mlsanityData to 14 Oct 2026

Dual-Stream Memory for Vision-Language-Action Policies

arXiv 2610.13262 ↗First posted 9 Oct 2026cs.RO cs.CVcited by 0

On the map: Vision-language-action models → Reasoning and chain-of-thought in VLA also fits: Efficient VLA inference and adaptation

Abstract (arXiv)

Memory-dependent manipulation requires a policy to track execution state and recover visual evidence that is no longer observable. We propose dual-stream memory for vision-language-action policies, assigning these requirements to complementary representations. A task-agnostic Gated DeltaNet compresses history into one temporal token, supervised by progress prediction and action-conditioned feature reconstruction. A task-conditioned retriever retains frames above a fixed relevance threshold; linear multi-scale fusion encodes each frame into sixteen spatial tokens. We then freeze the temporal compressor and jointly optimize retrieval and policy learning. Across sixteen RoboMME tasks, the complete model achieves 49.22% mean success versus 44.51% for the published FrameSamp+Modul reference, improving fourteen tasks and all four suite means. On four real-robot tasks, mean success reaches 62.5% versus 37.5%, with the largest gains on repeated actions requiring execution-stage discrimination. These results support combining execution context with selective visual access, while failures under cup swaps expose persistent identity tracking as an unresolved requirement.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Reasoning and chain-of-thought in VLA, one of 13 directions in Vision-language-action models. Builds on 40 of its 41 references that are in the map. First posted on arXiv 9 Oct 2026.

Builds on

40 of 41 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026NativeMEM: Native Memory Compression for Long-Horizon Robotic Manipulation7 Jul
  2. 2026RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policies4 Mar
  3. 2026MEM: Multi-Scale Embodied Memory for Vision Language Action Models4 Mar
  4. 2025MemoryVLA: Perceptual-Cognitive Memory in Vision-Language-Action Models for Robotic Manipulation26 Aug

Other references · 36

  1. 2017Attention Is All You Needcited by 60,361
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  4. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  5. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  6. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,968
  7. 2022Flow Matching for Generative Modelingcited by 5,702
  8. 2016Feature Pyramid Networks for Object Detectioncited by 5,036
  9. 2023Mamba: Linear-Time Sequence Modeling with Selective State Spacescited by 4,718
  10. 2022Flamingo: a Visual Language Model for Few-Shot Learningcited by 4,386
  11. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,109
  12. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 3,080
24 more
  1. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,887
  2. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,838
  3. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,407
  4. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,061
  5. 2020Transformers are RNNs: Fast Autoregressive Transformers with Linear Attentioncited by 2,022
  6. 2019Transformer-XL: Attentive Language Models Beyond a Fixed-Length Contextcited by 1,850
  7. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,761
  8. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,568
  9. 2024Octo: An Open-Source Generalist Robot Policycited by 1,484
  10. 2020MiniLM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformerscited by 1,463
  11. 2025DINOv3cited by 1,414
  12. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,410
  13. 2024Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Dualitycited by 1,336
  14. 2021Perceiver: General Perception with Iterative Attentioncited by 1,104
  15. 2019Compressive Transformers for Long-Range Sequence Modellingcited by 726
  16. 2021CALVIN: A Benchmark for Language-Conditioned Policy Learning for Long-Horizon Robot Manipulation Taskscited by 584
  17. 2022Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulationcited by 551
  18. 2024Gated Delta Networks: Improving Mamba2 with Delta Rulecited by 532
  19. 2024Parallelizing Linear Transformers with the Delta Rule over Sequence Lengthcited by 396
  20. 2024Learning to (Learn at Test Time): RNNs with Expressive Hidden Statescited by 355
  21. 2025LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristicscited by 254
  22. 2022Recurrent Memory Transformercited by 203
  23. 2023Vision Transformers with Mixed-Resolution Tokenizationcited by 36
  24. 2026RoboTTT: Context Scaling for Robot Policiescited by 19

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026T2Mem: Learning Test-Time Memory for Robotics ↗
  2. 2026MemBodied: Recurrent Associative Memory for Vision-Language-Action Models
  3. 2026Memory Without Curation: Learning Multi-Stream Latent Memory for Robot Policies
  4. 2026EvoMem-VLA: State-Evolution Memory for Long-Horizon Robot Manipulation
  5. 2026StateMem: Single-State Residual Memory with Adaptive Inference for Vision-Language-Action Policies
  6. 2026EventVLA: Event-Driven Visual Evidence Memory for Long-Horizon Vision-Language-Action Policies
  7. 2026MemoryVAM: Integrating Memory into Video Action Model for Robot Manipulation
  8. 2026ChainVLA: Chaining Vision-Language-Action Queries through a Unified Execution State for Long-Horizon Manipulation