Reaching the limit in autonomous racing: Optimal control versus reinforcement learning

强化学习 机器人学 敏捷软件开发 计算机科学 人工智能 控制器(灌溉) 杠杆(统计) 机器人 弹道 控制工程 控制理论(社会学) 控制(管理) 工程类 农学 软件工程 生物 物理 天文
作者
Yunlong Song,Ángel Romero,Matthias M. Müller,Vladlen Koltun,Davide Scaramuzza
出处
期刊:Science robotics [American Association for the Advancement of Science]
卷期号:8 (82) 被引量:51
标识
DOI:10.1126/scirobotics.adg1462
摘要

A central question in robotics is how to design a control system for an agile mobile robot. This paper studies this question systematically, focusing on a challenging setting: autonomous drone racing. We show that a neural network controller trained with reinforcement learning (RL) outperformed optimal control (OC) methods in this setting. We then investigated which fundamental factors have contributed to the success of RL or have limited OC. Our study indicates that the fundamental advantage of RL over OC is not that it optimizes its objective better but that it optimizes a better objective. OC decomposes the problem into planning and control with an explicit intermediate representation, such as a trajectory, that serves as an interface. This decomposition limits the range of behaviors that can be expressed by the controller, leading to inferior control performance when facing unmodeled effects. In contrast, RL can directly optimize a task-level objective and can leverage domain randomization to cope with model uncertainty, allowing the discovery of more robust control responses. Our findings allowed us to push an agile drone to its maximum performance, achieving a peak acceleration greater than 12 times the gravitational acceleration and a peak velocity of 108 kilometers per hour. Our policy achieved superhuman control within minutes of training on a standard workstation. This work presents a milestone in agile robotics and sheds light on the role of RL and OC in robot control.
最长约 10秒,即可获得该文献文件

科研通智能强力驱动
Strongly Powered by AbleSci AI
更新
PDF的下载单位、IP信息已删除 (2025-6-4)

科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
2秒前
秋暝寒衣完成签到,获得积分10
4秒前
完美世界应助bfs采纳,获得10
5秒前
橘子sungua完成签到,获得积分10
6秒前
111完成签到,获得积分10
6秒前
8秒前
10秒前
11秒前
可乐加冰完成签到,获得积分10
11秒前
David驳回了Ant应助
11秒前
12秒前
13秒前
量子星尘发布了新的文献求助10
14秒前
思源应助果粒多采纳,获得10
14秒前
15秒前
好滴捏发布了新的文献求助10
16秒前
bfs发布了新的文献求助10
17秒前
WN发布了新的文献求助10
18秒前
18秒前
慕青应助小白采纳,获得10
19秒前
AAACharlie发布了新的文献求助10
19秒前
热情的达发布了新的文献求助10
19秒前
orixero应助lucky李采纳,获得10
20秒前
20秒前
momo发布了新的文献求助10
21秒前
21秒前
22秒前
guo完成签到,获得积分10
23秒前
可期完成签到,获得积分10
24秒前
25秒前
wsj发布了新的文献求助10
25秒前
果粒多发布了新的文献求助10
26秒前
科目三应助ylq采纳,获得30
27秒前
liupc2019发布了新的文献求助20
28秒前
张雯思发布了新的文献求助10
31秒前
希望天下0贩的0应助momo采纳,获得10
31秒前
32秒前
33秒前
梦华完成签到 ,获得积分10
34秒前
35秒前
高分求助中
A new approach to the extrapolation of accelerated life test data 1000
ACSM’s Guidelines for Exercise Testing and Prescription, 12th edition 500
‘Unruly’ Children: Historical Fieldnotes and Learning Morality in a Taiwan Village (New Departures in Anthropology) 400
Indomethacinのヒトにおける経皮吸収 400
Phylogenetic study of the order Polydesmida (Myriapoda: Diplopoda) 370
基于可调谐半导体激光吸收光谱技术泄漏气体检测系统的研究 350
Robot-supported joining of reinforcement textiles with one-sided sewing heads 320
热门求助领域 (近24小时)
化学 材料科学 医学 生物 工程类 有机化学 生物化学 物理 内科学 纳米技术 计算机科学 化学工程 复合材料 遗传学 基因 物理化学 催化作用 冶金 细胞生物学 免疫学
热门帖子
关注 科研通微信公众号,转发送积分 3989297
求助须知:如何正确求助?哪些是违规求助? 3531418
关于积分的说明 11253893
捐赠科研通 3270097
什么是DOI,文献DOI怎么找? 1804884
邀请新用户注册赠送积分活动 882087
科研通“疑难数据库(出版商)”最低求助积分说明 809158