Solving shortest path problems with curvature constraints using beamlets

A new graph representation of a 2D environment is proposed in order to solve path planning problems with curvature constraints. We construct a graph such that the vertices correspond to feasible beamlets and the edges between beamlets capture not only distance information but directionality as well. The A* algorithm incorporated with new heuristic and cost function is implemented such that the curvature of the computed path can be constrained a priori. The proposed Beamlet* algorithm allows us to find paths with more strict feasibility guarantees, compared to other competing approaches, such as Basic Theta* or A* with post-smooth processing.