A performance criterion for the depth estimation of a robot visual control system

Abstract

We propose a performance criterion for the depth estimation of an active vision system. It is well-known that the linear velocity of the camera must satisfy some constraints for a success of the depth estimation. However, these constraints cannot ensure a good convergence performance of the depth estimation. Our criterion is an extension from an estimation result of a linear system. Furthermore, we use it to develop a visual servo control scheme that has good performance in both depth estimation and visual control. Finally, experiments and simulation results of a visual application verify our proposition.