An Entropy-Based LiDAR Loop Detection Algorithm

Since the existing loop-closure detection algorithm based on LiDAR is relatively simple in extracting point cloud features, and the problem of yaw angle needs to be considered in later matching, this paper proposes a multi-feature laser scanning context method based on entropy to achieve stable and reliable loop-closure detection. Firstly, the global descriptor is created by extracting multiple feature information of the LiDAR point cloud to build an image to record the three-dimensional information of the surrounding environment fully; At the same time, an image-matching algorithm based on entropy is used to make the loop detection rotation invariant, so that the yaw angle does not need to be modified. Relevant experiments have confirmed the superiority of the method proposed in this paper and the excellent performance of the experimental results.