Landing Sites Chosen In Midair by Autonomous Copters [video]

Many remotely piloted vehicles in use today demonstrate relatively little intelligence of its own. For a conventional fixed-wing aircraft including most UAVs, remote piloting is comparatively straightforward in some respects. Its flight path may have a known point of origin and a predetermined landing strip. Any ambiguities that are encountered mid-air are handled by remote pilots who use their own reasoning abilities to direct the machine away from danger.

For instance, the confusion that might arise when landing a machine like a helicopter in the middle of an unknown street (a la "Blackhawk Down") or on the top of a hospital building are minimized because a fixed-wing aircraft can't do that.

A research team would like to change that scenario by creating smarter robocopters. In work sponsored by the Army Telemedicine and Advanced Technology Research Center, a collaboration is underway between Carnegie Mellon University roboticists and aeronautics specialists at Piasecki Aircraft and Boeing. 

The investigators explain some of the challenges they faced in determining -- in real time -- which landing sites would be suitable.

To put this stream into a form the planning software can use, the system constantly updates two low-level interpretations. One is a high-resolution, two-dimensional mesh that encodes the shape of the terrain for landing; the other is a medium-resolution, three-dimensional representation of all the things the robot wants to avoid hitting during its descent. Off-the-shelf surveying software can create such maps, but it may take hours back in the lab to process the data. Our software creates and updates these maps essentially as fast as the data arrive.

Inference: Artificial intelligence will continue to co-evolve with autonomous flying machines.

Some of the technology is being pursued by Near Earth Autonomy, founded by one of the co-authors.

This story via IEEE Spectrum.