{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:C5WEWKNEC5LUFPWMHOWJIXE6AK","short_pith_number":"pith:C5WEWKNE","schema_version":"1.0","canonical_sha256":"176c4b29a4175742becc3bac945c9e02b60fc220bdbece9c53ac9c8be0c1bf93","source":{"kind":"arxiv","id":"2604.04834","version":2},"attestation_state":"computed","paper":{"title":"E-VLA: Event-Augmented Vision-Language-Action Model for Dark and Blurred Scenes","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Event-augmented VLA models restore robotic manipulation success in dark and blurred scenes via direct event fusion.","cross_cats":["cs.MM","cs.RO","eess.IV"],"primary_cat":"cs.CV","authors_text":"Hao Shi, Jiajun Zhai, Kailun Yang, Kaiwei Wang, Shangwei Guo","submitted_at":"2026-04-06T16:35:57Z","abstract_excerpt":"Robotic Vision-Language-Action (VLA) models generalize well for open-ended manipulation, but their perception is fragile under sensing-stage degradations such as extreme low light, motion blur, and black clipping. We present E-VLA, an event-augmented VLA framework that improves manipulation robustness when conventional frame-based vision becomes unreliable. Instead of reconstructing images from events, E-VLA directly leverages motion and structural cues in event streams to preserve semantic perception and perception-action consistency under adverse conditions. We build an open-source teleopera"},"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":true},"canonical_record":{"source":{"id":"2604.04834","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-04-06T16:35:57Z","cross_cats_sorted":["cs.MM","cs.RO","eess.IV"],"title_canon_sha256":"a172fe623d8b02d0cc8ebec9ee457523924914d082a250c7350a660b243af532","abstract_canon_sha256":"913b2537ba2df974ab24be41387af13d9e3b8e67b330c85ee562f53db9522bdc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-01T01:17:50.121891Z","signature_b64":"j9q4jDDICEXdKzw6cFJ3A8jb71TC1mNFCX+LCgLJrIeue6ayyA/uoLiatrYimY4RECTe/At5f43Sjlj3BdpWCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"176c4b29a4175742becc3bac945c9e02b60fc220bdbece9c53ac9c8be0c1bf93","last_reissued_at":"2026-07-01T01:17:50.121388Z","signature_status":"signed_v1","first_computed_at":"2026-07-01T01:17:50.121388Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"E-VLA: Event-Augmented Vision-Language-Action Model for Dark and Blurred Scenes","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Event-augmented VLA models restore robotic manipulation success in dark and blurred scenes via direct event fusion.","cross_cats":["cs.MM","cs.RO","eess.IV"],"primary_cat":"cs.CV","authors_text":"Hao Shi, Jiajun Zhai, Kailun Yang, Kaiwei Wang, Shangwei Guo","submitted_at":"2026-04-06T16:35:57Z","abstract_excerpt":"Robotic Vision-Language-Action (VLA) models generalize well for open-ended manipulation, but their perception is fragile under sensing-stage degradations such as extreme low light, motion blur, and black clipping. We present E-VLA, an event-augmented VLA framework that improves manipulation robustness when conventional frame-based vision becomes unreliable. Instead of reconstructing images from events, E-VLA directly leverages motion and structural cues in event streams to preserve semantic perception and perception-action consistency under adverse conditions. We build an open-source teleopera"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"even a simple parameter-free fusion, i.e., overlaying accumulated event maps onto RGB images, could substantially improve robustness in dark and blur-heavy scenes: on Pick-Place at 20 lux, success increases from 0% (image-only) to 60% with overlay fusion and to 90% with our event adapter; under severe motion blur (1000 ms exposure), Pick-Place improves from 0% to 20-25%, and Sorting from 5% to 32.5%.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the collected real-world RGB-event-action dataset and the chosen tasks/illumination settings are representative enough for the reported robustness gains to generalize to other robots, tasks, and VLA backbones without substantial retraining or hyperparameter retuning.