Optical flow based odometry for mobile robots supported by multiple sensors and sensor fusion

Abstract

This paper introduces an optical flow based odometry solution for indoor mobile robots. The indoor localization of mobile robots is an important issue according to the increasing mobile robot market and the needs of the industrial, service and consumer electronics sectors. The robot odometry calculated from the robot kinematics accumulates the position error caused by the wheel slip but an optical flow based measurement is independent from wheel slipping so both methods have different credibility which was considered during the sensor fusion and the development. The focus of the research was to design an embedded system with high accuracy on the possibly lowest price to serve the needs of the consumer electronics sector without the need of expensive camera and real-time embedded computer based high level robot localization solutions. The paper proposes the theoretical background, the implementation and the experimental results as well. The universal optical flow module can be implemented in any kind of indoor mobile robot to measure the position and the orientation of the robot during the motion, even in the case of a 3 DoF holonomic drive like kiwi drive. The application of omnidirectional wheels in mobile robotics requires high accurate position and orientation feedback methods contrary to differential drives.