Minimizing Mission Risk in Fuel-Constrained Unmanned Aerial Vehicle Path Planning
Petter Ögren, Maja Winstrand · Journal of Guidance Control and Dynamics · 2008
One of the main reasons for using a UAV instead of a manned aircraft is that the mission that need to be carried out is very dangerous. In such cases, we believe that the risk of losing the vehicle should be made an explicit and important part of the mission planning. The mentioned risks can often be estimated on a small scale, using detailed simulation models, but the tools to propagate such estimates to the overall mission planning level has been missing. In this paper we propose one way to take such local risk estimates into account in the mission planning, and at the same time make the tradeoffs against other path properties, such as fuelor time constraints, explicit. First we show how to rewrite the accumulated risk estimate into a form that matches the formulation of the so-called weight constrained shortest path problem. Then we approximately solve it using a series of regular shortest path problems. These are in turn easily solved by standard algorithms such as Dijkstras or A*. UAV path planning has been extensively studied in the literature. However, to the best of our knowledge, only Chaundry et al. give the operator an explicit estimate of the risk associated with the proposed mission path. Furthermore, only Zabarankin et al. enable the user to clearly state constraints in terms of available fuel or time. In contrast to both Chaundry et al. and Zabarankin et al. who focus on the interactions of the UAV with a single SAM battery, the proposed approach enables planning of missions dealing with multiple threats, in the same way as the papers by Beard and others. 8, 11 The organization of this paper is as follows. In Section II, we state the proposed risk vs fuel path planning problem as well as two other path planning problems from the litterature. We then show how the proposed problem can be rewritten and approximately solved in Section III. A detailed example illustrating the approach is presented in Section IV, and finally, the paper is concluded in Section V.