Modelling and Control of a Skid-Steering Mobile Robot for Indoor Trajectory Tracking Applications

The present MSc thesis addresses the problem of modelling and controlling a four-wheel Skid-Steering Mobile Robot (SSMR) used for indoor trajectory tracking applications. The industries in need of a simple and robust mobile robot have long been drawn to SSMR solutions, but their underlying dynamic complexity has repeatedly hampered their true potential. As this category of robots is forced to skid in order to turn, they are characterised by complex wheel-ground interactions, which primarily exhibit as non-linear time-varying frictional forces. Such phenomena profoundly affect the behaviour of an SSMR and are responsible for unreliable dynamic models and inaccurate motion control system. This research work builds on the literature resources currently available and advances a new modelling scheme containing an innovative friction model. Two trajectory tracking control systems are then presented: a Proportional-Integral-Derivative (PID) and a Linear Quadratic Regulator (LQR) optimal control. Furthermore, this research is characterised by the novel use of Artificial Intelligence (AI), namely Differential Evolution (DE), for optimisation purposes. DE algorithms proved to be an undoubtedly effective alternative to system identification techniques generally employed for unknown physical parameters estimation. Moreover, DE was applied to get around the traditional empirical controller tuning methods and, instead, to find a global optimum to the control problem. Ultimately, the promising experimental results acquired at the Beidou Research Institute - in Shanghai, China - are presented. They validate the modelling assumptions and serve as a proof of concept for the use of DE algorithms for controllers tuning.