{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/towards-optimally-decentralized-multi-robot","title":"Towards Optimally Decentralized Multi-Robot Collision Avoidance via Deep Reinforcement Learning","arxiv_id":"1709.10082","date":"2017-09-28","proceeding":null,"authors":["Pinxin Long","Tingxiang Fan","Xinyi Liao","Wenxi Liu","Hao Zhang","Jia Pan"],"abstract":"Developing a safe and efficient collision avoidance policy for multiple\nrobots is challenging in the decentralized scenarios where each robot generate\nits paths without observing other robots' states and intents. While other\ndistributed multi-robot collision avoidance systems exist, they often require\nextracting agent-level features to plan a local collision-free action, which\ncan be computationally prohibitive and not robust. More importantly, in\npractice the performance of these methods are much lower than their centralized\ncounterparts.\n  We present a decentralized sensor-level collision avoidance policy for\nmulti-robot systems, which directly maps raw sensor measurements to an agent's\nsteering commands in terms of movement velocity. As a first step toward\nreducing the performance gap between decentralized and centralized methods, we\npresent a multi-scenario multi-stage training framework to find an optimal\npolicy which is trained over a large number of robots on rich, complex\nenvironments simultaneously using a policy gradient based reinforcement\nlearning algorithm. We validate the learned sensor-level collision avoidance\npolicy in a variety of simulated scenarios with thorough performance\nevaluations and show that the final learned policy is able to find time\nefficient, collision-free paths for a large-scale robot system. We also\ndemonstrate that the learned policy can be well generalized to new scenarios\nthat do not appear in the entire training period, including navigating a\nheterogeneous group of robots and a large-scale scenario with 100 robots.\nVideos are available at https://sites.google.com/view/drlmaca","url_abs":"http://arxiv.org/abs/1709.10082v3","url_pdf":"http://arxiv.org/pdf/1709.10082v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"towards-optimally-decentralized-multi-robot","repo_url":"https://github.com/Acmece/rl-collision-avoidance","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"towards-optimally-decentralized-multi-robot","repo_url":"https://github.com/mit-acl/gym-collision-avoidance","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"collision-avoidance","task_name":"Collision Avoidance"},{"task_slug":"deep-reinforcement-learning","task_name":"Deep Reinforcement Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1709.10082","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}