Application of the extended Kalman filter to LIDAR pose estimation

The goal of this work is to investigate the benefits of using a well-known nonlinear motion estimator, an Extended Kalman Filter (EKF), in conjunction with the Iterative Closest Point algorithm (ICP}, in particular, for the purpose of tracking the pose of a target satellite using a chaser satellite equipped with a LIDAR sensor. To accomplish this goal, two different architectures for tracking the pose of a target satellite were first implemented in MATLAB Simulink, and then implemented and tested on the Canadian Space Agency Automated Robotics Test Bed (CART} at the Canadian Space Agency (CSA} using a Neptec Laser Camera System as a sensor. The two architectures are: a} a pose tracking architecture that accepts the estimated pose supplied by the EKF to provide an initial pose guess to the ICP algorithm; and b) a pose tracking architecture that uses the pose supplied by the pervious pose measurement from the ICP algorithm as the initial pose guess for the ICP algorithm. The pose estimator combine with the EKF was able to track an object with a higher rate of motion then the rate possible without a nonlinear estimator. When the EKF estimate of the target satellite's states converges, a decrease in the number of ICP iterations per sensor measurement was also observed. Furthermore, the EKF increased the robustness of the system allowing the system to continue tracking after blackout periods. The test results showed an increased level of robustness of the tracking architecture that utilizes a nonlinear estimator in conjunction with the ICP algorithm. The advantages of the use of EKF were observed both in a simulated environment and experimentation.