehicle settings, namely the distribution, processing, and sourcing of sensor technologies and sensor data. The article considers how that much of this sensor work aids not only the operation of autonomous vehicles but also their necessary “interoperation.” In studying four specific sensing methods from an operational perspective, I consider how the interoperation between sensing devices and subsequent algorithmic, object-recognition, and motion planning procedures is fundamental to the development of autonomous vehicles.

Sparse Adversarial Unsupervised Domain Adaptation With Deep Dictionary Learning for Traffic Scene Classification

In recent years, the accurate recognition of traffic scenes has played a key role in autonomous vehicle operations. However, most works in this area do not address the domain shift issue where the classification performance is degraded when the distribution of the source and target images are different due to weather changes. Also, lack of sparsity of current studies results in sample complexity and overfitting issues. To mitigate these challenges, this paper proposes a novel Sparse Adversarial Domain Adaptation (SADA) model for traffic scene classification. Our objective is to learn a scene classifier on a source domain with sunny weather and transfer its knowledge to a target domain with different weather (i.e., cloudy, rainy, and snowy) to enhance the classification performance on the target. First, a sparse representation is learned from the source traffic scenes via nonlinear dictionary learning in the latent space of a deep classifier. Then, a conditional generative adversarial network is devised to capture the distribution of the source sparse codes. Finally, a domain invariant sparse feature extractor is developed via a minimax game to align the sparse codes of the target domain images with the source; hence, providing domain adaptation for traffic scene classification using a deep neural network. Experimental results on a real-world dataset collected by the Honda Research Institute (HRI) indicate the superior performance of the proposed SADA compared to current traffic scene classification methods and state-of-the-art domain adaptation frameworks.

Problems related to the operation of autonomous vehicles in adverse weather conditions

Article introduces and discusses the sensors used in autonomous cars. The reliability of these devices is crucial for the proper operation of autonomous driving systems. The research works related to the issue of the performance of autonomous sensors in adverse weather conditions is discussed and critically analysed. The negative effects caused by bad weather conditions are characterised. The paper presents the result of author's own research concern on the effects of rain, snow and fog on lidar measurments. The results obtained are presented, detailing the most important threats from each weather phenomenon. Attempts currently being made to address these issues are presented as well. The paper concludes with a summary of the research results, the current state of knowledge and suggestions for future developments.

Noncooperative and Cooperative Urban Intelligent Systems: Joint Logistic and Charging Incentive Mechanisms

Autonomous vehicles (AVs) have become an emerging crucial component of the intelligent transportation system (ITS) in modern smart cities. In particular, coordinated operations of AVs can potentially enhance the quality of public services, e.g. logistic and AV charging services. However, the joint logistic and AV charging scenario involves the sophisticated interactions between a large number of complicated agents, dynamic logistic, and electricity prices in real-world systems. Since AVs are individuals owned by different parties, the design of attractive incentive to motivate them to provide multiple public services becomes a fundamental issue. In this paper, we develop an urban intelligent system (UIS) by exploiting the efficient incentive mechanisms, e.g. non-cooperative and cooperative game theoretic approaches, to motivate the AVs to provide logistic and charging services in UIS. For the non-cooperative game approach, we formulate the interaction between the selfish AVs and the aggregator as a Stackelberg game. Meanwhile, the aggregator, known as the leader in the game, aims to decide the logistic and electricity trading prices, and then the AVs, executed as the followers, determine their service schedules. Furthermore, considering that all the players are willing to cooperate, we develop a cooperative potential game for the selfless AVs to maximize the social welfare of the UIS. These case studies demonstrate the effectiveness and practicability of proposed incentive mechanisms that can motivate EVs to provide high quality logistic and charging services by maximizing their utilities. Also, both the proposed schemes provide significant system revenues than that of conventional system optimization-based approaches.

