The Problem Our experience in computational command and control has included automated planning for unmanned air vehicles (UAVs) in several DARPA-supported efforts. In these programs, we constructed algorithms and software to plan (and to re-plan) UAV behaviors in the presence of partial, incomplete and noisy information from the battlefield and deceptive actions of adversaries.
The Solution With increasing reliance on unmanned systems for military operations comes the challenge of managing the battle-space in an efficient and effective manner. Current operations are manpower intensive, in some cases requiring multiple human operators for each unmanned vehicle. Computational automation, properly developed, can provide rapid response to these issues and allow for more efficient manpower usage. Our system plans for tens of vehicles against hundreds of targets, areas of unexplored and unidentified enemy and civilian entities, and heterogeneous, layered air defense systems.
The Outcome Our risk-sensitive approaches to air defense suppression applications demonstrated reductions in losses of up to 50% in test engagements. Our algorithms for sensing planning in support of offensive operations outperformed human planners, when scored by human commanders, by more than 30% in test engagements.