MOUSE TYPE BALLBOT IDENTIFICATION AND CONTROL USING A CONVEX-CONCAVE OPTIMIZATION

This paper shows how to identify and control a mouse type
Ballbot. The Ballbot is an unstable complex system. Using
the first principle law to model the robot is not accurate enough.
An identification method is proposed. The model is transformed
to be MIMO state space model and using a convexconcave
optimization to design a robust PI+phase-lead controller.
Experiments on a design Ballbot are presented. It turns
out that an identification model of Bollbot and a PI+phase-lead
controller design framework is suitable for using to control a
Ballbot.