Traffic noise evaluation at intersection using roadside light detection and ranging sensor

Abstract

Traffic noise significantly impacts public health, necessitating precise evaluation for effective monitoring, particularly at intersections. This study proposes a novel framework using roadside light detection and ranging (LiDAR) sensor to assess traffic noise, marking the first application of LiDAR-based background point cloud segmentation for noise evaluation. First, a point cloud segmentation method was introduced, leveraging integrated algorithms to identify buildings and vegetation. Then, a noise propagation model was developed, incorporating direct, diffractive, and reflective paths to evaluate environmental effects. In addition, noise attenuation by vegetation was quantified using point cloud density. Finally, noise maps were generated to visualize intersection noise levels. Experimental results demonstrated the segmentation method achieved an accuracy of 80.63%. The evaluation achieved a mean absolute error (MAE) of 1.24 dB and a coefficient of determination of 0.78 compared to sound level meter measurements, showcasing the model’s effectiveness and potential in evaluating traffic noise.