Multi-Objective Design and Path Planning Optimization of Unmanned Aerial Vehicles (UAVs)

Abstract

This paper presents the topic of combined UAV design and path planning using biobjective optimization. Two optimization models, integrated and decomposed, are developed and solved for path planning and design optimization of UAVs. Both models minimize a weighted sum of risk and time. The risk is based on a simulation of a UAV’s crash after it loses power and is estimated by computing an expected number of on-the-ground human population fatalities which are located at or near the UAV crash site. The flight time is computed for the UAV as it travels from a start to an end point. The integrated model solves the design and path planning optimization problem all-at-once. The decomposed model has two subproblems: one for the design optimization of UAV and another for its path planning optimization. A new decomposed method for solving the path planning subproblem while considering the design optimization subproblem is presented, which solves the path planning subproblem for a discrete graph. Results for this new method are compared to an integrated all-at-once approach that does not treat the design and path planning subproblems separately, which can indicate the method proposed is less vulnerable to local minima in the path planning optimization subproblem.