{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:E7OLPPMG7RRX5MRQCZJVV3T2SO","short_pith_number":"pith:E7OLPPMG","schema_version":"1.0","canonical_sha256":"27dcb7bd86fc637eb23016535aee7a93b31c3db889b5ed803fc8762a04358330","source":{"kind":"arxiv","id":"2502.08643","version":2},"attestation_state":"computed","paper":{"title":"A Real-to-Sim-to-Real Approach to Robotic Manipulation with VLM-Generated Iterative Keypoint Rewards","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.RO","authors_text":"Hooshang Nayyeri, Li Fei-Fei, Shivansh Patel, Shubham Garg, Svetlana Lazebnik, Wenlong Huang, Xinchen Yin, Yunzhu Li","submitted_at":"2025-02-12T18:57:22Z","abstract_excerpt":"Task specification for robotic manipulation in open-world environments is challenging, requiring flexible and adaptive objectives that align with human intentions and can evolve through iterative feedback. We introduce Iterative Keypoint Reward (IKER), a visually grounded, Python-based reward function that serves as a dynamic task specification. Our framework leverages VLMs to generate and refine these reward functions for multi-step manipulation tasks. Given RGB-D observations and free-form language instructions, we sample keypoints in the scene and generate a reward function conditioned on t"},"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":"2502.08643","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2025-02-12T18:57:22Z","cross_cats_sorted":["cs.AI","cs.CV"],"title_canon_sha256":"561d9ac99436f53c78e9c40ae955212c8c27c1e9b218b5995be148f691dfe97b","abstract_canon_sha256":"c1b32c6877bbfe331a2305feba9e809b446b4b0513f6d09677d958fbf02d1ab4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:16:21.011095Z","signature_b64":"N/8ae1KDZOoVaJHViutqGjcBfPvVbKynFCRlWC3Q67f0CFNAXpTUtFlqU/1LI5e4CJ8fVVEwxPqn0B7X8CgQCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"27dcb7bd86fc637eb23016535aee7a93b31c3db889b5ed803fc8762a04358330","last_reissued_at":"2026-07-05T10:16:21.010612Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:16:21.010612Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Real-to-Sim-to-Real Approach to Robotic Manipulation with VLM-Generated Iterative Keypoint Rewards","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.RO","authors_text":"Hooshang Nayyeri, Li Fei-Fei, Shivansh Patel, Shubham Garg, Svetlana Lazebnik, Wenlong Huang, Xinchen Yin, Yunzhu Li","submitted_at":"2025-02-12T18:57:22Z","abstract_excerpt":"Task specification for robotic manipulation in open-world environments is challenging, requiring flexible and adaptive objectives that align with human intentions and can evolve through iterative feedback. We introduce Iterative Keypoint Reward (IKER), a visually grounded, Python-based reward function that serves as a dynamic task specification. Our framework leverages VLMs to generate and refine these reward functions for multi-step manipulation tasks. Given RGB-D observations and free-form language instructions, we sample keypoints in the scene and generate a reward function conditioned on t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.08643","kind":"arxiv","version":2},"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/2502.08643/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":"2502.08643","created_at":"2026-07-05T10:16:21.010668+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.08643v2","created_at":"2026-07-05T10:16:21.010668+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.08643","created_at":"2026-07-05T10:16:21.010668+00:00"},{"alias_kind":"pith_short_12","alias_value":"E7OLPPMG7RRX","created_at":"2026-07-05T10:16:21.010668+00:00"},{"alias_kind":"pith_short_16","alias_value":"E7OLPPMG7RRX5MRQ","created_at":"2026-07-05T10:16:21.010668+00:00"},{"alias_kind":"pith_short_8","alias_value":"E7OLPPMG","created_at":"2026-07-05T10:16:21.010668+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2507.00990","citing_title":"Robotic Manipulation by Imitating Generated Videos Without Physical Demonstrations","ref_index":92,"is_internal_anchor":false},{"citing_arxiv_id":"2512.01773","citing_title":"IGen: Scalable Data Generation for Robot Learning from Open-World Images","ref_index":51,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/E7OLPPMG7RRX5MRQCZJVV3T2SO","json":"https://pith.science/pith/E7OLPPMG7RRX5MRQCZJVV3T2SO.json","graph_json":"https://pith.science/api/pith-number/E7OLPPMG7RRX5MRQCZJVV3T2SO/graph.json","events_json":"https://pith.science/api/pith-number/E7OLPPMG7RRX5MRQCZJVV3T2SO/events.json","paper":"https://pith.science/paper/E7OLPPMG"},"agent_actions":{"view_html":"https://pith.science/pith/E7OLPPMG7RRX5MRQCZJVV3T2SO","download_json":"https://pith.science/pith/E7OLPPMG7RRX5MRQCZJVV3T2SO.json","view_paper":"https://pith.science/paper/E7OLPPMG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.08643&json=true","fetch_graph":"https://pith.science/api/pith-number/E7OLPPMG7RRX5MRQCZJVV3T2SO/graph.json","fetch_events":"https://pith.science/api/pith-number/E7OLPPMG7RRX5MRQCZJVV3T2SO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/E7OLPPMG7RRX5MRQCZJVV3T2SO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/E7OLPPMG7RRX5MRQCZJVV3T2SO/action/storage_attestation","attest_author":"https://pith.science/pith/E7OLPPMG7RRX5MRQCZJVV3T2SO/action/author_attestation","sign_citation":"https://pith.science/pith/E7OLPPMG7RRX5MRQCZJVV3T2SO/action/citation_signature","submit_replication":"https://pith.science/pith/E7OLPPMG7RRX5MRQCZJVV3T2SO/action/replication_record"}},"created_at":"2026-07-05T10:16:21.010668+00:00","updated_at":"2026-07-05T10:16:21.010668+00:00"}