Paper Detail

MambaAct: Efficient Mamba Action Experts for Flow-Matching Vision-Language-Action Models

Khuyen Pham, Tran Nguyen Le, Edoardo Fazzari, Cesare Stefanini

huggingface Score 6.5

Published 2026-10-16 · First seen 2026-10-23

General AI

Abstract

Vision-language-action (VLA) models have shown strong promise for generalist robotic manipulation, but their inference cost remains a major bottleneck for real-time control. This limitation is especially pronounced in flow-matching VLAs such as π_{0.5}, where the action expert is evaluated repeatedly during iterative denoising. We introduce MambaAct, an efficient Mamba action expert for flow-matching VLA policies. MambaAct retains the pretrained visual-language prefix encoder of π_{0.5}, while replacing its transformer-based action expert with a compact Mamba-based action expert. Within this expert, Mamba serves as the action mixer that models dependencies across the noisy continuous-action sequence. To preserve visual-language conditioning without repeated token-level cross-attention, MambaAct uses a lightweight Context Compressor to map Vision-language model (VLM) prefix tokens into a compact conditioning signal that gates and modulates each Mamba action block. MambaAct uses a 10times smaller action expert and reduces flow-matching denoising latency by 64.8% on LIBERO and 58.8% in real-world deployment, while improving mean LIBERO success by 1.08 percentage points across three independent fine-tuning seeds and maintaining gains on LIBERO-Plus and real-world benchmarks. Our code and model weights are publicly available at https://github.com/giakhuyendihoc/MambaAct.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
soon
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@misc{pham2026mambaact,
  title = {MambaAct: Efficient Mamba Action Experts for Flow-Matching Vision-Language-Action Models},
  author = {Khuyen Pham and Tran Nguyen Le and Edoardo Fazzari and Cesare Stefanini},
  year = {2026},
  abstract = {Vision-language-action (VLA) models have shown strong promise for generalist robotic manipulation, but their inference cost remains a major bottleneck for real-time control. This limitation is especially pronounced in flow-matching VLAs such as π\_\{0.5\}, where the action expert is evaluated repeatedly during iterative denoising. We introduce MambaAct, an efficient Mamba action expert for flow-matching VLA policies. MambaAct retains the pretrained visual-language prefix encoder of π\_\{0.5\}, while r},
  url = {https://huggingface.co/papers/2610.22951},
  keywords = {vision-language-action models, flow-matching, Mamba, action expert, cross-attention, context compressor, denoising latency, code available, huggingface daily},
  eprint = {2610.22951},
  archiveprefix = {arXiv},
}

Metadata

{}