{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:OD7HCS2EESJFOYR6NPBF7GG6V5","short_pith_number":"pith:OD7HCS2E","schema_version":"1.0","canonical_sha256":"70fe714b44249257623e6bc25f98deaf731716161f53c1afcef3cacbc59aea10","source":{"kind":"arxiv","id":"2106.08050","version":1},"attestation_state":"computed","paper":{"title":"Residual Reinforcement Learning from Demonstrations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Cordelia Schmid, Gabriel Dulac-Arnold, Jean Ponce (WILLOW), Julien Mairal (Thoth), Minttu Alakuijala (WILLOW, Thoth)","submitted_at":"2021-06-15T11:16:49Z","abstract_excerpt":"Residual reinforcement learning (RL) has been proposed as a way to solve challenging robotic tasks by adapting control actions from a conventional feedback controller to maximize a reward signal. We extend the residual formulation to learn from visual inputs and sparse rewards using demonstrations. Learning from images, proprioceptive inputs and a sparse task-completion reward relaxes the requirement of accessing full state features, such as object and target positions. In addition, replacing the base controller with a policy learned from demonstrations removes the dependency on a hand-enginee"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2106.08050","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-06-15T11:16:49Z","cross_cats_sorted":[],"title_canon_sha256":"4579dcc12e7ac6a4c3f9b5ad3af651264533ad2302bb33a8b12b51ac7e6cc9dc","abstract_canon_sha256":"4e5cb3f529617ca406515af19d6183ce72dad256963b472732110ef38e02877a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:49:37.273808Z","signature_b64":"X7k7s3zIn30/4YqJEnMJRu87TIa6gfIBoyLDc1/cFyfE2/jzLuLBFzgcPzd5BzDQdUGgYGlFDnltWzxfE6+rBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"70fe714b44249257623e6bc25f98deaf731716161f53c1afcef3cacbc59aea10","last_reissued_at":"2026-07-05T02:49:37.273310Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:49:37.273310Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Residual Reinforcement Learning from Demonstrations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Cordelia Schmid, Gabriel Dulac-Arnold, Jean Ponce (WILLOW), Julien Mairal (Thoth), Minttu Alakuijala (WILLOW, Thoth)","submitted_at":"2021-06-15T11:16:49Z","abstract_excerpt":"Residual reinforcement learning (RL) has been proposed as a way to solve challenging robotic tasks by adapting control actions from a conventional feedback controller to maximize a reward signal. We extend the residual formulation to learn from visual inputs and sparse rewards using demonstrations. Learning from images, proprioceptive inputs and a sparse task-completion reward relaxes the requirement of accessing full state features, such as object and target positions. In addition, replacing the base controller with a policy learned from demonstrations removes the dependency on a hand-enginee"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.08050","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2106.08050/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2106.08050","created_at":"2026-07-05T02:49:37.273376+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.08050v1","created_at":"2026-07-05T02:49:37.273376+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.08050","created_at":"2026-07-05T02:49:37.273376+00:00"},{"alias_kind":"pith_short_12","alias_value":"OD7HCS2EESJF","created_at":"2026-07-05T02:49:37.273376+00:00"},{"alias_kind":"pith_short_16","alias_value":"OD7HCS2EESJFOYR6","created_at":"2026-07-05T02:49:37.273376+00:00"},{"alias_kind":"pith_short_8","alias_value":"OD7HCS2E","created_at":"2026-07-05T02:49:37.273376+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.11743","citing_title":"TacCoRL: Integrating Tactile Feedback into VLA via Simulation","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09825","citing_title":"An Agency-Transferring Model-Free Policy Enhancement Technique","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28323","citing_title":"DexCompose: Reusing Dexterous Policies for Multi-Task Manipulation with a Single Hand","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29209","citing_title":"AnyBody: Free-Form Whole-Body Humanoid Control from Arbitrary Keypoint Guidance","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2409.00588","citing_title":"Diffusion Policy Policy Optimization","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OD7HCS2EESJFOYR6NPBF7GG6V5","json":"https://pith.science/pith/OD7HCS2EESJFOYR6NPBF7GG6V5.json","graph_json":"https://pith.science/api/pith-number/OD7HCS2EESJFOYR6NPBF7GG6V5/graph.json","events_json":"https://pith.science/api/pith-number/OD7HCS2EESJFOYR6NPBF7GG6V5/events.json","paper":"https://pith.science/paper/OD7HCS2E"},"agent_actions":{"view_html":"https://pith.science/pith/OD7HCS2EESJFOYR6NPBF7GG6V5","download_json":"https://pith.science/pith/OD7HCS2EESJFOYR6NPBF7GG6V5.json","view_paper":"https://pith.science/paper/OD7HCS2E","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.08050&json=true","fetch_graph":"https://pith.science/api/pith-number/OD7HCS2EESJFOYR6NPBF7GG6V5/graph.json","fetch_events":"https://pith.science/api/pith-number/OD7HCS2EESJFOYR6NPBF7GG6V5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OD7HCS2EESJFOYR6NPBF7GG6V5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OD7HCS2EESJFOYR6NPBF7GG6V5/action/storage_attestation","attest_author":"https://pith.science/pith/OD7HCS2EESJFOYR6NPBF7GG6V5/action/author_attestation","sign_citation":"https://pith.science/pith/OD7HCS2EESJFOYR6NPBF7GG6V5/action/citation_signature","submit_replication":"https://pith.science/pith/OD7HCS2EESJFOYR6NPBF7GG6V5/action/replication_record"}},"created_at":"2026-07-05T02:49:37.273376+00:00","updated_at":"2026-07-05T02:49:37.273376+00:00"}