Real-time control of a wheeled inverted pendulum based on an intelligent model free controller

This work applies a novel model free intelligent controller to control wheeled inverted pendulums (WIPs). The main purpose of this work is to develop an adaptive output recurrent cerebellar-model-articulation-controller (AORCMAC) for angle and position control of such a WIP without model information. Because the novel model free AORCMAC architecture captures system dynamics, it has a superior capability to the conventional cerebellar-model-articulation-controllers in terms of its efficient learning and dynamic response. Moreover, a virtual angle is applied to complete the angle and position control of the WIP simultaneously with the model information unknown. The experimental results indicate that the WIPs can stand upright and move forward and backward stably when using the proposed control scheme.