Special Reports

Autonomy out of necessity — not because it's cool

The military is moving toward greater autonomy — not because it's cool or trendy, but because it must, according to a roboticist with the Naval Research Laboratory.

When autonomy is expensive, difficult to implement, tough to verify and oversee, why the push for it? Signe Redfield, speaking at the annual MILCOM conference in Baltimore on Tuesday, said the reason for the autonomy push is "because there is a massive chunk of problems that are completely intractable" that are now doable with autonomy.

There is a clear distinction among autonomy and similar systems, said David H. Scheidt at the same panel. An autonomous system makes its own decision when faced with a problem to which it doesn't have the answer, noted Scheidt, a member of the principal professional staff at Johns Hopkins University's Applied Physics Laboratory, adding that a truly autonomous system is one where at design time, all possibilities were not considered.

He offered three main classes for working with autonomous systems:

  • Teleoperation, where there are essentially the same mechanisms of control that exist with manned systems. With a UAS, for example, there is a throttle or a joystick that allows a human to fly it.
  • Automatic, where a human provides a waypoint and the system navigates without the user having to manage the throttle and the rudder.
  • Autonomous,where a pilot is no longer needed. Although there is a human in the loop, the human acts as more of a supervisor than a pilot.

Scheidt also referenced the Defense Science Board’s recent report on autonomy, which came out with value propositions for why there is a need for autonomous systems: he highlighted decision speed — where decisions need to happen faster than human cognition — as a good argument for autonomy.

When there is more time for decision-making, there is less value in the proposition for autonomy. But the more data involved in a problem, the more complex it becomes and the more likely the need for autonomous systems because machines can process structured data faster than humans, he said, adding that unstructured data is still under development. He also noted that persistence has a high-value proposition for autonomy given that robots don’t have to sleep.

Also at the same MILCOM panel, Charles Shoemaker, the lead for autonomous systems at the Army's Communications-Electronics Research, Development and Engineering Center, outlined frequently cited motivations for moving toward autonomous control of unmanned ground systems: reduced operator workload; potential for single operator or crew to control multiple systems; new communications options such as autonomous relay; and reduced potential for spectrum saturation, such as wireless interference.

Recommended for you
Around The Web