 This paper proposes a faster and more accurate object detector for autonomous driving based on the YRL of five algorithm. It uses structural reparameterization, REP, to enhance the accuracy and speed of the model, as well as introduces neural architecture search to reduce redundancy in the multi-branch reparameterization module. A small object detection layer is also added to the network, while the coordinate attention mechanism is used to improve the recognition rate of small objects such as cars and pedestrians. Experimental results demonstrate that the detection accuracy of the proposed method on the KITI dataset reached 96.1% and the frame per second was 202, outperforming other mainstream algorithms and significantly improving the accuracy and real-time performance of unmanned driving object detection. This article was authored by Xiong Jiu, Ying Tong, Hong Mingqiao, and others.