Avant Guard

Players direct a team of robots to find the optimum level of autonomy.
Players direct a team of robots to find the optimum level of autonomy.
Players direct a team of robots to find the optimum level of autonomy.
Players direct a team of robots to find the optimum level of autonomy.
Players direct a team of robots to find the optimum level of autonomy.
Players direct a team of robots to find the optimum level of autonomy.

The introduction of autonomy in the battlefield raises profound and practical questions. Researchers assume a future airman may be the only human in a team of UAV scouts. They assume that each plane’s autonomy is limited by design, not by technology. If so, what level of UAV autonomy best executes on the human intent?

Ideally, players learn nothing during gameplay. It is the researchers who learn. They measure the degree to which one pattern of autonomy anticipates the player’s needs. With ten levels of autonomy in each of four domains (Orient, Observe, Decide, Act), the testbed offers the researcher 10,000 autonomy patterns.

Players guide a convoy through hostile urban terrain. They depend on sensor data from three fly-ahead UAVs, whose degree of independent action is set by researchers.

Four turn-key testbeds were installed in the Air Force Research Labs at Wright Patterson Air Force Base.