LiDAR Information based High-Speed-High-Accuracy Object Recognition Network for Autonomous Vehicles*
Xin Chen, Bin Zhang · 2024
Rapid advances in in-vehicle technology are leading the automotive industry in a safer, more efficient, and intelligent direction. Object recognition technology based on LiDAR sensor information enables highly accurate object location detection and plays an important role in autonomous driving systems. However, there is a trade-off between detection speed and accuracy, and the need for object recognition methods that are both fast and accurate is high. In this study, we propose a method to achieve both detection speed and accuracy by improving CenterPoint [1]. The proposed method uses a lightweight deep learning network with attention mechanisms. It is expected to promote the spread and development of autonomous driving technology and contribute to further innovation in the automotive industry.