A snow depth estimation method with LiDAR system in snow groomer

Providing high-resolution and real-time snow depth estimations in front of the snow groomer is necessary for the proper navigation and working efficiency improvement. However, accurate and real-time estimation of the snow depth during grooming operations in ski resorts is challenging due to the time-varying terrain of pistes. In this study, a real-time snow depth estimation method was established utilizing the LiDAR based multi-sensor perception system, where the estimated snow depth distribution was achieved by the elevation differences between the real-time local snow-covered grids and the snow-free reference map. A new multi-sensor fusion-based simultaneous localization and mapping (SLAM) approach especially for the ski resort environment was built to achieve the snow-covered and snow-free reference terrain maps. Also, we developed a snowfall denoising method and a piste extraction algorithm based on ski resort geometric features to improve the SLAM accuracy in snow-covered terrain. To validate the snow-depth estimation approach, a series of experiments were performed at the Wanlong Ski Resort and Linyu Ski Resort, China. Results indicates that the proposed snowfall denoising, snow piste extraction effectively improve the SLAM accuracy in snow-covered terrain map construction. The snow depth on flat ground was measured with an average error of 0.034 m, whereas the error on the slope was 0.042 m at the tested ski resorts. Additionally, the calculation period satisfies the requirements for real-time monitoring. The proposed method can provide precise and real-time depth estimations, facilitating the operational processes of snow groomers and enhancing the pistes maintenance efficiency.