Adaptive self-tuning path control system for an autonomous mobile robot

We consider the path control problem for the autonomous mobile robot B21. In particular we define the reference paths as time indexed functions. The error between the actual position and the reference path is described by a nonlinear time varying multivariable state space model. We propose an adaptive self-tuning control system taking into account the properties of the obtained state space model. We concentrate especially on the control design block. For every iteration of the outer control loop a nonlinear transformation of the state space system into a canonical form must be carried out. Afterwards we use the pole assignment method to design a controller. We have implemented the designed control system on the real platform B21. Several experiments have been carried out where it has been demonstrated that the system requirements are fulfilled. The closed-loop system accurately follows the reference path. As expected, initial disturbances are eliminated and the reference path is caught up. Even reference paths with very high dynamics are followed accurately by the closed-loop system after the transient oscillations and no tracking error remains