Investigation of Influence from Variation in Color on LiDAR Sensor for Perception of Environment in Autonomous Vehicles

Perception of the vehicle surrounding is one of the most vital tasks in achieving a vision of fully autonomous driving. To achieve this task, forward-looking sensors like RADAR, LiDAR, and camera are widely used for object detection and classification. Accurate range information at several closely distributed points from the object's outer surface gives the LiDAR sensor an advantage for estimating the shape and size of the opponent object. Though the LiDAR sensor can provide accurate range information in point cloud data, there are some variations in intensity values and no. of points reflected from objects with different colors. In this paper, we analyze the differences in the LiDAR intensities and no. of point clouds received from different colored objects by recording LiDAR data of three different colored cars at varied orientation angles and distances from the LiDAR sensor in similar environmental conditions and the impact it might have in perception of vehicle surrounding. The results show that the color of the object has a considerable influence on the LiDAR data and can affect the output of the pre-crash estimation algorithms. This highlights the need to further investigate this effect to improve the pre-crash estimation algorithms by considering the behavior of the variation in the LiDAR data based on the object color.