Proactive decision support for dynamic assignment and routing of Unmanned Aerial Systems

Abstract

Unmanned Aerial System (UAS) missions are executed by teams of operators with highly specialized training and roles; however, the task demands on each operator are highly variable, often resulting in uneven workloads among operators and sometimes in mishaps. Therefore, there is a need to develop anticipative and effective decision support algorithms that permit the evaluation of courses of action (COAs), while assuring that operators are attending to the right task at the right time and that task demands do not exceed the operator's cognitive capabilities in dynamic multi-mission environments. Motivated by the need to assist UAS operators in efficiently managing their workloads, this paper develops algorithms for the dynamic scheduling of UAS tasks by providing efficient COA recommendations in an unobtrusive manner. The dynamic scheduling of a set of UASs to search for targets with varying rewards is an NP-hard problem. We model this problem as an extension to the open vehicle routing problem (OVRP). Extensions to OVRP include risk propensity of human decision making, task deadlines, and multiple vehicle types. UAS operators would benefit greatly from the COA recommendations and the algorithms proposed in this paper by a) enhancing rapid planning and re-planning capabilities; b) proactive allocation of UASs, while balancing operator workloads; and c) adapting plans as new targets of opportunity appear or information is updated about a target and/or UAS. The proposed algorithms are embedded in the Supervisory Control Operations User Testbed (SCOUTTM), an experimental paradigm developed by the Naval Research Laboratory-Washington DC.