{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ZN7FNZKISECDPB5LY2UPOUPDKT","short_pith_number":"pith:ZN7FNZKI","schema_version":"1.0","canonical_sha256":"cb7e56e54891043787abc6a8f751e354f67adbbb24f1d5ca6d79fc30e9a5fbf0","source":{"kind":"arxiv","id":"2604.17473","version":2},"attestation_state":"computed","paper":{"title":"Dual-Anchoring: Addressing State Drift in Vision-Language Navigation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Explicit anchoring of instruction progress and memory landmarks prevents state drift in vision-language navigation agents.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Jianyi Liu, Jinjun Wang, Kailin Lyu, Kangyi Wu, Lin Zhao, Pengna Li, Qingrong He, Xi Lin","submitted_at":"2026-04-19T15:03:38Z","abstract_excerpt":"Vision-Language Navigation(VLN) requires an agent to navigate through 3D environments by following natural language instructions. While recent Video Large Language Models(Video-LLMs) have largely advanced VLN, they remain highly susceptible to State Drift in long scenarios. In these cases, the agent's internal state drifts away from the true task execution state, leading to aimless wandering and failure to execute essential maneuvers in the instruction. We attribute this failure to two distinct cognitive deficits: Progress Drift, where the agent fails to distinguish completed sub-goals from re"},"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":"2604.17473","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-04-19T15:03:38Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"f787b49f8f5add1533acb765e0f08df29ceaaaf04f3ba69578e5c0d32443d839","abstract_canon_sha256":"d1a10c8d8ab31f5ceca4fd60014da52a18497aa23214775c935ff899fe47b443"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-22T01:04:02.655901Z","signature_b64":"8d3lRR9VRbx0uFiPd4QhEJ5bTLhRl3Tl1m1F3ADFX2adjaphQhhE0pep6CdBu9LQwXvGi34PsD0Ijz5A2PuKDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cb7e56e54891043787abc6a8f751e354f67adbbb24f1d5ca6d79fc30e9a5fbf0","last_reissued_at":"2026-05-22T01:04:02.655163Z","signature_status":"signed_v1","first_computed_at":"2026-05-22T01:04:02.655163Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Dual-Anchoring: Addressing State Drift in Vision-Language Navigation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Explicit anchoring of instruction progress and memory landmarks prevents state drift in vision-language navigation agents.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Jianyi Liu, Jinjun Wang, Kailin Lyu, Kangyi Wu, Lin Zhao, Pengna Li, Qingrong He, Xi Lin","submitted_at":"2026-04-19T15:03:38Z","abstract_excerpt":"Vision-Language Navigation(VLN) requires an agent to navigate through 3D environments by following natural language instructions. While recent Video Large Language Models(Video-LLMs) have largely advanced VLN, they remain highly susceptible to State Drift in long scenarios. In these cases, the agent's internal state drifts away from the true task execution state, leading to aimless wandering and failure to execute essential maneuvers in the instruction. We attribute this failure to two distinct cognitive deficits: Progress Drift, where the agent fails to distinguish completed sub-goals from re"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"The Dual-Anchoring Framework explicitly anchors the instruction progress and history representations, achieving a 15.2% improvement in Success Rate and a remarkable 24.7% gain on long-horizon trajectories.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That adding structured progress tokens and retrospective landmark prediction will reliably prevent state drift without introducing new failure modes or requiring task-specific tuning that does not generalize.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Dual-Anchoring adds explicit progress tokens and retrospective landmark verification to VLN agents, cutting state drift and lifting success rate 15.2% overall with 24.7% gains on long trajectories.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Explicit anchoring of instruction progress and memory landmarks prevents state drift in vision-language navigation agents.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"1e8bddbaf26664ce70ecf57a7f08aabaaa91fdf532f1f83d772d33d5192e469f"},"source":{"id":"2604.17473","kind":"arxiv","version":2},"verdict":{"id":"9a1116ff-0d2b-454e-8101-82e8239168a4","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T06:46:38.680197Z","strongest_claim":"The Dual-Anchoring Framework explicitly anchors the instruction progress and history representations, achieving a 15.2% improvement in Success Rate and a remarkable 24.7% gain on long-horizon trajectories.","one_line_summary":"Dual-Anchoring adds explicit progress tokens and retrospective landmark verification to VLN agents, cutting state drift and lifting success rate 15.2% overall with 24.7% gains on long trajectories.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That adding structured progress tokens and retrospective landmark prediction will reliably prevent state drift without introducing new failure modes or requiring task-specific tuning that does not generalize.","pith_extraction_headline":"Explicit anchoring of instruction progress and memory landmarks prevents state drift in vision-language navigation agents."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.17473/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":"2604.17473","created_at":"2026-05-22T01:04:02.655278+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.17473v2","created_at":"2026-05-22T01:04:02.655278+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.17473","created_at":"2026-05-22T01:04:02.655278+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZN7FNZKISECD","created_at":"2026-05-22T01:04:02.655278+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZN7FNZKISECDPB5L","created_at":"2026-05-22T01:04:02.655278+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZN7FNZKI","created_at":"2026-05-22T01:04:02.655278+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2604.27620","citing_title":"SpaAct: Spatially-Activated Transition Learning with Curriculum Adaptation for Vision-Language Navigation","ref_index":71,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZN7FNZKISECDPB5LY2UPOUPDKT","json":"https://pith.science/pith/ZN7FNZKISECDPB5LY2UPOUPDKT.json","graph_json":"https://pith.science/api/pith-number/ZN7FNZKISECDPB5LY2UPOUPDKT/graph.json","events_json":"https://pith.science/api/pith-number/ZN7FNZKISECDPB5LY2UPOUPDKT/events.json","paper":"https://pith.science/paper/ZN7FNZKI"},"agent_actions":{"view_html":"https://pith.science/pith/ZN7FNZKISECDPB5LY2UPOUPDKT","download_json":"https://pith.science/pith/ZN7FNZKISECDPB5LY2UPOUPDKT.json","view_paper":"https://pith.science/paper/ZN7FNZKI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.17473&json=true","fetch_graph":"https://pith.science/api/pith-number/ZN7FNZKISECDPB5LY2UPOUPDKT/graph.json","fetch_events":"https://pith.science/api/pith-number/ZN7FNZKISECDPB5LY2UPOUPDKT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZN7FNZKISECDPB5LY2UPOUPDKT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZN7FNZKISECDPB5LY2UPOUPDKT/action/storage_attestation","attest_author":"https://pith.science/pith/ZN7FNZKISECDPB5LY2UPOUPDKT/action/author_attestation","sign_citation":"https://pith.science/pith/ZN7FNZKISECDPB5LY2UPOUPDKT/action/citation_signature","submit_replication":"https://pith.science/pith/ZN7FNZKISECDPB5LY2UPOUPDKT/action/replication_record"}},"created_at":"2026-05-22T01:04:02.655278+00:00","updated_at":"2026-05-22T01:04:02.655278+00:00"}