Road edge detection based on improved RANSAC and 2D LIDAR Data

For the road edge detection under the structured and semi-structured road environment, we model the road edge using a linear regression and propose an improved RANSAC algorithm to finish the model selection, and it is applied to the fitting and filtering of road edge points extracted from the 2D LIDAR (light detecting and ranging) data. First, road edge points are extracted using the discrete Kalman filter from each frame 2D LIDAR data. Then, based on GPS and gyroscope data, multiple continuous road edge points are associated to the same coordinate system by translation and rotation of the coordinate. Finally, the points are fitted and filtered adaptively by the model which is automatically selected by the improved RANSAC algorithm, and the real road edges are consequently extracted. Experimental results show the feasibility and practicability of the method.