{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004161343,0.001712723,0.001585422,0.001166819,0.0006483689,0.003229772,0.002386997,0.002518417,0.00315355],"category_scores_gemma":[0.01061673,0.0007127648,0.001044019,0.001196674,0.006457054,0.004544152,0.003358696,0.006008583,0.0007568391],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002281068,"about_ca_system_score_gemma":0.002378813,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003395956,"about_ca_topic_score_gemma":0.001926162,"domain_scores_codex":[0.997332,0.001013092,0.000201174,0.0004049534,0.0009038814,0.0001449289],"domain_scores_gemma":[0.9929807,0.005283514,0.0003616352,0.000491829,0.0006976945,0.0001846483],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00009241549,0.00007801083,0.0007589103,0.001410746,0.0001160258,0.00009189801,0.0002742282,0.3645077,0.0009031619,0.419975,0.005307472,0.2064845],"study_design_scores_gemma":[0.00003341114,0.0001625526,0.000208398,0.0005286781,0.00003241124,0.00006711175,0.00008039551,0.4642808,0.001509424,0.5074701,0.02557235,0.00005445625],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"review","genre_scores_codex":[0.001906133,0.02378568,0.9648454,0.00275301,0.0003241038,0.00004674229,0.00004261611,0.0003242813,0.005971925],"genre_scores_gemma":[0.4718748,0.1034508,0.4061493,0.003266368,0.003270364,0.0007009726,0.000344664,0.0006728695,0.01026978],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.004161343,"threshold_uncertainty_score":0.02200752,"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."}}