WGIT*: Workspace-Guided Informed Tree for Motion Planning in Restricted Environments

The motion planning of robots faces formidable challenges in restricted environments, particularly in the aspects of rapidly searching feasible solutions and converging toward optimal solutions. This article proposes workspace-guided informed tree (WGIT*) to improve planning efficiency and ensure high-quality solutions in restricted environments. Specifically, WGIT* preprocesses the workspace by constructing a hierarchical structure to obtain critical restricted regions and connectivity information sequentially. The refined workspace information guides the sampling and exploration of WGIT*, increasing the sample density in restricted areas and prioritizing the search tree exploration in promising directions, respectively. Furthermore, WGIT* utilizes gradually enriched configuration space information as feedback to rectify the guidance from the workspace and balance the information of the two spaces, which leads to efficient convergence toward the optimal solution. The theoretical analysis highlights the valuable properties of the proposed WGIT*. Finally, a series of simulations and experiments verify the ability of WGIT* to quickly find initial solutions and converge toward optimal solutions.