Enabling Dynamic Tracking in Vision-Language-Action Models via Time-Discrete and Time-Continuous Velocity Feedforward

TL;DR

This paper introduces methods to incorporate velocity feedforward terms into vision-language-action models for robot manipulation, improving responsiveness and safety in contact-rich tasks by bridging the gap between compliance and precision.

Contribution

It proposes two model-agnostic approaches—finite-difference and B-Spline—for integrating velocity targets into VLA policies, enhancing control performance without extensive architecture changes.

Findings

Finite difference approach improves task speed.

B-Spline method maintains high success rates.

Both methods are compatible with standard architectures.

Abstract

While vision-language-action (VLA) models have shown great promise for robot manipulation, their deployment on rigid industrial robots remains challenging due to the inherent trade-off between compliance and responsiveness. Standard Behavior Cloning (BC) approaches predict discrete poses at low frequencies, omitting the velocity and acceleration feedforward terms typically used by low-level compliant controllers. This requires to rely on high stiffness for accurate tracking, thereby sacrificing safe contact dynamics. In this paper, we demonstrate the importance of integrating velocity feedforward terms into VLA policies to resolve this trade-off. We propose two methods for extracting velocity targets from VLAs: a time-discrete finite-difference approximation that serves as a highly effective bridge for existing models, and a continuous Cubic B-Spline action space that natively yields $C^2$ continuous trajectories for high-frequency control. Crucially, both approaches are strictly model-agnostic and compatible with any standard action-chunking architecture, requiring modifications only to teleoperation, data processing, and the low-level controller. We fine-tune the $\pi_{0.5}$ model and evaluate both of our approaches on a demanding, contact-rich cube-in-hole task. Our results indicate that incorporating the velocity feedforward term via finite differences significantly improves task execution speed, while the continuous B-Spline approach maintains high overall success rates and provides a foundation for smoother higher-order derivatives without compromising compliance.