Fast and accurate object detector for autonomous driving based on improved YOLOv5

Autonomous driving is an important branch of artificial intelligence, and real-time and accurate object detection is key to ensuring the safe and stable operation of autonomous vehicles. To this end, this paper proposes a fast and accurate object detector for autonomous driving based on improved YOLOv5. First, the YOLOv5 algorithm is improved by using structural re-parameterization (Rep), enhancing the accuracy and speed of the model through training-inference decoupling. Additionally, the neural architecture search method is introduced to cut redundant branches in the multi-branch re-parameterization module during the training phase, which ameliorates the training efficiency and accuracy. Finally, a small object detection layer is added to the network and the coordinate attention mechanism is added to all detection layers to improve the recognition rate of the model for small vehicles and pedestrians. The experimental results show that the detection accuracy of the proposed method on the KITTI dataset reaches 96.1%, and the FPS reaches 202, which is superior to many current mainstream algorithms and effectively improves the accuracy and real-time performance of unmanned driving object detection.

Evaluation of autonomous underwater vehicle motion trajectory optimization algorithms

The operation of autonomous underwater vehicles (AUVs) relies on three major motions: yaw, theta, and depth, each requiring its own set of proportional integral derivative (PID) controller gain criteria. Thus, different issues arise, including the availability of multiple criteria for optimization algorithm evaluation, the importance of these criteria, the trade-off between criterion performance, and criterion critical values. These issues make the evaluation of optimization algorithms for AUV motion control a complex multicriteria decision-making (MCDM) problem. This research proposes a novel selection-integrated approach for AUV optimization algorithms in different motions using two MCDM methods: fuzzy-weighted zero-inconsistency (FWZIC) for criteria weighting and fuzzy decision by opinion score method (FDOSM) for optimization algorithm selection. The approach comprises three main phases: development of PID, FWZIC-based criteria weighting, and FDOSM-based optimization algorithm selection. In all three motion types – “depth”, “yaw”, and “theta” – Kp_θ had the highest weight value, with respective weights of 0.143, 0.149, and 0.142. In contrast, Ki_depth consistently received the lowest weight value across all three motion types, with respective weights of 0.057, 0.0598, and 0.057. Regarding the “depth” motion, the Archimedes optimization algorithm (AOA) was the highest-performing alternative with a score of 0.077, while the eagle strategy–particle swarm optimization algorithm was the worst alternative with a score of 0.022. The Cuckoo optimization algorithm was identified as the best alternative for the “yaw” motion with a score of 0.072, whereas black hole optimization had the lowest score of 0.036. The best alternative for the “theta” motion was AOA with a score of 0.067, and the worst algorithm was sunflower optimization with a score of 0.028. This developed approach was evaluated based on systematic ranking and sensitivity analysis, which confirmed the validity of the proposed work.

Multi-ASV Collision Avoidance for Point-to-Point Transitions Based on Heading-Constrained Control Barrier Functions with Experiment

Dear Editor Collision avoidance is critical for safe operations of multiple autonomous surface vehicles (ASVs). It is a challenging task to design collision-free control laws to ensure safety, especially in a crowded sea environment. This letter presents a collision-free point-to-point transition strategy for multiple ASVs subject to static obstacles, dynamic obstacles and neighboring ASVs based on control barrier functions (CBFs). Existing CBFs consider only the position and not the heading, which makes it difficult to apply them to under-actuated ASVs. To address it, a heading-constrained CBF is proposed to design a yaw rate control law for avoiding collision with static obstacles, dynamic obstacles and vehicle neighbors. A quadratic program (QP) problem is formulated to compute the optimal yaw rate for each ASV such that the reach-avoid task can be achieved with safety guarantees. Both hardware-in-loop (HIL) simulation and field experiment results are given to substantiate the efficacy of the proposed heading-constrained CBFs.

