On September 26, information from the Qichacha APP revealed that Hangzhou Unitree Robotics Co., Ltd. has recently made public a patent titled "A Multi-Floor Composite Navigation Method and Device for Robots." This particular patent is situated within the realm of robotic navigation technology. It specifically tackles the prevalent issues in current robotic multi-floor navigation systems, namely the limited capacity for information retention in grid maps and the inability to generate the most efficient planned paths.

To address these challenges, the patent introduces a novel navigation approach. The method commences with the segmentation of a 3D point cloud map according to different floors, resulting in distinct 3D point cloud maps for each level. This step significantly enhances the system's ability to retain information. Following this, each floor's 3D point cloud map is subjected to projection and rasterization processes, yielding multiple 2D maps that correspond to each floor.

Subsequently, the patent outlines the analysis of the interconnections and structural characteristics between floors, culminating in the creation of an inter-floor connection map. This map serves as a crucial component in the final stage of the method, where a topological path search is conducted. Leveraging both the inter-floor connection map and the 2D same-floor maps, this search identifies the most optimal planned path. Consequently, it facilitates precise navigation for robots across multiple floors within the same building, marking a significant advancement in the field of robotic navigation technology.