analiza aiqbench
This research demonstrates how Vision-Language-Action models can achieve high-performance, safe manipulation in contact-rich tasks without requiring on-board tactile sensors during inference. By distilling safety-aware reasoning into a compact token learned offline, HapticVLA outperforms baselines that rely directly on haptic feedback while lowering hardware costs and improving platform reproducibility. The approach effectively decouples complex tactile learning from deployment constraints, advancing the practicality of autonomous robotics through advanced machine learning techniques rather than new sensor technology.
HapticVLA: Contact-Rich Manipulation via Vision-Language-Action Model without Inference-Time Tactile Sensing
arXiv:2603.15257v2 Announce Type: replace Abstract: Tactile sensing is a crucial capability for Vision-Language-Action (VLA) architectures, as it enables dexterous and safe manipulation in contact-rich tasks. However, reliance on dedicated tactile hardware increases cost and reduces reproducibility across robotic platforms. We argue that tactile-aware manipulation can be learned offline and deployed without direct haptic feedback at inference. To this end, we present HapticVLA, which proceeds in two tightly coupled stages: Safety-Aware Reward-Weighted Flow Matching (SA-RWFM) and Tactile Distillation (TD). SA-RWFM trains a flow-matching action expert that incorporates precomputed, safety-aware tactile rewards penalizing excessive grasping force and suboptimal grasping trajectories. TD further transfers this tactile-aware capability into a conventional VLA: we distill a compact tactile token from the SA-RWFM teacher and train a student VLA to predict that token from vision and state modalities, enabling tactile-aware action generation at inference without requiring on-board tactile sensors. This design preserves contact-rich tactile-aware reasoning within VLA while removing the need for on-board tactile sensors during deployment. On real-world experiments, HapticVLA achieves a mean success rate of 86.7%, consistently outperforming baseline VLAs - including versions provided with direct tactile feedback during inference.
Abstract:Tactile sensing is a crucial capability for Vision-Language-Action (VLA) architectures, as it enables dexterous and safe manipulation in contact-rich tasks. However, reliance on dedicated tactile hardware increases cost and reduces reproducibility across robotic platforms. We argue that tactile-aware manipulation can be learned offline and deployed without direct haptic feedback at inference. To this end, we present HapticVLA, which proceeds in two tightly coupled stages: Safety-Aware Reward-Weighted Flow Matching (SA-RWFM) and Tactile Distillation (TD). SA-RWFM trains a flow-matching action expert that incorporates precomputed, safety-aware tactile rewards penalizing excessive grasping force and suboptimal grasping trajectories. TD further transfers this tactile-aware capability into a conventional VLA: we distill a compact tactile token from the SA-RWFM teacher and train a student VLA to predict that token from vision and state modalities, enabling tactile-aware action generation at inference without requiring on-board tactile sensors. This design preserves contact-rich tactile-aware reasoning within VLA while removing the need for on-board tactile sensors during deployment. On real-world experiments, HapticVLA achieves a mean success rate of 86.7%, consistently outperforming baseline VLAs - including versions provided with direct tactile feedback during inference.
Robotics (cs.RO)
arXiv:2603.15257 [cs.RO]
(or arXiv:2603.15257v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2603.15257 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.
Text integral sub licența deschisă a publicației · 272 cuvinte
Subiecte conexe
ForeTac-VLA: A Forecasting-Based Tactile-Vision-Language-Action Model for Contact-Rich Robotic Manipulation
arXiv · 21 sept. 2026
VT-Bridge: Bridging Pretrained Foundation VLAs to VTLAs via Lightweight Residual Adaptation
arXiv · 22 sept. 2026
HapTile: A Haptic-Informed Vision-Tactile-Language-Action Dataset for Contact-Rich Imitation Learning
arXiv · 4 iun. 2026
DeCAL: Towards Physically-Grounded Dexterous Vision-Language-Action Models via Contact-Aware Latent Co-Imagination
arXiv · 9 sept. 2026