Event
ForceDelta-VLA Force-Correction Distillation Framework Released
1 reports · 1 sources
Ju Dong and other authors released ForceDelta-VLA, a force-conditioned action-correction distillation framework: it uses the force-conditioned and force-free predictions of a frozen teacher model to construct an explicit force-correction target, and additionally sets a delayed correction target to handle reference-action mismatch. The authors report that on nine single-arm and dual-arm contact-intensive tasks, the system achieved an average success rate of 82.2%, compared with 54.4% for the ForceVLA baseline and 70.6% for direct execution by the first-stage temporal teacher; compared with ForceVLA, the full system reduced the average peak contact force of successful trials by about 26% on two platforms.
Report timeline
- ForceDelta-VLA: Distilling Force-Conditioned Action Corrections for Contact-Rich Manipulation
ForceDelta-VLA proposes a correction-distillation framework that constructs an explicit force-correction target from the force-conditioned and force-agnostic predictions of a frozen teacher model, and adds a separate delay-correction target to handle reference-action mismatch. Across nine single-arm and bimanual contact-rich tasks, the system achieves a mean success rate of 82.2%, compared with 54.4% for the ForceVLA baseline and 70.6% for direct execution of the Stage-1 temporal teacher. Compared with ForceVLA, the full system reduces the mean peak contact force of successful trials by about 26% on both platforms.
arXiv cs.RO ·