A Grid-Based Path Planning Approach for a Team of Two Vehicles with Localization Constraints

This paper proposes a path finding method for two un-manned aerial vehicles with localization constraints using a modified shortest path algorithm. A beacon vehicle has GPS or other absolute positioning information, and a target vehicle has only bearing information taken relative to the beacon vehicle or known stationary landmarks. By overlaying a grid on the map and discretizing the position uncertainty, the path planning problem for two vehicles can be formulated as a dynamic programming problem and solved using a modified form of the A* shortest path algorithm. Edge costs are found using a factored covariance method for an Extended Kalman filter based on results available in literature. In simulation, paths found from the dynamic programming method outperform a greedy algorithm.