Linear controller for the NXT ballbot with parameter variations using linear matrix inequalities

Abstract

Robotics, a classical but still burgeoning field buttressed by the fast innovating computer science and hardware industries revolution is still at the top area of the engineering academia. Unlike the traditional robot that has arms and wheels and is just like a mimic of human, ballbot is a robot who could achieve dynamic stability by balance itself on a ball. Its unique mechanism and performance shed some light on creative robot design and analysis. To develop a controllable stable Ballbot, modern control theory will be applied based on the dynamic force analysis of the robot system. During the past years, several ballbots has been developed by other institutes and universities (CMU (Carnegie Mellon University) in 2005, and TGU (Tohoku Gakuin University in 2008) ) . This solo project aims to develop a Ballbot using Lego NXT Mindstream toolkit and apply a controller based on the LMI analysis of the model to boost the stability and add robustness against the foreseeable parametric uncertainties to the controllers. Throughout the report, the model and methodology used by developing the robot will be discussed, the simulation and experimental results will be introduced and analyzed to discern the unique advantage of the application of LMI.