arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2305.04412cs.ROcs.AIcs.LG
收藏

带参数化技能与先验的高效自动驾驶强化学习

Efficient Reinforcement Learning for Autonomous Driving with Parameterized Skills and Priors

University of Toronto(多伦多大学);Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);SenseTime Research(商汤科技研究院);McGill University(麦吉尔大学)

Letian Wang, Jie Liu, Hao Shao, Wenshuo Wang, Ruobing Chen, Yu Liu, Steven L. Waslander

AI总结:

针对自动驾驶在交互式密集交通中学习效率低的问题,提出ASAP-RL算法,通过参数化运动技能和双重初始化利用专家先验,在稀疏奖励下实现更高学习效率和更好驾驶性能。

摘要:

当自动驾驶车辆部署在公共道路上时,它们将遇到无数且多样的驾驶情境。许多手动设计的驾驶策略难以扩展到现实世界。幸运的是,强化学习通过自动试错在许多任务中取得了巨大成功。然而,当涉及交互式密集交通中的自动驾驶时,强化学习智能体要么无法学习到合理的性能,要么需要大量数据。我们的见解是,人类学习驾驶时,会1)在高层技能空间而非低层控制空间进行决策,2)利用专家先验知识而非从零开始学习。受此启发,我们提出了ASAP-RL,一种高效的自动驾驶强化学习算法,该算法同时利用运动技能和专家先验。我们首先参数化运动技能,这些技能足够多样化,能够覆盖各种复杂驾驶场景和情境。提出了一种技能参数逆恢复方法,将专家演示从控制空间转换到技能空间。提出了一种简单但有效的双重初始化技术,以利用专家先验,同时绕过专家次优性和早期性能退化的问题。我们在给定简单且稀疏奖励的交互式密集交通驾驶任务上验证了我们提出的方法。实验结果表明,相对于先前以不同方式利用技能和先验的方法,我们的方法能够带来更高的学习效率和更好的驾驶性能。代码已开源以促进进一步研究。

英文摘要:

When autonomous vehicles are deployed on public roads, they will encounter countless and diverse driving situations. Many manually designed driving policies are difficult to scale to the real world. Fortunately, reinforcement learning has shown great success in many tasks by automatic trial and error. However, when it comes to autonomous driving in interactive dense traffic, RL agents either fail to learn reasonable performance or necessitate a large amount of data. Our insight is that when humans learn to drive, they will 1) make decisions over the high-level skill space instead of the low-level control space and 2) leverage expert prior knowledge rather than learning from scratch. Inspired by this, we propose ASAP-RL, an efficient reinforcement learning algorithm for autonomous driving that simultaneously leverages motion skills and expert priors. We first parameterized motion skills, which are diverse enough to cover various complex driving scenarios and situations. A skill parameter inverse recovery method is proposed to convert expert demonstrations from control space to skill space. A simple but effective double initialization technique is proposed to leverage expert priors while bypassing the issue of expert suboptimality and early performance degradation. We validate our proposed method on interactive dense-traffic driving tasks given simple and sparse rewards. Experimental results show that our method can lead to higher learning efficiency and better driving performance relative to previous methods that exploit skills and priors differently. Code is open-sourced to facilitate further research.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