Fleet sizing and charging infrastructure design for electric autonomous mobility-on-demand systems with endogenous congestion and limited link space

Autonomous vehicles are to revolutionize the way urban mobility demands are served, and they are most likely to be powered by electricity. To accurately quantify the benefits of replacing existing mobility services with autonomous electric vehicles, electric autonomous mobility-on-demand (EAMoD) systems need to be evaluated and designed with the congestion effects they may cause taken into account. In this work, the congestion effects are depicted through a discrete event simulation model with state-dependent link travel speed and limited link space that allow endogenous congestion to emerge, spread, and dissipate across the entire road network. Three mathematical models are integrated into the simulation model to optimally match vehicles with waiting requests, relocate empty vehicles to potential high-demand areas, and assign low charge vehicles to charging stations based on the workloads of the charging stations. Based on the simulation model, we apply Bayesian optimization to jointly design the fleet size and charging facility configuration considering the endogenous congestion incurred by the daily operation of autonomous electric vehicles. Contraction hierarchies are adopted to route vehicles to perform assigned tasks in real-time. The proposed solution method is tested on Manhattan below 60th street, which corresponds to the potential congestion pricing zone in New York City. Experiment results show that the proposed simulation-based optimization approach is computationally tractable, and can find a satisfactory solution to the fleet sizing and charging infrastructure design problem within a tight computation budget. Excluding congestion effects at system design stage would lead to up to 14% passenger loss due to long assignment waiting time. Although the charging facility cost accounts for only 1.8% of the total cost, the number of chargers can indirectly affect the percentage of passengers that can be served through its efficiency in recharging vehicles. A charging facility configuration aligned with the fleet size can help to improve service quality and vehicle utilization rate.

Trajectory Planning for an Autonomous Vehicle with Conflicting Moving Objects Along a Fixed Path – An Exact Solution Method

Trajectory planning for autonomous vehicles (AVs) by considering conflicting moving objects (CMOs) is a challenging problem to AV operations. This paper investigates an AV trajectory planning problem where the AV follows a given spatial path, and the trajectories of CMOs are predictable. With the spatial path fixed, the two-dimensional trajectory planning problem is reduced to a one-dimensional speed planning problem that decides the optimal speeds and accelerations of the AV along the spatial path. This paper first analyzes the conflict area caused by a single CMO in the space-time diagram, which reveals upper and lower bounds to the conflict area. Then a multi-area fusion algorithm is proposed to extend the upper and lower bound analyses to a relatively complex traffic scenario with multiple CMOs. To facilitate the computation of the investigated problem, a customized dynamic programming (DP) algorithm is developed, which employs the revealed upper and lower bounds and arriving time constraints to cut invalid trajectories at each stage. With this, the number of stages and states, as well as the computational time for solving the proposed problem, are largely reduced. A set of numerical experiments are conducted to evaluate the performance of the customized DP-based algorithm. The results show that the proposed customized DP-based algorithm can solve the investigated problem within milliseconds, which enables applications to real-time AV control. This much outperforms a state-of-the-art commercial solver, Gurobi, especially for complex traffic scenarios. We further implemented the proposed method in a real-world case study, and the results show that the trajectory generated by the proposed model has a great potential to enhance the future traffic system.

Policy and Strategy Evaluation of Ridesharing Autonomous Vehicle Operation: A London Case Study

To understand the dynamics of an autonomous ridesharing transport mode from the perspectives of different stakeholders, a single model of such a system is essential, because this will enable policymakers and companies involved in the manufacture and operation of shared autonomous vehicles (SAVs) to develop user-centered strategies. The model needs to be based on real data, network, and traffic information and applied to real cities and situations, particularly those with complex public transportation systems. In this paper, we propose a new agent-based model for SAV deployment that enables the parametric assessment of key performance indicators from the perspective of potential SAV users, vehicle manufacturers, operators, and local authorities. This has been applied to a case study of three regions in London: central, inner, and outer. The results show there is no linear correlation between an increased ridesharing acceptance level and average trip duration. Without a fleet rebalancing algorithm, over 80% of SAVs’ energy expenditure is on picking up customers. By reducing pickup distance, SAVs could be a contender for a nonpersonal transportation system based on trip energy comparisons. The results provide a picture of future SAV systems for potential users and offer suggestions as to how operators can devise an optimal transportation strategy beyond the question of fleet size and how policymakers can improve the overall transport network and reduce its environmental impact based on energy consumption. As a result of its flexibility and parametric capability, the model can be utilized to inform any local authority how SAV services could be deployed in any city.

