降雨条件对车载激光雷达性能影响的试验研究

激光雷达是自动驾驶车辆最为关键的传感器之一，被广泛用于车辆定位、目标检测与跟踪等任务。然而，激光雷达的点云数据会受到恶劣天气（如雨、雾、雪等）的严重影响，致使自动驾驶全天候行驶仍然面临着巨大挑战。为了量化评估恶劣天气对激光雷达性能的影响，分析了降雨环境下激光雷达的性能，基于构建的场地降雨模拟系统控制降雨量，通过多视角的静、动态试验定性与定量分析激光雷达测距精度、典型目标点密度、有效检测距离等性能参数与降雨量之间的关系。试验结果表明：车辆作为目标物时，目标物上的激光点云受降雨的影响最大，相较于无雨环境，中雨时打在汽车上的激光点数降低幅度超过了60%，检测距离下降了69%，并且随着降雨量的增大激光雷达对目标的有效检测距离持续下降；试验方法和结果对于测试评价自动驾驶性能及提升降雨环境下的激光感知能力具有重要意义。

Experimental Study of Influence of Rain on Performance of Automotive LiDAR

Light Detection and Ranging (LiDAR) is one of the most critical sensors for autonomous vehicles, which is widely employed for vehicle localization, object detection, and tracking tasks. However, LiDAR data can be severely impacted by adverse weather (e.g., rain, fog, snow). This brings a tremendous challenge to autonomous driving in all weather conditions. Therefore, it is essential to quantitatively assess the impact of adverse weather on LiDAR performance. This article focuses on analyzing the performance of LiDAR in a rainfall environment. Based on the constructed site rainfall simulation system to control rainfall, the relationship between LiDAR performance parameters (e.g., ranging accuracy, typical object point density, effective detection distance) and rainfall was qualitatively and quantitatively analyzed through multi-view static and dynamic tests. The experimental results show that the LiDAR point cloud on the object is most affected by rainfall when the vehicle is used as the object. Compared with a no-rainy environment, the number of LiDAR points hit on the vehicle during moderate rain decreases by more than 60%, and detection distance decreases by 69%. Moreover, the effective detection distance of LiDAR for the object continued to decrease with the increase of rainfall. The experimental methods and results are of immense significance for testing and evaluating autonomous driving performance and improving LiDAR perception ability in a rainfall environment.