{"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/learning-under-misspecified-objective-spaces","title":"Learning under Misspecified Objective Spaces","arxiv_id":"1810.05157","date":"2018-10-11","proceeding":null,"authors":["Andreea Bobu","Andrea Bajcsy","Jaime F. Fisac","Anca D. Dragan"],"abstract":"Learning robot objective functions from human input has become increasingly\nimportant, but state-of-the-art techniques assume that the human's desired\nobjective lies within the robot's hypothesis space. When this is not true, even\nmethods that keep track of uncertainty over the objective fail because they\nreason about which hypothesis might be correct, and not whether any of the\nhypotheses are correct. We focus specifically on learning from physical human\ncorrections during the robot's task execution, where not having a rich enough\nhypothesis space leads to the robot updating its objective in ways that the\nperson did not actually intend. We observe that such corrections appear\nirrelevant to the robot, because they are not the best way of achieving any of\nthe candidate objectives. Instead of naively trusting and learning from every\nhuman interaction, we propose robots learn conservatively by reasoning in real\ntime about how relevant the human's correction is for the robot's hypothesis\nspace. We test our inference method in an experiment with human interaction\ndata, and demonstrate that this alleviates unintended learning in an in-person\nuser study with a 7DoF robot manipulator.","url_abs":"http://arxiv.org/abs/1810.05157v4","url_pdf":"http://arxiv.org/pdf/1810.05157v4.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":"learning-under-misspecified-objective-spaces","repo_url":"https://github.com/andreea7b/beta_adaptive_pHRI","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1810.05157","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}