aiqbench analysis
This research introduces LexiconVLA, a framework that enables Vision-language-action models to reuse atomic action codebooks across unseen tasks without requiring skill-specific experts or deployment-time updates. By grounding reusable patterns in visual motion and effects, the approach significantly boosts success rates on held-out tasks while maintaining performance on known ones. The method demonstrates practical viability through real-robot experiments showing improved stepwise execution and failure recovery capabilities.
LexiconVLA: Learning Reusable Atomic Action Codebooks for Unseen Tasks
arXiv:2609.36774v1 Announce Type: new Abstract: Vision-language-action (VLA) models struggle to reuse recurring interactions in unseen tasks. Our diagnostic study reveals that reliable task completion does not imply consistent execution of constituent atomic actions across task contexts. We present LexiconVLA, a retrievable atomic-action lexicon for cross-task reuse. Global and detail codebooks capture shared interaction structure and fine-grained execution variation, respectively, preserving both reusable patterns and execution details. Visual-Atomic Action Alignment couples trajectory reconstruction from visual state changes with visual outcome prediction from action codes, grounding the lexicon in motion and its effects. We learn these codebooks with trajectory reconstruction and visual alignment on our AtomAction Dataset of 57,803 segments from 69 tasks. A planner and scene-aware adapter translate new goals into code-conditioned subtasks for a shared policy, without skill-specific experts or deployment-time parameter updates. Across five policy backbones on 26 RLBench tasks, LexiconVLA largely maintains performance on 18 seen tasks while improving success on 8 tasks held out from policy training. With BridgeVLA, unseen-task success rises from 16.67% to 34.17% (+17.50 percentage points), and overall success reaches 71.08%, the highest among methods with reported results. Real-robot experiments demonstrate stepwise execution and failure recovery.
Abstract:Vision-language-action (VLA) models struggle to reuse recurring interactions in unseen tasks. Our diagnostic study reveals that reliable task completion does not imply consistent execution of constituent atomic actions across task contexts. We present LexiconVLA, a retrievable atomic-action lexicon for cross-task reuse. Global and detail codebooks capture shared interaction structure and fine-grained execution variation, respectively, preserving both reusable patterns and execution details. Visual-Atomic Action Alignment couples trajectory reconstruction from visual state changes with visual outcome prediction from action codes, grounding the lexicon in motion and its effects. We learn these codebooks with trajectory reconstruction and visual alignment on our AtomAction Dataset of 57,803 segments from 69 tasks. A planner and scene-aware adapter translate new goals into code-conditioned subtasks for a shared policy, without skill-specific experts or deployment-time parameter updates. Across five policy backbones on 26 RLBench tasks, LexiconVLA largely maintains performance on 18 seen tasks while improving success on 8 tasks held out from policy training. With BridgeVLA, unseen-task success rises from 16.67% to 34.17% (+17.50 percentage points), and overall success reaches 71.08%, the highest among methods with reported results. Real-robot experiments demonstrate stepwise execution and failure recovery.
Robotics (cs.RO)
arXiv:2609.36774 [cs.RO]
(or arXiv:2609.36774v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.36774 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 · 280 words
Related coverage
Dual-Process Atomic Skill Learning: Decoupling Semantic Reasoning and Real-Time Control
arXiv · Sep 14, 2026
SkillMemo: Expert-guided Skill Memory Framework for Compositional Embodied Manipulation
arXiv · Aug 7, 2026
Learning Reusable Hybrid Motion Priors for Humanoid Locomotion from Motion Imitation
arXiv · Jul 28, 2026
MoMo: Dial Motion Mode in Robot Manipulation with Spatiotemporal Action Tokenization
arXiv · Jul 30, 2026