Robust Decision Making for Autonomous Vehicles at Highway On-Ramps: A Constrained Adversarial Reinforcement Learning Approach

He, Xiangkun; Lou, Baichuan; Yang, Haohan; Lv, Chen

doi:10.1109/tits.2022.3229518

Cited by 48 publications

(11 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The lateral deviation of the preview point and the deviation of the yaw or heading angle are selected as the state quantities to guarantee the vehicle's path tracking, and the vehicle kinematics equations based on the yaw and heading angles are expressed through the dynamics parameters of the vehicle, respectively [26][27][28][29].…”

Section: Vehicle Kinematics Modeling Based On Yaw and Heading Anglesmentioning

confidence: 99%

Path Tracking Control Based on T-S Fuzzy Model for Autonomous Vehicles with Yaw Angle and Heading Angle

He,

Wu,

et al. 2024

Machines

View full text Add to dashboard Cite

Existing vehicle-road models used for road tracking do not take into account the side slip angle, which leads to a reduction in road tracking accuracy in scenarios where the vehicle is at a large side slip angle, such as an emergency lane change. Consequently, this study presents a path-tracking control technique based on the T-S fuzzy model of heading angle vehicle autonomy. In this paper, based on the yaw angle-based vehicle tracking model, a heading angle-based tracking model considering the side slip angle is constructed. Second, since the vehicle speed varies with time, this paper selects the membership function of the vehicle speed to establish the T-S fuzzy model of autonomous vehicle based on the yaw angle and heading angle, respectively, and ensures the robustness and stability over the whole parameter space by the linear parameter variation robust H∞ controller. Then, cost functions based on the yaw angle and heading angle augmented error systems are created separately to optimize the system’s overall performance. Ultimately, simulation and experimentation confirm that the algorithm for control, which is based on the fuzzy model of the heading angle vehicle, has superior autonomous trajectory performance.

show abstract

Section: Vehicle Kinematics Modeling Based On Yaw and Heading Anglesmentioning

confidence: 99%

Path Tracking Control Based on T-S Fuzzy Model for Autonomous Vehicles with Yaw Angle and Heading Angle

He,

Wu,

et al. 2024

Machines

View full text Add to dashboard Cite

show abstract

“…Consideration for moving obstacles is an inevitable aspect of future work. With the evolution and increasing accessibility of autonomous navigation technologies, the rapid detection and avoidance of mobile objects become an imperative concern [35]. Future re search focus will be on the development of new sensors, construction of more accurate obstacle-prediction models, and implementation of faster path-planning algorithms.…”

Section: Conclusion and Further Workmentioning

confidence: 99%

High-Efficiency Navigation of Nonholonomic Mobile Robots Based on Improved Hybrid A* Algorithm

Chi

et al. 2023

Applied Sciences

View full text Add to dashboard Cite

With the development of automation technologies, autonomous robots are increasingly used in many important applications. However, precise self-navigation and accurate path planning remain a significant challenge, particularly for the robots operating in complex circumstances such as city centers. In this paper, a nonholonomically constrained robot with high-precision navigation and path planning capability was designed based on the Robot Operating System (ROS), and an improved hybrid A* algorithm was developed to increase the processing efficiency and accuracy of the global path planning and navigation of the robot. The performance and effectiveness of the algorithm were evaluated by using randomly constructed maps in MATLAB and validated in a practical circumstance. Local path planning and obstacle avoidance were carried out based on the model predictive control (MPC) theory. Compared with the conventional A* + DWA (dynamic window approach) method, the average searching time was reduced by 12.62~24.5%, and the average search length was reduced by 9.25~9.5%. In practical navigating tests, the average search time was reduced by 18~24%, and the average search length was reduced by 10.3~12%, while the overall path was smoother. The results demonstrate that the improved algorithm can enable precise and efficient navigation and path planning of the robot in complex circumstances.

show abstract

“…Algorithm 2 outlines the shared reward function design in detail where a y denotes vehicle lateral acceleration, k represents dynamic factor [42], g represents gravity acceleration, and µ denotes adhesion coefficient. The comfort in reward function is set based on the results of [43]. In terms of safety, not only collision but also vehicle dynamic stability is considered.…”

Section: Reward Functionmentioning

confidence: 99%

Robust Multiagent Reinforcement Learning toward Coordinated Decision-Making of Automated Vehicles

He,

Chen,

2023

SAE Int. J. Veh. Dyn., Stab., and NVH

Self Cite

View full text Add to dashboard Cite

<div>Automated driving is essential for developing and deploying intelligent transportation systems. However, unavoidable sensor noises or perception errors may cause an automated vehicle to adopt suboptimal driving policies or even lead to catastrophic failures. Additionally, the automated driving longitudinal and lateral decision-making behaviors (e.g., driving speed and lane changing decisions) are coupled, that is, when one of them is perturbed by unknown external disturbances, it causes changes or even performance degradation in the other. The presence of both challenges significantly curtails the potential of automated driving. Here, to coordinate the longitudinal and lateral driving decisions of an automated vehicle while ensuring policy robustness against observational uncertainties, we propose a novel robust coordinated decision-making technique via robust multiagent reinforcement learning. Specifically, the automated driving longitudinal and lateral decisions under observational perturbations are modeled as a constrained robust multiagent Markov decision process. Meanwhile, a nonlinear constraint setting with Kullback–Leibler divergence is developed to keep the variation of the driving policy perturbed by stochastic perturbations within bounds. Additionally, a robust multiagent policy optimization approach is proposed to approximate the optimal robust coordinated driving policy. Finally, we evaluate the proposed robust coordinated decision-making method in three highway scenarios with different traffic densities. Quantitatively, in the absence of noises, the proposed method achieves an approximate average enhancement of 25.58% in traffic efficiency and 91.31% in safety compared to all baselines across the three scenarios. In the presence of noises, our technique improves traffic efficiency and safety by an approximate average of 30.81% and 81.02% compared to all baselines in the three scenarios, respectively. The results demonstrate that the proposed approach is capable of improving automated driving performance and ensuring policy robustness against observational uncertainties.</div>

show abstract

Robust Decision Making for Autonomous Vehicles at Highway On-Ramps: A Constrained Adversarial Reinforcement Learning Approach

Cited by 48 publications

References 42 publications

Path Tracking Control Based on T-S Fuzzy Model for Autonomous Vehicles with Yaw Angle and Heading Angle

Path Tracking Control Based on T-S Fuzzy Model for Autonomous Vehicles with Yaw Angle and Heading Angle

High-Efficiency Navigation of Nonholonomic Mobile Robots Based on Improved Hybrid A* Algorithm

Robust Multiagent Reinforcement Learning toward Coordinated Decision-Making of Automated Vehicles

Contact Info

Product

Resources

About