Adaptive control of a parallel robot via backstepping technique

Abstract

Parallel robots have attracted more and more attention in recent years due to their kinematical and mechanical advantages. However the complicated high nonlinear model with unknown parameters and singularities make the control of a parallel robot much more difficult than a serial robot. Nonlinear control has been made great progress since backstepping technique was developed. Backstepping technique is a recursive design procedure and feasible for lower triangular nonlinear systems. Moreover, the adaptive backstepping is able to handle nonlinear systems with unknown parameters, which turns out to be a suitable control design methodology for parallel robots. The adaptive backstepping technique is applied to set point and tracking control of a planar parallel robot in this thesis. The dynamic model of the robot is characterized by a set of differential algebraic equations (DAEs) and further reduced to a set o f ordinary differential equations (ODEs). The inverse kinematics is also under investigation. For set point control, a model-based adaptive controller is designed based on backstepping technique, and an adaptive PD controller is also constructed for comparison. For tracking control, adaptive backstepping controller is designed based on the model with unknown parameters. The adaptive PD controller is also implemented for comparison. The performances o f the controllers are tested by experiments. Desired trajectories such as circle, line, and square are tracked in experiments for two cases: with no load and with load at the end effector. It is shown that adaptive controllers can achieve less steady state errors in set point control, and smaller tracking errors in tracking control than non-adaptive controllers, especially when there is a load attached to the end effector.