A Nonlinear State Estimation Framework for Field Mobile Robots

Abstract

Field mobile robots (FMRs) require an accurate navigation system to operate properly. A standard navigation system normally uses differential global navigation satellite system (DGNSS) to determine the absolute position of the mobile robot. However, DGNSS signal is prone to noise and has a relatively low sample rate. To complement the DGNSS measurement, other sensors with higher sample rate such as inertial measurement unit (IMU), and encoder can be employed. These sensors can be fused to obtain an accurate position estimation using Kalman filter. This paper presents a new approach for DGNSS/encoder integration using exogenous Kalman filter (XKF) for a car like vehicle. The XKF consists of a cascade of a nonlinear observer (NLO) and a linearized Kalman filter (LKF). The NLO provides a global stability property while the LKF handle the uncertainty. The algorithm is implemented in the FMR model that represents ground mobile robot dynamics and is compared with the extended Kalman filter (EKF). Simulation results show the XKF algorithm has better estimation errors compared to the EKF.