aiqbench analysis
Assistron introduces a shared autonomy framework that integrates off-the-shelf Vision-Language-Action models to reduce human effort in robotic manipulation without requiring fine-tuning. By leveraging flow matching guidance and phase-aware interaction detection, the system autonomously handles macro-movements while soliciting human intervention only at critical failure points. This approach preserves broad behavioral priors against catastrophic forgetting, offering a scalable paradigm for assistive robotics that bridges the gap between pure autonomy and teleoperation.
Assistron: Bayesian Shared Autonomy with Off-the-shelf Vision-Language-Action Models
arXiv:2606.23147v2 Announce Type: replace Abstract: We propose Assistron, a shared autonomy model that leverages Vision-Language-Action (VLA) models to assist the user in daily activities. Our approach is grounded in two core principles: (1)~minimizing human cognitive and physical effort by leveraging VLA-driven autonomy for macro-movements, and (2)~prioritizing human intervention specifically at critical failure points. Driven by the user's verbal language commands, Assistron utilizes the VLA to autonomously execute macro-reaching trajectories, saving users' effort. In contact-rich interactions where VLAs tend to fail, Assistron employs a phase-aware interaction detection mechanism and solicits the user to intervene, in turn adjusting the VLA's action generation via flow matching guidance. Critically, our formulation eliminates the need for VLA fine-tuning, protecting its broad behavioral priors from catastrophic forgetting and ensuring the model does not become a narrow specialist. We validate our approach on a comprehensive multi-task scene recovery benchmark encompassing diverse daily manipulation skills. Empirical results demonstrate that Assistron significantly improves task success rates over pure autonomous baselines while significantly reducing human cognitive and physical workload compared to traditional teleoperation, offering a scalable, smooth, and effortless paradigm for assistive manipulation. The code is available in https://github.com/mousecpn/Assistron.git.
Abstract:We propose Assistron, a shared autonomy model that leverages Vision-Language-Action (VLA) models to assist the user in daily activities. Our approach is grounded in two core principles: (1)~minimizing human cognitive and physical effort by leveraging VLA-driven autonomy for macro-movements, and (2)~prioritizing human intervention specifically at critical failure points. Driven by the user's verbal language commands, Assistron utilizes the VLA to autonomously execute macro-reaching trajectories, saving users' effort. In contact-rich interactions where VLAs tend to fail, Assistron employs a phase-aware interaction detection mechanism and solicits the user to intervene, in turn adjusting the VLA's action generation via flow matching guidance. Critically, our formulation eliminates the need for VLA fine-tuning, protecting its broad behavioral priors from catastrophic forgetting and ensuring the model does not become a narrow specialist. We validate our approach on a comprehensive multi-task scene recovery benchmark encompassing diverse daily manipulation skills. Empirical results demonstrate that Assistron significantly improves task success rates over pure autonomous baselines while significantly reducing human cognitive and physical workload compared to traditional teleoperation, offering a scalable, smooth, and effortless paradigm for assistive manipulation. The code is available in this https URL.
Robotics (cs.RO)
arXiv:2606.23147 [cs.RO]
(or arXiv:2606.23147v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.23147 Focus to learn more arXiv-issued DOI via DataCite
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 · 277 words
Related coverage
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · Oct 13, 2026
Vertical Advantage: Transforming Industries with Lakebase and Agentic AI
Databricks · Aug 27, 2026
Capability-Aware Arbitration for Semantic Intent-Based Shared Control
arXiv · Sep 24, 2026
AnyGoal: Vision-Language Guided Multi-Agent Exploration for Training-Free Lifelong Navigation
arXiv · Jun 16, 2026