Bridging the Morphology Gap: Adapting VLA Models to Dexterous Manipulation via Intent-Conditioned Fine-Tuning
Bridging the Morphology Gap: Adapting VLA Models to Dexterous Manipulation via Intent-Conditioned Fine-Tuning
This research introduces InDex, a framework that adapts pretrained vision-language-action (VLA) models for dexterous manipulation by explicitly conditioning on grasp intent.
The study presents InDex, a framework for adapting VLA models to dexterous manipulation via intent-conditioned fine-tuning, reporting task success rates in simulation and on held-out objects without recalibration.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Environment
- SimulationOpen source S4
Reported numbers
number of simulated tasks
4
View original evidence
control with low-rate co-adaptation. Across four simulated tasks, InDex with $\pi_{0.5}$ achieves 85.8% mean task success, improving the native policy by 35.5 percentage points, and reaches 69.0% on held-out objects with similar geometry without fine-tuning or recalibration. Improvements also hold across three VLA backbones and separately trained real-world policies, where mean success increases from 71.0% to 83.0%.
Open source S4number of VLA backbones
3
View original evidence
control with low-rate co-adaptation. Across four simulated tasks, InDex with $\pi_{0.5}$ achieves 85.8% mean task success, improving the native policy by 35.5 percentage points, and reaches 69.0% on held-out objects with similar geometry without fine-tuning or recalibration. Improvements also hold across three VLA backbones and separately trained real-world policies, where mean success increases from 71.0% to 83.0%.
Open source S4
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
control with low-rate co-adaptation. Across four simulated tasks, InDex with $\pi_{0.5}$ achieves 85.8% mean task success, improving the native policy by 35.5 percentage points, and reaches 69.0% on held-out objects with similar geometry without fine-tuning or recalibration. Improvements also hold across three VLA backbones and separately trained real-world policies, where mean success increases from 71.0% to 83.0%.
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org