Adaptive Odometry and IMU Sensor Fusion for KUKA youBot Mobile Robot Using Analytical Time Update

Abstract

Odometry and inertial measurement units are fundamental components of modern vehicle localization systems along with the low-sampling-rate absolute position sensors like GPS and LIDAR. This paper considers their tight fusion and adaptive filtering methods to determine their bias and deviation online. The performance of CMAUKF (Covariance Matching-based Adaptive Kalman Filter) and the WRWAUKF (Windowing and Random Weighting Adaptive Unscented Kalman Filter) is investigated via numerical simulations. Furthermore, a robot-specific exact analytical formula is utilized in the time update step instead of applying the generic unscented transformation in order to decrease the computational cost of the methods.