Speed Bump Recognition for Autonomous Vehicles Based on Semantic Segmentation

Abstract

In this paper, we combine 3Dmax and PreScan to make data sets and use the recognition algorithm to get the region of interest (ROI) of the speed bumps area. Then, we realize the speed bump recognition for autonomous vehicles. Besides, we do the post-processing operations on the area corresponding to the disparity map. Finally, we provide an innovative method to improve the accuracy of the distance measurement of the speed bumps. The results show that the method in this paper can accurately recognize the speed bumps and measure the distance under the requirements of the automotive embedded system.