{"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/imitation-from-observation-learning-to","title":"Imitation from Observation: Learning to Imitate Behaviors from Raw Video via Context Translation","arxiv_id":"1707.03374","date":"2017-07-11","proceeding":null,"authors":["YuXuan Liu","Abhishek Gupta","Pieter Abbeel","Sergey Levine"],"abstract":"Imitation learning is an effective approach for autonomous systems to acquire\ncontrol policies when an explicit reward function is unavailable, using\nsupervision provided as demonstrations from an expert, typically a human\noperator. However, standard imitation learning methods assume that the agent\nreceives examples of observation-action tuples that could be provided, for\ninstance, to a supervised learning algorithm. This stands in contrast to how\nhumans and animals imitate: we observe another person performing some behavior\nand then figure out which actions will realize that behavior, compensating for\nchanges in viewpoint, surroundings, object positions and types, and other\nfactors. We term this kind of imitation learning \"imitation-from-observation,\"\nand propose an imitation learning method based on video prediction with context\ntranslation and deep reinforcement learning. This lifts the assumption in\nimitation learning that the demonstration should consist of observations in the\nsame environment configuration, and enables a variety of interesting\napplications, including learning robotic skills that involve tool use simply by\nobserving videos of human tool use. Our experimental results show the\neffectiveness of our approach in learning a wide range of real-world robotic\ntasks modeled after common household chores from videos of a human\ndemonstrator, including sweeping, ladling almonds, pushing objects as well as a\nnumber of tasks in simulation.","url_abs":"http://arxiv.org/abs/1707.03374v2","url_pdf":"http://arxiv.org/pdf/1707.03374v2.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":"imitation-from-observation-learning-to","repo_url":"https://github.com/wyndwarrior/imitation_from_observation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"deep-reinforcement-learning","task_name":"Deep Reinforcement Learning"},{"task_slug":"imitation-learning","task_name":"Imitation Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"translation","task_name":"Translation"},{"task_slug":"video-prediction","task_name":"Video Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1707.03374","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}