{"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/emergence-of-locomotion-behaviours-in-rich","title":"Emergence of Locomotion Behaviours in Rich Environments","arxiv_id":"1707.02286","date":"2017-07-07","proceeding":null,"authors":["Nicolas Heess","Dhruva TB","Srinivasan Sriram","Jay Lemmon","Josh Merel","Greg Wayne","Yuval Tassa","Tom Erez","Ziyu Wang","S. M. Ali Eslami","Martin Riedmiller","David Silver"],"abstract":"The reinforcement learning paradigm allows, in principle, for complex\nbehaviours to be learned directly from simple reward signals. In practice,\nhowever, it is common to carefully hand-design the reward function to encourage\na particular solution, or to derive it from demonstration data. In this paper\nexplore how a rich environment can help to promote the learning of complex\nbehavior. Specifically, we train agents in diverse environmental contexts, and\nfind that this encourages the emergence of robust behaviours that perform well\nacross a suite of tasks. We demonstrate this principle for locomotion --\nbehaviours that are known for their sensitivity to the choice of reward. We\ntrain several simulated bodies on a diverse set of challenging terrains and\nobstacles, using a simple reward function based on forward progress. Using a\nnovel scalable variant of policy gradient reinforcement learning, our agents\nlearn to run, jump, crouch and turn as required by the environment without\nexplicit reward-based guidance. A visual depiction of highlights of the learned\nbehavior can be viewed following https://youtu.be/hx_bgoTF7bs .","url_abs":"http://arxiv.org/abs/1707.02286v2","url_pdf":"http://arxiv.org/pdf/1707.02286v2.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":"emergence-of-locomotion-behaviours-in-rich","repo_url":"https://github.com/Sohojoe/ActiveRagdollAssaultCourse","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}},{"paper_slug":"emergence-of-locomotion-behaviours-in-rich","repo_url":"https://github.com/Unity-Technologies/marathon-envs","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}},{"paper_slug":"emergence-of-locomotion-behaviours-in-rich","repo_url":"https://github.com/llSourcell/OpenAI_Five_vs_Dota2_Explained","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"emergence-of-locomotion-behaviours-in-rich","repo_url":"https://github.com/magnusja/ppo","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"emergence-of-locomotion-behaviours-in-rich","repo_url":"https://github.com/pat-coady/trpo","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"emergence-of-locomotion-behaviours-in-rich","repo_url":"https://github.com/tensorlayer/RLzoo","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"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=1707.02286","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}