{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:D4P2DV6LIIIIRGCRDXKHAA5O37","short_pith_number":"pith:D4P2DV6L","schema_version":"1.0","canonical_sha256":"1f1fa1d7cb42108898511dd47003aedfee63f763e5a2e5f753c1aedf64a975dc","source":{"kind":"arxiv","id":"2508.01539","version":1},"attestation_state":"computed","paper":{"title":"HALO: Human Preference Aligned Offline Reward Learning for Robot Navigation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.RO","authors_text":"Amogha Thalihalla Sunil, Dinesh Manocha, Gershom Seneviratne, Jianyu An, Jonathan Deepak Kannan, Kasun Weerakoon, Mohamed Bashir Elnoor, Sahire Ellahy","submitted_at":"2025-08-03T01:46:10Z","abstract_excerpt":"In this paper, we introduce HALO, a novel Offline Reward Learning algorithm that quantifies human intuition in navigation into a vision-based reward function for robot navigation. HALO learns a reward model from offline data, leveraging expert trajectories collected from mobile robots. During training, actions are uniformly sampled around a reference action and ranked using preference scores derived from a Boltzmann distribution centered on the preferred action, and shaped based on binary user feedback to intuitive navigation queries. The reward model is trained via the Plackett-Luce loss to a"},"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":"2508.01539","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.RO","submitted_at":"2025-08-03T01:46:10Z","cross_cats_sorted":[],"title_canon_sha256":"7096a60e5e4aa8d09a86b0557462dc86be89ab3f187edaddeb39c41bd9546a27","abstract_canon_sha256":"bc2e9e1224c1999b6cc3401d3812d2799b24d510a12aa392ce1dfc2ae4c82519"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:47:40.689594Z","signature_b64":"+9fJwy3g2rCNKtVNvHJPgTCaBfp1VW2U8w/jy3iMKrJ7v8W2VPyotKoHhEBxdZ9DchPAMatZAdXJMBp1ICfEAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1f1fa1d7cb42108898511dd47003aedfee63f763e5a2e5f753c1aedf64a975dc","last_reissued_at":"2026-07-05T11:47:40.689079Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:47:40.689079Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HALO: Human Preference Aligned Offline Reward Learning for Robot Navigation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.RO","authors_text":"Amogha Thalihalla Sunil, Dinesh Manocha, Gershom Seneviratne, Jianyu An, Jonathan Deepak Kannan, Kasun Weerakoon, Mohamed Bashir Elnoor, Sahire Ellahy","submitted_at":"2025-08-03T01:46:10Z","abstract_excerpt":"In this paper, we introduce HALO, a novel Offline Reward Learning algorithm that quantifies human intuition in navigation into a vision-based reward function for robot navigation. HALO learns a reward model from offline data, leveraging expert trajectories collected from mobile robots. During training, actions are uniformly sampled around a reference action and ranked using preference scores derived from a Boltzmann distribution centered on the preferred action, and shaped based on binary user feedback to intuitive navigation queries. The reward model is trained via the Plackett-Luce loss to a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.01539","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/2508.01539/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":"2508.01539","created_at":"2026-07-05T11:47:40.689143+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.01539v1","created_at":"2026-07-05T11:47:40.689143+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.01539","created_at":"2026-07-05T11:47:40.689143+00:00"},{"alias_kind":"pith_short_12","alias_value":"D4P2DV6LIIII","created_at":"2026-07-05T11:47:40.689143+00:00"},{"alias_kind":"pith_short_16","alias_value":"D4P2DV6LIIIIRGCR","created_at":"2026-07-05T11:47:40.689143+00:00"},{"alias_kind":"pith_short_8","alias_value":"D4P2DV6L","created_at":"2026-07-05T11:47:40.689143+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18426","citing_title":"VEGA: Learning Navigation VLAs from In-the-Wild Egocentric Video with Geometric Trajectory Supervision","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12603","citing_title":"From Imitation to Alignment: Human-Preference Flow Policies for Long-Horizon Sidewalk Navigation","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2603.17510","citing_title":"Interpreting Context-Aware Human Preferences for Multi-Objective Robot Navigation","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/D4P2DV6LIIIIRGCRDXKHAA5O37","json":"https://pith.science/pith/D4P2DV6LIIIIRGCRDXKHAA5O37.json","graph_json":"https://pith.science/api/pith-number/D4P2DV6LIIIIRGCRDXKHAA5O37/graph.json","events_json":"https://pith.science/api/pith-number/D4P2DV6LIIIIRGCRDXKHAA5O37/events.json","paper":"https://pith.science/paper/D4P2DV6L"},"agent_actions":{"view_html":"https://pith.science/pith/D4P2DV6LIIIIRGCRDXKHAA5O37","download_json":"https://pith.science/pith/D4P2DV6LIIIIRGCRDXKHAA5O37.json","view_paper":"https://pith.science/paper/D4P2DV6L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.01539&json=true","fetch_graph":"https://pith.science/api/pith-number/D4P2DV6LIIIIRGCRDXKHAA5O37/graph.json","fetch_events":"https://pith.science/api/pith-number/D4P2DV6LIIIIRGCRDXKHAA5O37/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/D4P2DV6LIIIIRGCRDXKHAA5O37/action/timestamp_anchor","attest_storage":"https://pith.science/pith/D4P2DV6LIIIIRGCRDXKHAA5O37/action/storage_attestation","attest_author":"https://pith.science/pith/D4P2DV6LIIIIRGCRDXKHAA5O37/action/author_attestation","sign_citation":"https://pith.science/pith/D4P2DV6LIIIIRGCRDXKHAA5O37/action/citation_signature","submit_replication":"https://pith.science/pith/D4P2DV6LIIIIRGCRDXKHAA5O37/action/replication_record"}},"created_at":"2026-07-05T11:47:40.689143+00:00","updated_at":"2026-07-05T11:47:40.689143+00:00"}