Deep Reinforcement Learning in Decision-Making of Autonomous Driving: A Survey
Jizhou Cai
DOI: https://doi.org/10.59429/esta.v10i5.1414
Keywords: Autonomous Driving; Deep Learning; Reinforcement Learning
Abstract
Deep reinforcement learning (DRL) is a burgeoning sub-field in the realm of artificial intelligence that combines the benefits of deep learning (DL) and reinforcement learning (RL). By integrating these two methods, deep reinforcement learning has effectively addressed previously complex problems related to autonomous driving system (ADs) and has played a vital role in their development. Specifically, deep learning enhances reinforcement learning’s ability to handle extensive high-dimensional data, which is critical for ADs. In this review, we mainly concentrate on the application of DRL algorithms in ADs, focusing primarily on decision-making processes. The review will be- gin by introducing deep learning and reinforcement learning independently before delving into the current applications, future prospects, and challenges facing deep reinforcement learning in this field. Finally, we will conclude with a summary of this review.
References
Carlier, & Mathilde. (2023). Topic: Automotive industry worldwide. Retrieved from https://www.statista.com/topics/1487/automotive-industry/#topicOverview.
Caroselli. (2022, Jun). What percentage of car accidents are caused by human error?: Pittsburgh law blog. Caroselli, Beachler amp; Coleman, LLC. Retrieved from https://www.cbmclaw.com/what-percentage-of-car-accidents-are-caused-by-human-error/.
Chen, X., & Chen, S. (2015). Autonomous vehicles: A review. International Journal of Automotive Engineering, 6 (1).
Cheng, Y., & Song, Y. (2020). Autonomous decision-making generation of uav based on soft actor-critic algorithm. In 2020 39th Chinese control conference (ccc) (p. 7350-7355).