Remote Handling operations for ex-vessel transportation in fusion power plants: A Fast Geometric Path-Finding Heuristic for Single-Robot Systems Using Constrained Delaunay Triangulation

This paper proposes a path planning algorithm, the Constrained Delaunay Triangulation Rapidly-exploring Random Tree Star (CDT-RRT*), for determining a feasible path between a start and goal node on a 2D map. While exploiting the advantages of Constrained Delaunay Triangulation (CDT) for the discretization of the map, the algorithm manages to find an optimized sequence of triangles and sequence of points belonging to some edges of this set of triangles. The resulting geometrical path is computationally faster than RRT* but with lower performances in terms of length. To improve the aspect related to path length, a final version of CDT-RRT* is presented in this paper. The addition of backtracking leads to high performances in terms of computational time while improving the path length parameter, making the resulting CDT-RRT* average value comparable to the one of RRT*. One of the main advantages of CDT-RRT* is its scalability, thus, although the nuclear fusion scenario constitutes the primary case study, the proposed methodology is also generalizable to industrial warehouses and other constrained environments.