A Minimum Risk Approach for Path Planning of UAVs

Abstract

In the last years, the Flight Mechanics Research Group of Politecnico di Torino started a wide research activity, focused on exploration and implementation of path planning algorithms for commercial autopilots, typically adopted on unmanned vehicles. Different path planning approaches were implemented in a Matlab/Simulink based tool, generating waypoints sequences with four methods: geometric predefined trajectories, manual waypoints definition, automatic waypoints distribution (i.e. optimizing camera payload capabilities) and, finally, a comprehensive A*-based approach. The tool was also integrated with functions managing the maps used for planning. In this paper, two approaches to path planning in presence of orographic obstacles are detailed. The first algorithm is subdivided in three phases: the generation of a risk map associated with the ground orography, the transformation of the map in a digraph analyzed with the A* algorithm (to obtain the path with minimum risk/minimum distance) and finally the smoothing phase, to obtain a flyable waypoint sequence, realized with the Dubins curves. The structure of this method was defined and implemented, but its optimization is still in progress. The second algorithm is based on the same risk map, but optimizing polynomial curves with a genetic algorithm. This method produces a flyable waypoints sequence, minimizing a cost function reflecting path length and collision risk. An extension of this path solver, including aircraft performance constraints, was also considered. This method was tested on sample domains and its computational cost has still to be evaluated before the implementation in the tool.