Extended Kalman filter with manifold state representation for navigating a maneuverable melting probe
Joachim Clemens, Kerstin Schill · International Conference on Information Fusion · 2016
We present a multi-sensor fusion approach based on an extended Kalman filter for estimating position and attitude in 3-dimensional space. The corresponding state is represented by a manifold that allows for a robust and mathematically consistent estimation of the state and uncertainty. The filter is used for navigating a maneuverable melting probe through deep ice, which is designed for in-situ sample analysis missions to icy bodies of our solar system. It is equipped with a variety of sensors for inertial navigation and absolute positioning that account for the fact that no global navigation satellite system (GNSS) can be used. The probe and navigation technologies were successfully tested on European and Antarctic glaciers. We demonstrate the performance of our algorithm and compare it to a particle filter approach in different simulated and real-world scenarios.