| Vehicle tracking is an important component of autonomy in the robotics field, requiring integration of hardware and software, and the application of advanced algorithms. Sensors are often plagued with noise and require filtering. Additionally, no single sensor is sufficient for effective tracking. Data from multiple sensors is needed in order to perform effective tracking. The Kalman Filter provides a convenient and efficient solution for filtering and fusing sensor data as well as estimating noise error covariances. Consequently, it has been essential in tracking algorithms since its introduction in 1960.;This thesis presents an application of the Kalman filter to tracking of a custom four-wheel-drive four-wheel-steering vehicle using a limited sensor suite. Sensor selection is discussed, along with the characteristics of the sensor noise as related to meeting the requirements of the Kalman filter for guaranteeing optimality. The filter requires the development of a dynamical model, which is derived using empirical data methods and evaluated. Tracking results are presented and compared to unfiltered data. |
