Cone Detection and Location for Formula Student Driverless Race

This paper proposes a cone detection algorithm based on HSV color and the least-squares method. The algorithm first converts the RGB image of the cone into HSV format and then binarizes it. Next, filter the binarized image by methods of floodfill and morphology. The contour of the cone is detected, and then the slope is determined by the least-squares method for the final determination. This article also describes the principle of camera calibration and the perspective transformation method to extract the position of the identified cone. Finally, it is verified by experiments that cone recognition is effective and the location information is highly reliable.