aiqbench analysis
MambaAct addresses the high inference latency of flow-matching vision-language-action models by substituting a transformer-based action expert with a more efficient Mamba architecture. This modification reduces denoising latency by approximately 65% while maintaining or slightly improving success rates on standard robotic manipulation benchmarks like LIBERO. The approach effectively demonstrates how state-space models can optimize real-time control without sacrificing the conditioning capabilities of pretrained visual-language prefixes.
MambaAct: Efficient Mamba Action Experts for Flow-Matching Vision-Language-Action Models
arXiv:2610.22951v1 Announce Type: new 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 $\pi_{0.5}$, 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 $\pi_{0.5}$, 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 $10\times$ 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.
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 $\pi_{0.5}$, 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 $\pi_{0.5}$, 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 $10\times$ 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 this https URL.
Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2610.22951 [cs.RO]
(or arXiv:2610.22951v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2610.22951 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 281 words
Related coverage
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · Oct 13, 2026
Q-VGM: Q-Guided Value-Gradient Matching for Flow-Matching VLA Policies
arXiv · Jun 9, 2026
Mamba-based Selective State Space Modeling Improves the Accuracy-Complexity Tradeoff of SmolVLA Vision-Language-Action Experts
arXiv · Aug 25, 2026
[AINews] Meta Muse Wonder & Wonder Full: MSL’s Claude 5.5-tier Frontier Model
Latent Space · Oct 13, 2026