3D Scene reconstruction and motion planning for an autonomous mobile robot in complex outdoor scenes

Abstract

3D environment modeling and motion planning are two essential tasks for an autonomous mobile robot's outdoor navigation. In order to implement mobile robot large-scale motion planning, an edge-feature-based ICP algorithm is presented to complete 3D scene registration and reconstruction. According to the respective advantages of geometric features and terrain features, a novel motion planning approach is proposed based on the reconstructed scene. This method integrates the planar features and elevation information in one map to build the 3D roadmap, and then search a safety path with multiple constraints. A series of experiment results show the method's validity and practicability.