Modeling of an Automotive Grade LIDAR Sensor

Abstract

Automobiles use LIDAR sensor to detect different objects around the vehicle. For system analysis and study, this paper is proposing a LIDAR sensor model, which takes into consideration the impact of the real world information. Impact of the environment on the LIDAR sensor is modeled and all the possible parameters of the LIDAR are modeled as configurable parameters. Option to import 3D objects of any shape, type or dimension via FBX file format is also incorporated. 3D objects in FBX format shall be converted to a set of triangles first, which approximate the surface mesh of the 3D object. Ray cast modeling is then used to detect whether in a vertical distribution of LIDAR beams, an intersection occurs between LIDAR beam and any of the triangular face. If there is a collision, the collision point shall be saved as the point cloud data. This will be repeated around the sensor, and all such point cloud data points shall be appended to the final point cloud data. These point cloud data is then subjected to segmentation and object detection using belief theory. Either the processed point cloud data in object information format or the unprocessed raw point cloud data can be produced as the output by the proposed LIDAR sensor model.