Optimizing UAV Mission Control Published June 15, 2010 By Maria Callier Office of Scientific Research ARLINGTON, Virginia -- Air Force Research Laboratory-sponsored engineers at Boston University are pursuing a theoretical approach to unmanned aerial vehicle fleet automation that would save Air Force pilots substantial effort by enabling UAVs to adapt more rapidly in response to unforeseen events and, ultimately, to demand less human supervision. While unmanned systems currently rely on low-level automation for basic functions such as navigation, stabilization, and trajectory, systems operation remains nonetheless quite labor-intensive for pilots given the variable flying conditions experienced by UAVs. Led by Drs. David Castañón and Christos Cassandras, the BU team's work seeks to address this shortfall by incorporating mission control and decision making into the realm of UAV automation. The focus on mission control involves the optimization of midlevel control approaches that go beyond simply improving stability and tracking trajectories. Among the research areas of interest are functions such as automated task partitioning among the member UAVs of a given UAV team, or fleet. In this scenario, participating (or designated) UAVs would monitor the success of individual activities, replanning (and otherwise responding) to accommodate contingencies or failures in executing the specific task(s) at hand. To date, the BU researchers have developed mathematical algorithms enabling nearly optimal decisions under realistic model conditions. They have thus far based their approach on the need to accommodate a number of uncertainties requiring complex computations that are nearly impossible to implement in real-time systems. Accordingly, the team has exploited classes of models for which fast algorithms can be developed and then extended to generate decisions in models that are more complex and, as such, better reflect UAV problems (and desired automation features) of interest. While much of their work is rooted in mathematical analysis, the BU researchers have also developed a robotics test scenario for evaluating their approach. BU students at all levels--graduate and undergraduate alike--are involved in this testing, which uses teams of small robots equipped with sensors to represent the UAVs. In these tests, the robots must prove their capacity to function in a midlevel control environment while subjected to the distraction of unforeseen events (e.g., loss of team members, arrival of new tasks, and/or discovery of new information). As the BU researchers learn more about the various environments in which UAVs operate, they will continue to hone their results, with the long-term goal of increasing the level of self-sufficiency available to future AF UAV fleets.