MinkUNeXt-SI: Improving point cloud-based place recognition including spherical coordinates and LiDAR intensity

Place recognition is crucial for autonomous navigation, requiring accuracy despite seasonal, weather, or environmental changes. This paper presents our method, MinkUNeXt-SI, which, starting from a LiDAR point cloud, preprocesses the input data to obtain its spherical coordinates and normalized intensity values. It then produces a robust place recognition descriptor. To that end, a deep learning approach that combines Minkowski convolutions and a U-net architecture with skip connections is used. MinkUNeXt-SI’s high recall results of up to 99.15% demonstrate that this method reaches and surpasses state-of-the-art performance while it also generalizes satisfactorily to other datasets. Additionally, we showcase the capture of a custom dataset and its use in evaluating our solution, which also achieves recall results above 85%. These results illustrate that our method improves upon the state-of-the-art by close to +10% in this growing field. Both the code of our solution and the runs of our dataset are publicly available for reproducibility purposes.