Team Planning for Unmanned Vehicles in the Risk-Aware Mixed-Initiative Dynamic Replanning System

The Navy envisions using heterogeneous teams of unmanned aerial vehicles (UAVs) and unmanned undersea vehicles (UUVs) for future missions. The operator must interact with both the team as an entity as well as with individual vehicles. To accomplish these missions, the team must autonomously plan activities to meet the team mission objectives while abiding by operating constraints. As part of the Office of Naval Research´s (ONR) Intelligent Autonomy program, Draper is developing the risk-aware mixed-initiative dynamic replanning (RMDR) system. The RMDR operator specifies team-level tasking, approves generated plans, and may re-task the team during the mission. The operator tasking includes Search tasks with value and precedence constraints, UUV Sentry tasks, UAV communication relay tasks, as well as vehicle, task, and region constraints. The RMDR team planner allocates the tasking among the available vehicles to accomplish the team objectives, abiding by the constraints. This paper focuses on the team-level planning capabilities and algorithms, specifically the Search task optimization and the Communication Relay task assignment. The Search task algorithm allocates Search tasks to vehicles using a mixed- integer optimization formulation that minimizes the total mission time and maximizes mission value. The communication relay task assignment algorithm calculates communication locations and time intervals for UAVs to maximize communication coverage for the UUV tasks.