Adaptive homography-based visual servo for micro unmanned surface vehicles

Abstract

In this paper, a novel adaptive homography-based visual servo (AHBVS) scheme is proposed to regulate a micro unmanned surface vehicle (MUSV) to the desired pose in the presence of both unknown image depth and unmatched dynamics. By virtue of homography decomposition technique, the scaled pose errors are directly retrieved from the live and desired images which are captured by a monocular camera. On the basis of the MUSV characteristics, a completely new visual servo system is firstly derived from visual measurements including both kinematics and dynamics. Different from kinematics solutions, the dynamics-level AHBVS controllers adapting to unknown image depth and compensating unmatched dynamics are developed by incorporating backstepping technique and Lyapunov synthesis, and thereby facilitating practical implementations. Lyapunov analysis proves that the proposed AHBVS scheme renders the closed-loop visual servo system globally uniformly asymptotically stable (GUAS). Simulation results demonstrate remarkable performance on a prototype MUSV.