Event-based Data Scheduling for Invariant Extended Kalman Filtering

We investigate the communication data scheduling problem for remote estimation by using the invariant extended Kalman filter (InEKF) for robotic localization with a Lidar sensor, where the robot needs to consider the limited communication energy and bandwidth to decide whether to send the measurement to a terminal end. To address this problem, we first propose an event-based scheduling scheme for InEKF and derive the corresponding exact MMSE estimator. We also derive an approximate MMSE estimator and evaluate the communication frequency in a closed form. In numerical experiments, we illustrate that the proposed scheme can strike a balance between the communication usage and the estimation quality by choosing an appropriate event-triggering threshold.