Implementation and comparison of tracking algorithms for differential drive robots

Abstract

Trajectory tracking is an integral component of most approaches to robot motion planning. There are a variety of techniques that have been developed over the years. However, without clear guidance in the literature, users are often left to guess what algorithm may work best for them. In this thesis, we quantitatively compare three trajectory tracking controllers for a differential drive wheeled robot: Linear Quadratic Regulator, Model Predictive Control and Sliding Mode Control. We compare the performance of these controllers through the metrics of tracking error, control energy, and rate of convergence. Through both simulations and experiments, our results show that LQR and MPC outperform SMC. In addition, SMC is difficult to tune and less robust than the other two. While LQR and MPC have similar performance. LQR is easier to implement while MPC enforces the physical constraints of the robot hardware.