{"id":"W3195968524","doi":"10.1146/annurev-control-042920-020211","title":"Safe Learning in Robotics: From Learning-Based Control to Safe Reinforcement Learning","year":2022,"lang":"en","type":"article","venue":"Annual Review of Control Robotics and Autonomous Systems","topic":"Fault Detection and Control Systems","field":"Engineering","cited_by":660,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs; Canadian Institute for Advanced Research","keywords":"Reinforcement learning; Robot learning; Artificial intelligence; Computer science; Leverage (statistics); Machine learning; Robotics; Robot; Mobile robot","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.001660106,0.0004408624,0.00155127,0.0002840442,0.0002970051,0.00008210028,0.000290246,0.0001106535,0.00009919357],"category_scores_gemma":[0.0003142445,0.0004542482,0.0002321943,0.0003844745,0.00003322137,0.000115028,0.00005452057,0.0009514383,0.00004938739],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003211321,"about_ca_system_score_gemma":0.0001151294,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006146813,"about_ca_topic_score_gemma":0.00002677666,"domain_scores_codex":[0.9960051,0.0007905352,0.001581156,0.0004438226,0.0005851061,0.000594345],"domain_scores_gemma":[0.9983737,0.00047359,0.000434384,0.0003082599,0.0001665476,0.0002435216],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00006751193,0.00002796913,0.0009209095,0.001511273,0.0001460946,0.0000168192,0.0002667527,0.9924967,0.0003104138,0.0002885573,0.0001648013,0.003782207],"study_design_scores_gemma":[0.00282517,0.0006093041,0.0002010628,0.001924891,0.0001162594,0.00001039324,0.0007810866,0.8377615,0.000006101798,0.000003353219,0.15533,0.0004309266],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.006512772,0.3611277,0.5995567,0.004103024,0.005158684,0.01082208,0.0002547893,0.001467716,0.01099658],"genre_scores_gemma":[0.9961773,0.001547628,0.00005895902,0.0005407502,0.0001498367,0.0004027051,0.00004565819,0.00007862024,0.0009985992],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9896645,"threshold_uncertainty_score":0.9997909,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.005279335206119093,"score_gpt":0.2147044786788494,"score_spread":0.2094251434727303,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}