Optimal Configuration Method of Multi-source Sensors for Autonomous Vehicles Considering Three-dimensional Perception

The proliferation of sensors and computer technology accelerates the advancement of autonomous driving technology. Manufacturers are continuously introducing various autonomous driving systems that rely on perception sensors and vehicle computers for environment detection and real-time data processing. In the process of seeking the optimal configuration of multi-source sensors for self-driving cars, this paper proposes a joint optimal configuration scheme of Solid-State (SS) LiDARs and High Definition (HD) cameras based on three-dimensional perception. By analyzing the sensing performance of SS LiDARs and HD cameras, a multi-sensor joint configuration optimization model is established, and through the quantification of the 3D sensing area range, it aims to achieve the goal of maximizing the coverage of the sensing area under the condition of ensuring redundant configuration of critical areas. In this model, the cost and computational resources required for sensors are considered as the key constraints to find the number and the placement of SS LiDARs and HD cameras installed in the optimal configuration. To address this problem, the Particle Swarm Optimization (PSO) algorithm is selected for model solving and the superiority of the proposed method is verified through comparative simulation experiments. The results are beneficial in providing decision support for the joint optimization design of perception schemes in autonomous vehicles, thereby facilitating the creation of sensor configurations that are both more efficient and more effective.