Deep Reinforcement Learning for Autonomous Driving: A Survey
By B. R. Kiran, Ibrahim Sobh, Victor Talpaert, Patrick Mannion, Ahmad A. Al Sallab, Senthil Kumar Yogamani, Patrick Pérez. 2020
2,407 citations recorded by the source corpus.
With the development of deep representation learning, the domain of reinforcement learning (RL) has become a powerful learning framework now capable of learning complex policies in high dimensional environments. This review summarises deep reinforcement learning (DRL) algorithms and provides a taxonomy of automated driving tasks where (D)RL methods have been employed, while addressing key computational challenges in real world deployment of autonomous driving agents. It also delineates adjacent domains such as behavior cloning, imitation learning, inverse reinforcement learning that are related but are not classical RL algorithms. The role of simulators in training agents, methods to validate, test and robustify existing solutions in RL are discussed.
Subfields: Artificial Intelligence, Automotive Engineering, Control and Systems Engineering, Reinforcement Learning, Robotics & Control
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
In Automotive Engineering
In Control and Systems Engineering
- Peter Stone
- W. B. Knox
- Lesia Semenova
- Roberto Mart'in-Mart'in
- Dacheng Tao
- X. Guan
- Tao Jiang
- Felix Schmitt
In Reinforcement Learning
- Peter Stone
- W. B. Knox
- Lilian Weng
- Roberto Mart'in-Mart'in
- Dacheng Tao
- X. Guan
- Tao Jiang
- Holger Banzhaf
In Robotics & Control
- Roberto Mart'in-Mart'in
- Dacheng Tao
- W. B. Knox
- Peter Stone
- Lilian Weng
- Jixiang Guo
- Mingming Gong
- Yasin Yilmaz
Citations in the corpus
- Interpretable Machine Learning: Fundamental Principles and 10 Grand Challenges (2021, background): “Recently, deep reinforcement learning algorithms have achieved state-of-the-art performance in mastering the game of Go and various Atari games [273, 210] and have been actively used in robotic control [156, 283], simulated autonomous navigation and self-driving cars [150, 265], dialogue summarizati”
- Milestones in Autonomous Driving and Intelligent Vehicles: Survey of Surveys (2023, neutral): “DATASETS The publicity of various kinds of autonomous driving datasets has made a substantial contribution to the advancement in this area, especially for perception and E2E planning tasks.”
- A Relational Intervention Approach for Unsupervised Dynamics Generalization in Model-Based Reinforcement Learning (2022, background): “However, solving real-world problems with RL is still a challenging problem because the sample efficiency of RL is low while the data in many applications is limited or expensive to obtain (Gottesman et al., 2018; Lu et al., 2018; 2020; Kiran et al., 2020).”
- Reward (Mis)design for Autonomous Driving (2021, background): “Additionally, this paper is not a general survey of reinforcement learning for autonomous driving; such surveys have been conducted by Zhu and Zhao [57] and Kiran et al. [25].”
- BEHAVIOR: Benchmark for Everyday Household Activities in Virtual, Interactive, and Ecological Environments (2021, neutral): “…kinds of embodied AI methods, here we evaluate two reinforcement learning (RL) algorithms that have demonstrated excellent results in simpler embodied AI tasks with continuous or discrete action spaces [61, 62, 21, 63–67]: Soft-Actor Critic (SAC [16]) and Proximal-Policy Optimization (PPO [17]).”
- Exploration in Deep Reinforcement Learning: A Survey (2022, background): “impressive performance in Atari games [2], Mujoco simulator [3], controllers tuning [4], autonomous landing [5], self-driving cars [6] and healthcare [7].”
- Deep Reinforcement Learning for Intelligent Transportation Systems: A Survey (2020, background): “applications for autonomous vehicles is presented in [10], where authors discuss recent works with the challenges of real-world deployment of such RL-based autonomous driving methods.”
- End-to-End Autonomous Driving: Challenges and Frontiers (2023, neutral): “The others focus on specific topics in this domain, such as imitation learning [44], [45], [46] or reinforcement learning [47], [48].”
- A Survey on Offline Reinforcement Learning: Taxonomy, Review, and Open Problems (2022, background): “driving [12], [81], healthcare [11], [82], dialog systems [83],”
- A Survey on Trajectory-Prediction Methods for Autonomous Driving (2022, neutral): “trajectory prediction tasks of AVs [163], [164].”
- Sensor and Sensor Fusion Technology in Autonomous Vehicles: A Review (2021, background): “One approach to providing a more reliable and accurate obstacle detection is to enhance existing sensor fusion algorithms through deep learning approaches or deep reinforcement learning approaches [215].”
- Deep Reinforcement Learning for Robotics: A Survey of Real-World Successes (2024, background): “In some domains, such as urban autonomous driving, DRL-based solutions remain limited to simulation or strictly confined field tests (8).”
- A Review of Deep Reinforcement Learning for Smart Building Energy Management (2020, background): “communications and networking, autonomous driving, autonomous IoT, cyber security, and multi-agent systems can be found in [24], [33]–[36], [52]–[56].”
- A Survey on Multimodal Large Language Models for Autonomous Driving (2023, background): “Deep Reinforcement Learning (DRL) has additionally played a pivotal role in enhancing the control strategies of autonomous vehicles, refining motion planning, and decision-making processes to adapt to dynamic and uncertain driving conditions [16,75,78,93],.”