Robust Scheduling and Routing for Collaborative Human/Unmanned Aerial Vehicle Surveillance Missions

Abstract

A supervisory mission in which a team of unmanned vehicles visits a set of targets and collects sensory data to be analyzed in real time by a remotely located human operator is considered. A framework is proposed to simultaneously construct the operator’s task-processing schedule and each vehicle’s target visitation route, with the dual goal of moderating the operator’s task load and preventing unnecessary vehicle loitering. The joint scheduling/routing problem is posed as a mixed-integer (nonlinear) program that can be equivalently represented as a mixed-integer linear program through expansion of the solution space. In single-vehicle missions, it is shown that an alternative linearization that does not increase the problem size exists. Next, a dynamic solution strategy that incrementally constructs suboptimal schedules and routes by solving a comparatively small, mixed-integer linear program whenever the operator finishes a task is introduced. Using a scenario-based extension, this dynamic framework is then modified to provide robustness to uncertainty in operator processing times. The flexibility and utility of these algorithms are explored in simulated missions.