Aiming at the problem of accurate real-time positioning of mobile robot

visual sensor Kinect is used as information acquisition source

a visual odometer method which fuses optical flow direct method and feature point matching method for Kalman fusion is proposed. Optical flow direct method is used to estimate the small scale motion between frames. at the same time

the feature point matching method is used to estimate the motion between key frames

and correct the localization error brought by the optical flow method. The results show that the fusion algorithm can overcome poor precision

error accumulation of optical flow method and low real-time processing speed of feature point method

highlight the high precision advantages of feature point matching and eliminate the accumulation of error with local map. This fusion method can provide more accurate real-time localization information

it has a certain robustness.