analiza aiqbench
This research introduces ForceDelta-VLA, a distillation framework that decouples task-level motion from contact-dependent adjustments in Vision-Language-Action policies. By leveraging paired predictions from frozen teacher models, the method enables lightweight correction policies that significantly improve success rates and reduce peak contact forces in complex manipulation tasks. This advancement addresses a key bottleneck in deploying robust AI agents for physical interaction where precise force control is critical.
ForceDelta-VLA: Distilling Force-Conditioned ActionCorrections for Contact-Rich Manipulation
arXiv:2609.18242v1 Announce Type: new Abstract: Force-aware Vision-Language-Action (VLA) policies improve contact-rich manipulation, but typically combine task-level motion and contact-dependent adjustment in a single action prediction. Demonstrations provide no explicit labels for decomposing that prediction into a reusable reference action and a correction. We present ForceDelta-VLA, a correction-distillation framework that constructs an explicit force-correction target using paired predictions from a frozen teacher's force-conditioned and learned force-agnostic modes. A separate delay-correction target accounts for reference-action mismatch and the change in reference state. Training uses asynchronous schedule replay with the cached task context available during execution. The resulting lightweight policy adjusts the reference actions using recent force history and robot state, responding to contact changes between reference-action updates without regenerating complete action chunks. Across nine single-arm and bimanual contact-rich tasks, ForceDelta-VLA achieves an 82.2% mean success rate, compared with 54.4% for the original ForceVLA baseline. Direct execution of our Stage-1 Temporal Teacher achieves 70.6%. Relative to ForceVLA, the complete system reduces mean peak contact force over successful trials by approximately 26% on both platforms.
Authors:Ju Dong, Yu Fu, Jian Chen, Yimeng Liu, Haocheng Zhao, Lei Zhang, Kaixin Bai, Liding Zhang, Diwen Zheng, Alois Christian Knoll, Angela P. Schoellig, Jianwei Zhang
Abstract:Force-aware Vision-Language-Action (VLA) policies improve contact-rich manipulation, but typically combine task-level motion and contact-dependent adjustment in a single action prediction. Demonstrations provide no explicit labels for decomposing that prediction into a reusable reference action and a correction. We present ForceDelta-VLA, a correction-distillation framework that constructs an explicit force-correction target using paired predictions from a frozen teacher's force-conditioned and learned force-agnostic modes. A separate delay-correction target accounts for reference-action mismatch and the change in reference state. Training uses asynchronous schedule replay with the cached task context available during execution. The resulting lightweight policy adjusts the reference actions using recent force history and robot state, responding to contact changes between reference-action updates without regenerating complete action chunks. Across nine single-arm and bimanual contact-rich tasks, ForceDelta-VLA achieves an 82.2% mean success rate, compared with 54.4% for the original ForceVLA baseline. Direct execution of our Stage-1 Temporal Teacher achieves 70.6%. Relative to ForceVLA, the complete system reduces mean peak contact force over successful trials by approximately 26% on both platforms.
Robotics (cs.RO)
arXiv:2609.18242 [cs.RO]
(or arXiv:2609.18242v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.18242 arXiv-issued DOI via DataCite (pending registration)
Text integral sub licența deschisă a publicației · 209 cuvinte
Subiecte conexe
HapticWAM: Distilling Imagined Touch into a World-Action Model without Inference-Time Tactile Sensing
arXiv · 22 sept. 2026
Beyond Dexterity: Why Contact May Define the Next Era of Robotics
IEEE Spectrum · 9 iun. 2026
TAO-Force: Unifying Force-Aware Perception and Fast-Slow Control for Contact-Rich Manipulation
arXiv · 17 sept. 2026
TacForeSight: Force-Guided Tactile World Model for Contact-Rich Manipulation
arXiv · 10 iun. 2026