Experimental Performance Analysis of a Self-Driving Vehicle Using High-Definition Maps

Abstract

In this paper, we present our first experimental results on self-driving vehicles using high-definition maps. We will start with the CAN bus based Drive-By-Wire subsystem, on board computers, and the sensors used in this study. Then, we discuss the sensor fusion and vehicle guidance algorithms, and finally present our experimental results. We also have a video recording of our experiment, and its YouTube link is shared in the paper. All of the experimental results and plots presented in the paper, and the video link refer to the same self-driving experiment that we did on the Florida Polytechnic University campus. The autonomous vehicle (AV) methodology adopted in this work has some similarities with the Cruise AV’s approach and the use of high definition (HD) maps. The research vehicle used in this work is equipped with radar, lidar, camera, GPS, and IMU sensors, but in this work we use only the GPS, wheel rotation and camera sensors. After presenting our first experimental AV results, we comment on sensor fusion related issues, and possible future steps for improvement.