Sensor Data Fusion using Unscented Kalman Filter for VOR-based Vision Tracking System for Mobile Robots

This paper presents sensor data fusion using Unscented Kalman Filter (UKF) to implement high performance vestibulo-ocular reflex (VOR) based vision tracking system for mobile robots. Information from various sensors is required to be integrated using an efficient sensor fusion algorithm to achieve a continuous and robust vision tracking system. We use data from low cost accelerometer, gyroscope, and encoders to calculate robot motion information. The Unscented Kalman Filter is used as an efficient sensor fusion algorithm. The UKF is an advanced filtering technique which outperforms widely used Extended Kalman Filter (EKF) in many applications. The system is able to compensate for the slip errors by switching between two different UKF models built for slip and no-slip cases. Since the accelerometer error accumulates with time because of the double integration, the system uses accelerometer data only for the slip case UKF model. Using sensor fusion by UKF, the position and orientation of the robot is estimated and is used to rotate the camera mounted on top of the robot towards a fixed target. This concept is derived from the vestibule-ocular reflex (VOR) of the human eye. The experimental results show that the system is able to track the fixed target in various robot motion scenarios including the scenario when an intentional slip is generated during robot navigation.