Pilots could be usurped by neural networks
The next thing you know they'll be writing news...
An adaptive neural network could eventually replace airline pilots, according to researchers at the Georgia Institute of Technology, NASA and Boeing.
In 1989, 111 people died in a DC-10 plane when it crashed after suffering a hydraulic failure. Investigators concluded that had it not been for the skill of the pilots, there would have been no survivors, let alone the 185 people that lived to tell the tale. The pilots managed to avert a total disaster by using the engine's throttles to steer the plane.
After this crash, engineers took a closer look at what the pilots' handiwork. The result was a new system that allows a plane to fly and land on autopilot, using only the power of the engines. The autopilot can even cope with a wing going missing mid-flight, or damage to the airframe, the researchers said.
Following successful testing in an F-15 fighter jet, the autopilot is being combined with an adaptive neural network, developed by Anthony Calise, and engineer at Georgia Tech. He says that the product of this union will be capable of responding to single and multiple failures, including full loss of control surfaces on a 747.
Rather un-imaginatively named the "Integrated Neural Flight and Propulsion Control System" or INFPCS, the network will react to incoming data within a few tenths of a second.
It has not been decided whether the technology has a place on passenger aircraft, as yet. ®
Sponsored: Benefits from the lessons learned in HPC