{"paper":{"title":"MemoryVLA: Perceptual-Cognitive Memory in Vision-Language-Action Models for Robotic Manipulation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"MemoryVLA adds a perceptual-cognitive memory bank to vision-language-action models to supply temporal context for long-horizon robotic manipulation.","cross_cats":["cs.CV"],"primary_cat":"cs.RO","authors_text":"Bin Xie, Erjin Zhou, Fengrong Liu, Gao Huang, Haoqiang Fan, Hao Shi, Lin Sun, Tiancai Wang, Xiangyu Zhang, Yingfei Liu","submitted_at":"2025-08-26T17:57:16Z","abstract_excerpt":"Temporal context is essential for robotic manipulation because such tasks are inherently non-Markovian, yet mainstream VLA models typically overlook it and struggle with long-horizon, temporally dependent tasks. Cognitive science suggests that humans rely on working memory to buffer short-lived representations for immediate control, while the hippocampal system preserves verbatim episodic details and semantic gist of past experience for long-term memory. Inspired by these mechanisms, we propose MemoryVLA, a Cognition-Memory-Action framework for long-horizon robotic manipulation. A pretrained V"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"On 12 real-world tasks spanning general skills and long-horizon temporal dependencies, MemoryVLA achieves 84.0% success rate, with long-horizon tasks showing a +26 improvement over state-of-the-art baseline.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the adaptive retrieval, fusion, and redundancy-merging operations in the Perceptual-Cognitive Memory Bank will reliably supply temporally relevant context without introducing noise or stale entries that degrade action generation.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"MemoryVLA introduces a perceptual-cognitive memory bank and working-memory retrieval mechanism into VLA models, raising success rates on long-horizon robotic tasks by up to 26 points over prior baselines.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"MemoryVLA adds a perceptual-cognitive memory bank to vision-language-action models to supply temporal context for long-horizon robotic manipulation.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"a83598c45f7663ac45a09f74942c4c937c4139903d790b76888d527ad51e71a7"},"source":{"id":"2508.19236","kind":"arxiv","version":2},"verdict":{"id":"729cc02f-cd30-434c-9c26-de96826b76da","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T20:39:39.773638Z","strongest_claim":"On 12 real-world tasks spanning general skills and long-horizon temporal dependencies, MemoryVLA achieves 84.0% success rate, with long-horizon tasks showing a +26 improvement over state-of-the-art baseline.","one_line_summary":"MemoryVLA introduces a perceptual-cognitive memory bank and working-memory retrieval mechanism into VLA models, raising success rates on long-horizon robotic tasks by up to 26 points over prior baselines.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the adaptive retrieval, fusion, and redundancy-merging operations in the Perceptual-Cognitive Memory Bank will reliably supply temporally relevant context without introducing noise or stale entries that degrade action generation.","pith_extraction_headline":"MemoryVLA adds a perceptual-cognitive memory bank to vision-language-action models to supply temporal context for long-horizon robotic manipulation."},"references":{"count":41,"sample":[{"doi":"","year":null,"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","ref_index":1,"cited_arxiv_id":"2303.08774","is_internal_anchor":true},{"doi":"","year":null,"title":"Qwen Technical Report","work_id":"bb1fd52f-6b2f-437c-9516-37bdf6eb9be8","ref_index":2,"cited_arxiv_id":"2309.16609","is_internal_anchor":true},{"doi":"","year":null,"title":"RT-1: Robotics Transformer for Real-World Control at Scale","work_id":"e11bda85-8531-46bc-a07f-d0ade3643ab1","ref_index":3,"cited_arxiv_id":"2212.06817","is_internal_anchor":true},{"doi":"","year":null,"title":"RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control","work_id":"ff438a8a-8003-4fae-9131-acd418b3597b","ref_index":4,"cited_arxiv_id":"2307.15818","is_internal_anchor":true},{"doi":"","year":null,"title":"AgiBot World Colosseo: A Large-scale Manipulation Platform for Scalable and Intelligent Embodied Systems","work_id":"f797e9ec-510f-43a7-8a0c-18009ce332e5","ref_index":5,"cited_arxiv_id":"2503.06669","is_internal_anchor":true}],"resolved_work":41,"snapshot_sha256":"b812842533576896938c3bdbb8ddee537924dfed431ff75851c6312045c12ade","internal_anchors":20},"formal_canon":{"evidence_count":2,"snapshot_sha256":"ca37bcb2a8e6f0961e446b609276285aa249012eab828fb798de3ca4ed563667"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}