Survey on Path Planning Based on Deep Reinforcement Learning

Lihan Xu, Wenzhi Zhang
Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, PMLR 278:685-695, 2025.

Abstract

In recent years, deep reinforcement learning (DRL) has demonstrated significant potential in the field of path planning and control, offering breakthrough solutions for path planning in dynamic and complex environments. DRL has been widely applied in UAV obstacle avoidance, autonomous vehicle path optimization, multi-robot coordination, and complex terrain navigation, demonstrating ad-vantages such as superior path quality, improved smoothness, and enhanced safety. This paper provides a systematic review of recent advances and applications of DRL core techniques. Value-based methods (e.g. DQN) significantly improve decision-making efficiency through optimized reward design and network architectures. Policy gradient algorithms (such as PPO, DDPG, and TD3) achieve high-precision control in continuous action spaces. The Actor-Critic framework, combined with double Q-networks and delayed update mechanisms (e.g. TD3), further expands the application scenarios. Future research should focus on enhancing cross-scenario generalization capabilities and improving deployment efficiency at the industrial level, thereby promoting the practical application of DRL in autonomous driving and industrial robotics.

Cite this Paper

BibTeX
@InProceedings{pmlr-v278-xu25a, title = {Survey on Path Planning Based on Deep Reinforcement Learning}, author = {Xu, Lihan and Zhang, Wenzhi}, booktitle = {Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing}, pages = {685--695}, year = {2025}, editor = {Zeng, Nianyin and Pachori, Ram Bilas and Wang, Dongshu}, volume = {278}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v278/main/assets/xu25a/xu25a.pdf}, url = {https://proceedings.mlr.press/v278/xu25a.html}, abstract = { In recent years, deep reinforcement learning (DRL) has demonstrated significant potential in the field of path planning and control, offering breakthrough solutions for path planning in dynamic and complex environments. DRL has been widely applied in UAV obstacle avoidance, autonomous vehicle path optimization, multi-robot coordination, and complex terrain navigation, demonstrating ad-vantages such as superior path quality, improved smoothness, and enhanced safety. This paper provides a systematic review of recent advances and applications of DRL core techniques. Value-based methods (e.g. DQN) significantly improve decision-making efficiency through optimized reward design and network architectures. Policy gradient algorithms (such as PPO, DDPG, and TD3) achieve high-precision control in continuous action spaces. The Actor-Critic framework, combined with double Q-networks and delayed update mechanisms (e.g. TD3), further expands the application scenarios. Future research should focus on enhancing cross-scenario generalization capabilities and improving deployment efficiency at the industrial level, thereby promoting the practical application of DRL in autonomous driving and industrial robotics.} }
Endnote
%0 Conference Paper %T Survey on Path Planning Based on Deep Reinforcement Learning %A Lihan Xu %A Wenzhi Zhang %B Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing %C Proceedings of Machine Learning Research %D 2025 %E Nianyin Zeng %E Ram Bilas Pachori %E Dongshu Wang %F pmlr-v278-xu25a %I PMLR %P 685--695 %U https://proceedings.mlr.press/v278/xu25a.html %V 278 %X In recent years, deep reinforcement learning (DRL) has demonstrated significant potential in the field of path planning and control, offering breakthrough solutions for path planning in dynamic and complex environments. DRL has been widely applied in UAV obstacle avoidance, autonomous vehicle path optimization, multi-robot coordination, and complex terrain navigation, demonstrating ad-vantages such as superior path quality, improved smoothness, and enhanced safety. This paper provides a systematic review of recent advances and applications of DRL core techniques. Value-based methods (e.g. DQN) significantly improve decision-making efficiency through optimized reward design and network architectures. Policy gradient algorithms (such as PPO, DDPG, and TD3) achieve high-precision control in continuous action spaces. The Actor-Critic framework, combined with double Q-networks and delayed update mechanisms (e.g. TD3), further expands the application scenarios. Future research should focus on enhancing cross-scenario generalization capabilities and improving deployment efficiency at the industrial level, thereby promoting the practical application of DRL in autonomous driving and industrial robotics.
APA
Xu, L. & Zhang, W.. (2025). Survey on Path Planning Based on Deep Reinforcement Learning. Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, in Proceedings of Machine Learning Research 278:685-695 Available from https://proceedings.mlr.press/v278/xu25a.html.

Related Material