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"E-VLA integrates event streams directly into VLA models via lightweight fusion, raising Pick-Place success from 0% to 60-90% at 20 lux and from 0% to 20-25% under severe motion blur.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Event-augmented VLA models restore robotic manipulation success in dark and blurred scenes via direct event fusion.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"3d65153ec6afbf62319dbeb006c5f4f070803558fd013aadbc044e31e796e2dc"},"source":{"id":"2604.04834","kind":"arxiv","version":2},"verdict":{"id":"f892352b-f3f6-4ec2-bd84-593ebb0a2bec","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T20:18:16.271874Z","strongest_claim":"even a simple parameter-free fusion, i.e., overlaying accumulated event maps onto RGB images, could substantially improve robustness in dark and blur-heavy scenes: on Pick-Place at 20 lux, success increases from 0% (image-only) to 60% with overlay fusion and to 90% with our event adapter; under severe motion blur (1000 ms exposure), Pick-Place improves from 0% to 20-25%, and Sorting from 5% to 32.5%.","one_line_summary":"E-VLA integrates event streams directly into VLA models via lightweight fusion, raising Pick-Place success from 0% to 60-90% at 20 lux and from 0% to 20-25% under severe motion blur.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the collected real-world RGB-event-action dataset and the chosen tasks/illumination settings are representative enough for the reported robustness gains to generalize to other robots, tasks, and VLA backbones without substantial retraining or hyperparameter retuning.","pith_extraction_headline":"Event-augmented VLA models restore robotic manipulation success in dark and blurred scenes via direct event fusion."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.04834/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":2,"snapshot_sha256":"5eaf2cd823fab9ac8be449304adc1b70a8e444b12286df165c1119bdd010c273"},"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":"2604.04834","created_at":"2026-07-01T01:17:50.121454+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.04834v2","created_at":"2026-07-01T01:17:50.121454+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.04834","created_at":"2026-07-01T01:17:50.121454+00:00"},{"alias_kind":"pith_short_12","alias_value":"C5WEWKNEC5LU","created_at":"2026-07-01T01:17:50.121454+00:00"},{"alias_kind":"pith_short_16","alias_value":"C5WEWKNEC5LUFPWM","created_at":"2026-07-01T01:17:50.121454+00:00"},{"alias_kind":"pith_short_8","alias_value":"C5WEWKNE","created_at":"2026-07-01T01:17:50.121454+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2606.30113","citing_title":"SA-VLA: State-aware tokenizer for improving Vision-Language-Action Models' performance","ref_index":18,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/C5WEWKNEC5LUFPWMHOWJIXE6AK","json":"https://pith.science/pith/C5WEWKNEC5LUFPWMHOWJIXE6AK.json","graph_json":"https://pith.science/api/pith-number/C5WEWKNEC5LUFPWMHOWJIXE6AK/graph.json","events_json":"https://pith.science/api/pith-number/C5WEWKNEC5LUFPWMHOWJIXE6AK/events.json","paper":"https://pith.science/paper/C5WEWKNE"},"agent_actions":{"view_html":"https://pith.science/pith/C5WEWKNEC5LUFPWMHOWJIXE6AK","download_json":"https://pith.science/pith/C5WEWKNEC5LUFPWMHOWJIXE6AK.json","view_paper":"https://pith.science/paper/C5WEWKNE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.04834&json=true","fetch_graph":"https://pith.science/api/pith-number/C5WEWKNEC5LUFPWMHOWJIXE6AK/graph.json","fetch_events":"https://pith.science/api/pith-number/C5WEWKNEC5LUFPWMHOWJIXE6AK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/C5WEWKNEC5LUFPWMHOWJIXE6AK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/C5WEWKNEC5LUFPWMHOWJIXE6AK/action/storage_attestation","attest_author":"https://pith.science/pith/C5WEWKNEC5LUFPWMHOWJIXE6AK/action/author_attestation","sign_citation":"https://pith.science/pith/C5WEWKNEC5LUFPWMHOWJIXE6AK/action/citation_signature","submit_replication":"https://pith.science/pith/C5WEWKNEC5LUFPWMHOWJIXE6AK/action/replication_record"}},"created_at":"2026-07-01T01:17:50.121454+00:00","updated_at":"2026-07-01T01:17:50.121454+00:00"}