This paper introduces a slope detection method based on point cloud data from 3D LiDAR for quadruped robots in unknown environments. For quadruped robots, they need to adjust their gaits according to different slope angles to avoid some potential dangers. 3D LiDAR is used to gather point cloud data, which is superior to 2D LiDAR in speed and accuracy. In this paper, a slope detection method using bilateral filtering and RANSAC algorithms is discussed. The experiments of slope detection are fulfilled with the consideration of different angles of slopes and different orientations to the slopes, and the results demonstrate that errors of angle estimation is small.