Low-Cost Radar for Object Tracking in Autonomous Driving: A Data-Fusion Approach

• Published in: 2018 IEEE 87th Vehicular Technology Conference (VTC Spring) pp. 1 - 5
• Main Authors: Aldrich, Ryan, Wickramarathne, Thanuka
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2018
• Subjects: Kalman filters; Laser radar; Mathematical model; Radar tracking; Three-dimensional displays; Vehicles
• ISSN: 2577-2465
• DOI: 10.1109/VTCSpring.2018.8417751

Automotive LiDAR and Radar sensor fusion is quickly developing as a technique used to provide ground truth data for radar development. Radar sensors in an automotive application are known to have many sources of noise such as bumper fascia distortion, non-linear component behavior plus angle and range resolution error from known Additive Gaussian White Noise sources in the Radio Frequency front end of the radar. State of the art LiDAR sensors such as the Velodyne HDL-32E and HDL-64E can be mounted outside of the bumper fascia in a test vehicle configuration and are not subject to the same amount of internal noise as an RF system. This means that LiDAR point cloud data can be used for reference system measurements in a possible data fusion system that can be used to characterize radar accuracy and performance. Representative LiDAR and Radar track data was collected and then was input into multiple instances of a Kalman filter that provided estimates used as a basis for comparison between the two modalities.