Pose estimation for mobile robots attracts a lot of attention in recent years. In order to remove process and measurement noise, a number of filtering approaches are available to use: the extended Kalman filter (EKF), the unscented Kalman filter (UKF), and several variants of the particle filter (PF). This thesis quantitatively explores and compares the performance of the different filtering techniques applied to mobile robot pose estimation. The main criteria compared are the magnitude of the error of pose estimation, the computational complexity, and the robustness of each filter to non-linear/non-Gaussian noise. All filters are applied on both an experimental environment of a differential wheeled robot and a simulated environment of a three-wheeled robot. The simulation and experimental results indicate that the bootstrap particle filter has the best state estimation accuracy and the most computational cost. The UKF performs better than the EKF and they both have much less computational cost than the particle filter.