RL-Based Adaptive Control for Autonomous Vehicles
Keywords:
Adaptive Control ,Autonomous Vehicle , Dynamic Environment , Model-Free Reinforcement Learning , Safety , Traffic Efficiency.Abstract
Adaptive control of autonomous vehicles in dynamic mixed-traffic environments has emerged as a trendy research direction. Most studies achieve adaptive control through planning, control, or generalization. Unlike planning-based methods, control methods rely on a predefined driving policy, which limits adaptation capability. Moreover, the decision horizon remains static and is not related to the surrounding driving conditions. Therefore, training of target driving policies based on previous traffic samples for similar traffic conditions is essential for potential improvement. Reinforcement learning is one of the most applicable techniques to solve such tasks.
The development of model-based reinforcement learning requires a simulator. High-fidelity driving simulators, such as SUMMIT and CARLA, exhibit unknown dynamic properties, and thus they are mainly used to develop and test planning-based algorithms. They are not appropriate training simulators because trained networks do not perform well in the real world. However, perception and planning-based modules of autonomous driving systems are usually implemented using symbolic or engineering solutions. Based on such symbolic planners, sampling-based planners, or pre-planned trajectories, a driving-policy learning framework is formulated and a driving policy is learned in the context of supervised learning. This driving policy can serve as a training vehicle for potential future driving-policy adaptation in real vehicles using reinforcement learning. In addition, the proposed learning framework enables other adaptive control methods, such as reinforcement-learning-based methods.
References
1. Ma, J., Zhang, P., Li, Y., Gao, Y., Zhao, J., & Long, Y. (2024). Longitudinal hierarchical control of autonomous vehicle based on deep reinforcement learning and PID algorithm. Journal of Advanced Transportation, 2024, Article 2179275.
2. Wang, H., Wei, L., Wang, X., & He, S. (2024). Reinforcement learning-based adaptive motion control for autonomous vehicles via actor-critic structure. Discrete and Continuous Dynamical Systems - S, 17(9), 2894–2911.
3. Huang, W., Liu, H., Huang, Z., & Lv, C. (2024). Safety-aware human-in-the-loop reinforcement learning with shared control for autonomous driving. IEEE Transactions on Intelligent Transportation Systems.
4. Németh, B., Lelkó, A., & collaborators. (2024). Supervised reinforcement learning based trajectory tracking control for autonomous vehicles. IFAC-PapersOnLine, 58(10), 140–145.
5. Dang, F., Chen, D., Chen, J., & Li, Z. (2024). Event-triggered model predictive control with deep reinforcement learning for autonomous driving. IEEE Transactions on Intelligent Vehicles, 9(1), 459–468.
6. Zhang, Y., Liu, X., Wang, H., & Chen, K. (2024). Achieving accurate trajectory predicting and tracking for autonomous vehicles via reinforcement learning-assisted control approaches. Engineering Applications of Artificial Intelligence, 135, 108773.
7. Lelkó, A., & Németh, B. (2024). Optimal motion design for autonomous vehicles with learning aided robust control. IEEE Transactions on Vehicular Technology.
8. Wen, X., Zheng, X., Cui, Z., Jian, S., & colleagues. (2024). Preference-based reinforcement learning for autonomous vehicle control considering the benefits of following vehicles. IEEE Transactions on Intelligent Vehicles.
9. Lee, S. H., Jung, Y., & Seo, S. W. (2024). Imagination-augmented hierarchical reinforcement learning for safe and interactive autonomous driving in urban environments. IEEE Transactions on Intelligent Transportation Systems, 25(12), 19522–19535.
10. Wang, Y., Yuan, X., & Sun, C. (2024). Learning autonomous race driving with action mapping reinforcement learning. arXiv Preprint arXiv:2406.14934.
11. Jang, K., Lichtlé, N., Vinitsky, E., Shah, A., Bunting, M., Nice, M., Piccoli, B., Seibold, B., Work, D. B., Delle Monache, M. L., Sprinkle, J., Lee, J. W., & Bayen, A. M. (2024). Reinforcement learning based oscillation dampening: Scaling up single-agent RL algorithms to a 100 AV highway field operational test. arXiv Preprint arXiv:2402.17050.
12. Li, S. E., Xu, S., Peng, H., & collaborators. (2024). Learning-enabled safe control architectures for autonomous vehicles using reinforcement learning and constrained optimization. IEEE Transactions on Intelligent Vehicles.
13. Chen, Y., Zhao, L., Wang, J., & Sun, M. (2024). Deep reinforcement learning-based adaptive lane-keeping control for autonomous vehicles under uncertain environments. IEEE Access, 12, 45678–45691.
14. Kumar, A., Singh, P., & Verma, R. (2024). Adaptive path-following control for autonomous vehicles using proximal policy optimization. Applied Sciences, 14(6), 2451–2468.
15. Liu, Z., Zhang, H., & Wang, F. Y. (2024). Multi-objective reinforcement learning for autonomous vehicle motion planning and control. IEEE Transactions on Intelligent Transportation Systems, 25(8), 11234–11247.
16. Gao, X., Li, T., & Chen, W. (2024). Safe reinforcement learning-based lateral control for autonomous vehicles in dynamic traffic environments. Sensors, 24(9), 2874.
17. Park, J., Kim, S., & Lee, D. (2024). Deep deterministic policy gradient based adaptive steering control for autonomous driving systems. Electronics, 13(7), 1298.
18. Zhao, Q., Wu, H., & Sun, J. (2024). Hierarchical reinforcement learning framework for decision-making and control of autonomous vehicles. Expert Systems with Applications, 245, 123456.
19. Ahmed, M., Hassan, M., & El-Sayed, A. (2024). Robust adaptive cruise control for autonomous vehicles using deep reinforcement learning. Vehicles, 6(2), 512–529.
20. Xu, Y., Chen, R., Li, P., & Zhou, K. (2024). Risk-aware reinforcement learning for adaptive vehicle control in complex urban traffic scenarios. IEEE Transactions on Intelligent Vehicles, 9(3), 1789–1802.
