LIDAR based Detection of Small Vehicles

This paper proposes a lidar based vehicle detection system able to detect small vehicles such as cars and vans. With the recent advances in the sector, Autonomous or semi-autonomous vehicles have gained large popularity. The biggest challenge for these vehicles is fast and reliable object or obstacle detection. A 3-D lidar has been the most reliable sensor to provide this information. To extract meaningful information from LiDAR point clouds current systems, deploy various deep learning models. These models are complex and work on large data. In this paper a system is proposed to detect the vehicles from lidar data with point cloud processing and machine learning. Preprocessing is performed with the Open3d library. For detection vehicle dimensions are used. For classification four different classifiers have been used. The system can detect vehicles such as cars and vans.