A sensor fusion methodology for obstacle avoidance robot

Obstacle detection and navigation of dynamic environments is a challenge in mobile robotics. To address this challenge, this paper presents an efficient sensor fusion methodology to detect the size and location of obstacles and navigate the mobile robot with high accuracy. This is done by leveraging upon the unique advantages of accuracy in both ultrasonic sensor and a Kinect sensor for near-field and far-fields respectively. Further, an efficient Kalman filter is implemented to reduce the systematic errors in encoder data to track robot pose of the robot in real-time and reach the destination with high accuracy. Implemented on differential drive-based mobile robot, the proposed system has been validated with a high efficiency of detecting obstacles and reaching the destination with an accuracy of 5cm.