Dynamic Point Removal

In the field of robotics, point clouds have become a key representation for maps. The task of dynamic point removal aims to eliminate points corresponding to dynamic objects from point clouds, thereby enhancing the performance of downstream tasks such as localization and global path planning. This study proposes an easily scalable unified benchmark framework that includes both reconstructed state-of-the-art methods and novel evaluation metrics, providing an in-depth analysis of the limitations of existing techniques and validating them using datasets from multiple sensor types. All relevant code and datasets are made publicly available to facilitate further research and application.