Linear Programming and Language Processing for Human-Unmanned Aerial-Vehicle Team Missions

Abstract

This paper presents a manned-vehicle/unmanned-aerial-vehicle (UAV) mission system that enables an operator in a manned aircraft to issue mission level commands to an autonomous aircraft in real time. A natural language interface allows the manned and unmanned vehicle to communicate in languages understood by both agents. A task scheduler transforms the commands into a dynamic mission plan consisting of task waypoints. These are then given to a mixed-integer linear programming (MILP)-based trajectory optimizer, which safely guides the vehicle through a partially known environment in real time. The MILP trajectory planning formulation and its implementation are discussed in detail. Integrated simulation and June 2004 flight-test results that used an F-15 and an autonomous T-33 equipped with Boeing's unmanned combat air vehicle (UCAV) avionics package are presented. These activities were part of the Capstone Demonstration of the Defense Advanced Research Projects Agency-sponsored Software Enabled Control effort. The flight tests mark the first time that an onboard MILP- based guidance system was used to control a UAV. They also mark the first time that a natural language interface was used by a manned vehicle to task a UAV in real time.