LMI-based trajectory planning for closed-loop control of robotic systems with visual feedback

Abstract

Closed-loop robot control based on visual feedback is an important research area, with useful applications in various fields. Planning the trajectory to be followed by the robot allows one to take into account multiple constraints during the motion, such as limited field of view of the camera and limited workspace of the robot. This paper proposes a strategy for path-planning from an estimate of the point correspondences between the initial view and the desired one, and an estimate of the camera intrinsic parameters. This strategy consists of generating a parametrization of the trajectories connecting the initial location to the desired one via polynomials. The trajectory constraints are then imposed by using suitable relaxations and LMIs (linear matrix inequalities). Some examples illustrate the proposed approach.