Robust monocular visual odometry using optical flows for mobile robots

Abstract

A robust optical flow-based visual odometry method using a single onboard camera is proposed in this paper. To improve the quality of the noisy optical flows, a correction method across multiple frames is developed. Furthermore, the optical flows in the plane at infinity are detected and removed as these optical flows have very low signal to noise ratio for robot translation estimation. Finally, a RANSAC approach for robot ego-motion estimation is proposed. Physical experiments are carried out and the results show that the proposed method is able to estimate the camera trajectory robustly with reasonable accuracy.