A Risk-aware Path Planning Method for Unmanned Aerial Vehicles

This paper proposes a risk-aware path planning method for Unmanned Aerial Vehicles, with the aim to generate safe flight paths minimizing the risk to the population. The proposed approach consists of two phases: first, an off-line path planning computes the optimal global path in a static environment considering the risk; then, taking into account a dynamic risk-map, an on-line path planning adjusts and adapts the off-line path. The risk-map is a location-based map, in which each cell represents a specific location with an associated risk-cost. The off-line path planning is performed by the riskA* algorithm. It is based on the well-known A* algorithm, enhanced considering the minimization of the risk-cost. The off-line path planning is executed in a static environment and it has no time constraints. On the contrary, the on-line path planning needs to adapt the path in a short time, thus a fast response constitutes a critical design parameter. The on-line path planning is performed by a novel algorithm, called Borderland. Borderland uses a check and repair routine, then it identifies and adjusts only the portions of path involved by changes in the dynamic risk-map. Simulation results corroborate the validity of our approach.