Mobile Robot Vision Tracking System Using Dead Reckoning & Active Beacons

Abstract This paper presents a new vision tracking system for mobile robot by integrating information received from encoders, inertial sensors, and active beacons. The proposed system accurately determines mobile robot position and orientation using relative and absolute position estimates, and rotates the camera towards the target during locomotion. Among the implemented sensors, the encoder data give relatively accurate robot motion information except when wheels slip. On the other hand inertial sensors have the problem of integration of noisy data, while active beacons are slow when compared to other sensors. The designed system compensates the sensors limitations and slip error by switching between two Kalman filters, built for slip and no-slip cases. Each Kalman filter uses different sensors combination and estimates robot motion respectively. The slip detector is used to detect the slip condition by comparing the data from the accelerometer and encoder to select the either Kalman filter as the output of the system. Based on the proposed sensor fusion method, a vision tracking system is implemented on a two-wheeled robot. The experimental results depict that proposed system is able to locate robot position with significantly reduced position errors and successful tracking of the target for various environments and robot motion scenarios.