Passivity-Based Adaptive Control for Visually Servoed Robotic Systems

TL;DR

This paper develops passivity-based adaptive control schemes for visually servoed robotic systems with uncertain parameters, ensuring image-space tracking errors converge to zero without depth invertibility assumptions.

Contribution

It introduces two novel passivity-based adaptive control methods for visual servoing that handle uncertainties without requiring depth invertibility.

Findings

Image-space tracking errors converge to zero

Proposed controllers handle parameter uncertainties

Numerical simulations confirm effectiveness

Abstract

This paper investigates the visual servoing problem for robotic systems with uncertain kinematic, dynamic, and camera parameters. We first present the passivity properties associated with the overall kinematics of the system, and then propose two passivity-based adaptive control schemes to resolve the visual tracking problem. One scheme employs the adaptive inverse-Jacobian-like feedback, and the other employs the adaptive transpose Jacobian feedback. With the Lyapunov analysis approach, it is shown that under either of the proposed control schemes, the image-space tracking errors converge to zero without relying on the assumption of the invertibility of the estimated depth. Numerical simulations are performed to show the tracking performance of the proposed adaptive controllers.