A sensor based indoor mobile localization and navigation using Unscented Kalman Filter

Abstract

Localization is the most important function to mobile vehicle in indoor environments. The precise positioning of the mobile object can provide higher mobility with more operation capability. The main challenge for indoor navigation is to solve higher accuracy heading and position in real time. In this paper, a low-cost MEMS hardware is designed and fabricated to focus on its accelerations and orientations by appropriate sensors. An auxiliary architecture of the Wireless Sensor Network (WSN) is added to improve the tracking accuracy in system operation. A sensor node, spacing around 10 to 20 meters, is implemented as a positioning and navigation network in the small area. The proposed system measures the radio signal strength from each node using the Unscented Kalman Filter (UKF). By this algorithm, the linearization process of a nonlinear model can be neglected. The evaluation of the Jacobians is not requested to get higher order accuracy. The more accurate estimation can reach, the better parameter tuning of the UKF is observed. The proposed algorithm incorporating with MEMS hardware has lead to some good indoor test results.