Autonomous UAV Sensor Planning, Scheduling and Maneuvering: An Obstacle Engagement Technique

Abstract

An uninhabited aerial vehicle (UAV) equipped with an electro-optical payload is tasked to collect over a set of discrete regions of interest. By considering the discrete regions to be obstacles that must be engaged, rather than avoided, a new mathematical technique emerges. To frame the anti-obstacle-avoidance problem, we use Kronecker indicator functions to localize the totality of constraints associated with the discrete regions. A rich class of payoff functionals can be defined using nonsmooth constructs. We show that the integrated sensor planning, scheduling and UAV maneuvering problem can be framed under a single unified mathematical framework. The price for this unification is nonsmooth calculus. The practical viability of the new problem formulation is demonstrated by solving a sample problem using DIDO — a guess-free, advanced MATLAB optimal control toolbox for solving dynamic optimization problems.