A FLAT MODEL PREDICTIVE CONTROLLER FOR TRAJECTORY TRACKING IN IMAGE BASED VISUAL SERVOING

Abstract Image-Based Visual Servoing (IBVS) is a control strategy using visual information to control the motion of robotic systems. Classical IBVS can not take into account either the mechanical constraints (joint and actuator limitations) or the visibility constraints, very important in this scheme. Model Predictive Control (MPC) is well adapted to deal with these drawbacks. However, applied to fast systems (e.g. mobile robots), the computational time is a great challenge for real time applications. One way to reduce this time is to use the concept of differential flatness. In this paper, a new IBVS strategy based on a flat MPC approach is proposed. The capabilities of this approach in terms of trajectory tracking and obstacle avoidance are pointed out. Applied to mobile robot trajectory tracking, a simulation experiment shows the efficiency and the robustness of this new control scheme. The computational time required by the proposed solution is compared with the nonlinear solution and easily enables a real-time application.