The article describes an approach to the analysis of a local indoor navigation map derived from a lidar point cloud represented as a projection on a horizontal plane. The article analyzes the applicability of the graph theory and binary image processing methods for structuring and uniting the elements of a noisy and segmental local navigation map layout. On an initial map layout (walls, free space and internal environment objects), stand-alone interior elements are detected, and the points of wall borders are filtered and closed up. Mathematical morphology methods and shortest path and connected component separation algorithms are used to complete these tasks. The authors have developed a trajectory generation algorithm consisting of several turning and straight-line motion commands.
展开▼