Object Detection for Autonomous Vehicle with LiDAR Using Deep Learning

Abstract

This paper presents an object detection for Autonomous Vehicle (AV) using deep learning algorithm. Currently, most AVs use the camera for visualization to detect surrounding objects. However, the performance of a sensor, such as a camera with visual perception, is diminished in dim light, for instance at night-time due to the less light environment. Thus, the study attempts to employ the Light Detection and Ranging (LiDAR) sensor that uses light in the form of a pulsed laser to calculate ranges and ultimately detect objects. The use of LiDAR with the recent deep learning algorithm, namely You Only Look Once (YOLO) v2, was simulated on the Robot Operating System (ROS) in the Linux environment. The collected data has undergone several filtering processes, which includes noise removal, downsampling, and transformation. The study then applies the model on real-time data from the LiDAR sensor to perform object detection. The results show that YOLOv2 can identify the objects better compared to Single Shot Detection (SSD) algorithm.