Research on Risk Detection of Autonomous Vehicle Based on Rapidly-Exploring Random Tree

There is no doubt that the autonomous vehicle is an important developing direction of the auto industry, and, thus, more and more scholars are paying attention to doing more research in this field. Since path planning plays a key role in the operation of autonomous vehicles, scholars attach great importance to this field. Although it has been applied in many fields, there are still some problems, such as low efficiency of path planning and collision risk during driving. In order to solve these problems, an automotive vehicle-based rapid exploration random tree (AV-RRT)-based non-particle path planning method for autonomous vehicles is proposed. On the premise of ensuring safety and meeting the requirements of the vehicle’s kinematic constraints through the expansion of obstacles, the dynamic step size is used for random tree growth. A non-particle collision detection (NPCD) collision detection algorithm and path modification (PM) path modification strategy are proposed for the collision risk in the turning process, and geometric constraints are used to represent the possible security threats, so as to improve the efficiency and safety of vehicle global path driving and to provide reference for the research of driverless vehicles.

Factors influencing safety perceptions of sharing roadways with autonomous vehicles among vulnerable roadway users

Introduction: The operation of autonomous vehicles (AVs) on public roadways affects the safety of vulnerable roadway users, such as pedestrians and bicyclists. This research contributes to the literature by investigating vulnerable roadway users’ safety perceptions on road sharing with AVs. Method: This study analyzed the survey responses of pedestrians and bicyclists in Pittsburgh, Pennsylvania, collected by Bike Pittsburgh (Bike PGH) in 2017 and 2019. First, this study investigates how pedestrians and bicyclists perceive safety regarding road sharing with AVs. Second, the study examines how the safety perceptions of pedestrians and bicyclists regarding AVs might be changing over time. Non-parametric tests were applied to compare the safety perceptions of pedestrians and bicyclists across different characteristics, experiences, and attitudes, considering the ordinal nature of the AV safety perception data. An ordered probit model was estimated to better understand the factors influencing safety perceptions regarding road sharing with AVs. Results: The study findings suggest that higher exposures to AVs are associated with improved safety perceptions. In addition, respondents with a stricter attitude toward AV regulations perceive road sharing with AVs as less safe. Respondents whose opinion regarding AVs did not worsen due to the pedestrian/bicyclist involved AV accident in Arizona have higher safety perceptions. Practical Applications: Policymakers can use the findings of this study in developing guidelines to ensure safe road sharing and develop strategies to sustain active transportation usage in the future AV era.

Framework for environment perception: Ensemble method for vision-based scene understanding algorithms in agriculture.

The safe and reliable operation of autonomous agricultural vehicles requires an advanced environment perception system. An important component of perception systems is vision-based algorithms for detecting objects and other structures in the fields. This paper presents an ensemble method for combining outputs of three scene understanding tasks: semantic segmentation, object detection and anomaly detection in the agricultural context. The proposed framework uses an object detector to detect seven agriculture-specific classes. The anomaly detector detects all other objects that do not belong to these classes. In addition, the segmentation map of the field is utilized to provide additional information if the objects are located inside or outside the field area. The detections of different algorithms are combined at inference time, and the proposed ensemble method is independent of underlying algorithms. The results show that combining object detection with anomaly detection can increase the number of detected objects in agricultural scene images.

A Constraint Optimization path planning for Dynamic Environment

Optimal path planning is a major challenge for the operation of autonomous unmanned ground vehicles. In dynamic mobile path planning, the goal is to find a collision-free path for a robot from an initial position to a goal position in an environment with obstacles. However, in a dynamic environment, the challenge is bigger than static. In this paper, a modified ABC and an adaptive constraint PSO algorithm are proposed to deal with the dynamic environment; the algorithms were chosen to have a good balance between exploration and exploitation, with a penalty function used to have a collision-free path.