Metraipolis › archive › dc_2aecb547dc93
SimpleMemVLA: A Simple but Effective Native-Video Memory for Vision-Language-Action Models — arXiv:2609.05533
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.05533
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T13:58:38.656Z
- SHA-256 of the archived text
0011ec8ff45dd03582a8ee39c3f95c2cd6057b3d7826662d69048adacfa170f7- Stable URL of this page
https://metraipolis.com/documento/dc_2aecb547dc93
How to cite
SimpleMemVLA: A Simple but Effective Native-Video Memory for Vision-Language-Action Models — arXiv:2609.05533 — arXiv (en), https://arxiv.org/abs/2609.05533 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2aecb547dc93 (Metraipolis archive dc_2aecb547dc93, sha256 0011ec8ff45dd03582a8ee39c3f95c2cd6057b3d7826662d69048adacfa170f7)
BibTeX
@misc{dc2aecb547dc93,
title = {SimpleMemVLA: A Simple but Effective Native-Video Memory for Vision-Language-Action Models — arXiv:2609.05533},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.05533},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2aecb547dc93; sha256 0011ec8ff45dd03582a8ee39c3f95c2cd6057b3d7826662d69048adacfa170f7},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.05533. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Long-horizon manipulation is partially observable: the information needed to choose the next action may appear only in observations from minutes earlier. Existing memory mechanisms for VLAs, such as retrieval banks, learned compressors and recurrent states, must decide what to keep from the past before knowing what a future decision will require. They were motivated by the assumption that minute-scale history is too large to process directly, which no longer holds for modern VLM backbones. We propose SimpleMemVLA, a VLA without a dedicated memory module that uses the backbone's native video context directly as memory. It keeps the sampled history intact in the timestamped video format the backbone was pretrained to process, routes the evidence it finds to a standard flow-matching action head through the hidden states of a generated sub-task, and prefills the history shared by consecutive decisions during action execution, keeping latency close to that of a single-frame VLA. SimpleMemVLA achieves state-of-the-art results on four memory benchmarks without loss on general-purpose control, and with the same backbone and training setup it outperforms retrieval, compression and recurrent-state methods by a wide margin. History interventions show that the policy reads specific evidence from its past and follows edited histories without parameter updates, a visual form of in-context learning. On a physical dual-arm robot, it completes two tasks whose decisive evidence disappears before the robot acts. Code available at this https URL Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-09 in cs.LG). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Machine Learning (cs.LG); Robotics (cs.RO). Pagina del lavoro: https://arxiv.org/abs/2609.05533 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-16T13:39:43.116Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).