ACRA Format Instructions for Submissions

This research tested different ways of tuning the process noise covariance matrix of a GPS/IMU extended Kalman filter. This Kalman filter was developed for a retrofit robot driver. Because of the constraints of this application, the Kalman filter had no process model for the outputs of the robot’s controller. In this situation, a significant proportion of the process noise would be caused by the robot’s planned manoeuvres. Since the robot driver controls these manoeuvres in a somewhat predictable manner, the optimisation of the process noise covariances was carried out offline. Evaluating the performance of a given process noise covariance matrix was achieved by comparing the true values of states to the state estimates gained while running the Kalman filter with simulated GPS and IMU data. The process noise covariance matrix was optimised using a genetic algorithm. Three hypotheses were tested using this optimisation procedure and all were found to be true. The hypotheses were that multiobjective optimisation should be used, that the vehicle manoeuvres were the key source of process noise for the robot driver and that different manoeuvres require different process noise covariance matrices. The proposed tuning procedure for a robot driver’s GPS/IMU Kalman filter was successfully field tested.