Unmodeled dynamics and data driven balance control for a class of underactuated mechanical systems

Unmodeled dynamics and data driven balance control strategy are presented in this paper for a class of underactuated mechanical systems with two freedoms. The idea behind the method are as follows. First, the underactuated system is divided into two subsystems. Linear models for each subsystem are constructed from the experimental datum. The proportional derivative (PD) controller can be designed by this linear model. The unmodeled dynamic compensator is designed to deal with modelling error between the linear model and the real model. Second, to control two outputs of the underactuated systems with one input at the same time, a coordinative control scheme is introduced for weighting the control inputs of the two subsystems. Finally, in order to value the proposed control scheme, the Pendubot (pendulum robot) is selected as the experimental platform to verify the method. Experimental results show that the proposed control strategy can be easily applied and has higher control precision than the existing methods.