FURI | Summer 2021

Autonomous Vehicle Object Detection on LiDAR Point Cloud Using Deep Learning Model

Data icon, disabled. Four grey bars arranged like a vertical bar chart.

In order to improve the current advanced driver assistant system, more accurate input information of the vehicle’s surround is required. The project serves to improve this input information as LiDAR sensor data is not affected by the lighting of the surrounding environment. The project can also be applied in security systems where cameras’ ability to collect clear images is limited. The system successfully runs two different pre-trained machine learning models PointPillars and PointSeg to perform object detection and tracking. However, the detection results’ accuracy needs to be improved, so more effort will be put into training the pre-trained models.

Student researcher

Vu Nguyen

Vu Trung Nghia Nguyen

Electrical engineering

Hometown: Ho Chi Minh city, Ho Chi Minh, Vietnam

Graduation date: Spring 2023