Model Predictive Control Using Dynamic Model Decomposition Applied to Two-Wheeled Inverted Pendulum Mobile Robot

Abstract

In this paper, we discuss the locomotion control for the two-wheeled inverted pendulum (TWIP) mobile robot. The robot in consideration involves two independent driving wheels sharing the same axle as well as one inverted pendulum in the middle acting as the main body. Instead of considering the entire TWIP mobile robot as a whole, following the idea of Dynamic Model Decomposition, we decompose the robot into the body and the two wheels, with interaction forces and moments connecting them. The effect is that we can thus enjoy lower-dimensional dynamics for each subsystem while their composition maintaining the equivalence to the full-order robot model. Based on that, we further propose a corresponding model predictive control framework via quadratic programming, which considers linearly approximated body dynamics with constrained wheel reaction forces as inputs. The overall methodology was successfully implemented on a TWIP mobile robot in the simulation environment. The simulation results show that the robot is capable of station keeping, disturbance rejection, velocity tracking, and path following.