Object detection with LiDAR has proven to be an efficient way to develop automated guided vehicles (AGVs). Snow noised point cloudsare de-noised with the help of several filters that work with different characteristics of snow flakes.

In the first part of this work, literature is reviewed on approaches to handle snow-noise inobject recognition. Secondly, based on the literature this work presents a method that classifies between different densities of snowfall to use each filter for its best use case. The filters were tested for a file with very high snow point density.

Furthermore, an own approach is presented that determines the snow in 3D-lidar data and distinguishes it from obstacles. The filters are tested in the second part of the work and compared with each other.