Bedlam OCU™: Adversarial Reasoning System for Unmanned Platforms

Primordial, Carnegie Mellon University, iRobot, and AeroVironment propose Bedlam OCU™—an adversarial reasoning system for unmanned platforms. Team Primordial will start with our existing state-of-the-art Surveyor OCU, which already fully or partially meets 78% of the topic’s requirements including enabling UAV/UGV collaboration, autonomously tracking targets, predicting adversary movement, providing a mission execution framework, enabling mission planning, mitigating cognitive load, and having a practical fielding plan. Using Surveyor OCU, Team Primordial has conducted numerous live, semi-autonomous ground vehicle tracking experiments using AeroVironment Raven UAVs and iRobot PackBot UGVs. From this proven baseline, we will develop Bedlam OCU—a handheld version of the Surveyor OCU with upgraded adversary prediction and target reacquisition modules. The upgraded adversary prediction module will account for terrain features, evasive maneuvers, and likely destinations. The upgraded target reacquisition module will enable an unmanned platform to autonomously reacquire a target if tracking fails. Primordial will test the upgraded modules using simulations incorporating computer- and human-controlled adversaries executing various strategies such as maximizing speed, minimizing visibility, or randomizing movements. Working with iRobot and AeroVironment, Primordial will also develop a transition plan for ensuring Bedlam OCU’s ultimate fielding. Finally, Primordial will deliver a report detailing Bedlam OCU’s design, development, and testing.