Impact of trajectory constraints on beailc and coilc convergence rates

Abstract

Iterative learning control (ILC) is a suitable control method for industrial robot applications where they are required to execute repetitive tasks with high precision. In this paper, the impact of trajectory constraints on convergence rates of two constrained state space ILC algorithms is studied. Taking into account that in reality, robot’s operating space is limited, as well as the ILC’s transient error growth problem, the following constrained state space ILC algorithms were applied to the nonlinear 3DoF robot manipulator model: Bounded Error Algorithm (BEA) and Constrained Output Algorithm (CO). Both algorithms force the output trajectory to stay inside the predetermined boundaries defined by the safest distance from the desired trajectory and the coordinate limit, making their convergence rates closely dependent on the selection of these boundaries. Herein, tracking simulations of the desired trajectories defined in the generalized coordinates space were conducted in MATLAB and Simulink environments, with the same feedback and learning parameters applied to both algorithms but with different sets of values for state space boundaries, hypercylinder radius eps for BEA and, the maximum and the minimum values of the joints’ generalized coordinates Qimax and Qimin for CO algorithm set in the way that the simulation results are comparable. Simulation results, analysis of the constraint parameters influence on the convergence rates, and their comparisons for the previously mentioned algorithms are shown later in this paper.