Outdoor Localization System with Augmented State Extended Kalman Filter and Radio-Frequency Received Signal Strength

Abstract

In this paper, we propose a mobile robot localization system based on the Received Signal Strength Indicator (RSSI) of three radio module beacons, to enhance or replace GPS positioning in scenarios with little or no GPS availability, such as urban environments with blocked sky view (near buildings or tall trees). Our approach uses an Extended Kalman Filter with augmented states to estimate a robot's trajectory by fusing sensor data from odometry, gyroscope, GPS (when available), and three RSSI from the beacons, which have its biases corrected in the Kalman filter itself. Our contribution is a low-cost localization system to be used in outdoor environments where other sensors suffer from a lack of availability and uncertainty. To test our approach, we experiment by collecting several datasets in an outdoor environment with a wheeled ground robot, where it travels in a predefined trajectory with a total length of 48.78 meters. We use these datasets to estimate the robot's trajectory offline, where our approach achieves a best mean trajectory error of 0.308 meters, an 82.58% improvement over the baseline GPS error.