Obstacle Detection and Recognition Using Stereo Vision and Radar Data Alignment for USV

Abstract

Environment perception technology is the first step and key technology for autonomous navigation of unmanned surface vehicle (USV) under the condition of over-the-horizon. The detection and location of marine obstacle targets are an important research content of environmental perception technology. This paper proposes a method for aligning the information of the stereo vision sensor and the millimeter-wave radar sensor without united calibration to detect and recognize the marine obstacle targets. This method extracts the position and velocity of the target as a feature vector and designs a matching evaluation function to find the best matching pair from two sensors’ data information. Finally, the article verifies the real-time and accuracy of the method by experiments.