New pilot safety system could save lives

October 17, 2000

THE feat of two heroic pilots who saved 185 lives when they crash-landed a crippled airliner has inspired the development of a new computer-controlled safety system that could one day allow damaged airliners to land themselves.

In 1989, hydraulic failure left a United Airlines DC-10 with nothing but two of its engines to control steering and altitude. The plane crash-landed at Sioux City airport in Iowa, killing 111 people. But aviation safety experts say 185 other passengers would have perished if it weren't for the pilots' clever use of the engine throttles to guide the plane.

Impressed by the pilots' skills, aerospace engineers at the Georgia Institute of Technology, NASA and Boeing have developed a system that will let the autopilot fly and land the plane using only engine power. "It can even cope with damage to the airframe or part of its wing shot off," says aerospace engineer Anthony Calise at Georgia Tech.

The system is designed to detect any damage to either the engines or any of a plane's flight control surfaces, and instantly adjust the remaining control surfaces or power resources to compensate. Their system has already been successfully tested on an F-15 fighter and an MD-11 passenger jet. In both cases the planes landed without human assistance using power-only control (POC), says NASA researcher John Kaneshige. It has also been tested in simulators of a variety of other aircraft, including Boeing 747s and the X36, a prototype space shuttle replacement.

But POC is limited in that it doesn't make use of any parts of control surfaces that might still function. So it is now being combined with an adaptive neural network developed by Calise. The resulting system, with the unwieldy name of the Integrated Neural Flight and Propulsion Control System (INFPCS), should be capable of responding to "single failures or multiple failures, all the way down to full loss of control surfaces on the 747," says Calise.

Other neural networks have been developed in the past to do a similar job, but these require offline "training" and take up to four seconds before the neural network reconfigures the plane and compensates for the damage (New Scientist, 24 April 1999, p 5). Calise's neural network has been designed to react within a few tenths of a second, fast enough for the reconfiguration to go unnoticed by the pilots flying the aircraft-and far faster than a person could react.

Rather than constantly scanning for failures or damage to the plane, the INFPCS compares what the pilot is doing with the aircraft's behaviour. If they don't match, it assumes a failure has occurred and attempts to compensate. "It uses a desired handling module," says Kaneshige. "If it's not behaving in a particular way [the INFPCS] modifies it."

But the development of INFPCS raises a disturbing question. "If you don't need to have a pilot trying to make these desperately difficult manoeuvres, then why have a pilot at all?" asks Tom Anderson of the Centre for Software Reliability at the University of Newcastle upon Tyne.

But don't expect your next holiday flight to be equipped with an INFPCS system, warns Kaneshige. "It's still to be determined whether this is a passenger-carrying technology."
Author: Duncan Graham-Rowe

New Scientist issue: 21 October 2000


New Scientist

Related Neural Network Articles from Brightsurf:

Performance test for neural interfaces
Freiburg researchers develop guidelines to standardize analysis of electrodes.

Neural cartography
A new x-ray microscopy technique could help accelerate efforts to map neural circuits and ultimately the brain itself.

New neural network differentiates Middle and Late Stone Age toolkits
The change from Middle Stone Age (MSA) to Later Stone Age (LSA) marks a major cultural change amongst our hunter-gatherer ancestors, but distinguishing between these two industrial complexes is not straightforward.

Early neural activity associated with autism
Researchers at the University of California, Los Angeles, have found evidence of signature brain activity in infants that predicted ASD symptoms later at 18 months old.

Recovering data: NIST's neural network model finds small objects in dense images
In efforts to automatically capture important data from scientific papers, computer scientists at the National Institute of Standards and Technology (NIST) have developed a method that can accurately detect small, geometric objects such as triangles within dense, low-quality plots contained in image data.

Get excited by neural networks
Scientists at The University of Tokyo introduced a new method for inferring the energy of the excited states of electrons in materials using machine learning.

Performing optical logic operations by a diffractive neural network
Optical logic operations, as the basis of optical computing, hold huge potentials to many applications such as cryptographically secured wireless communication and real-time wavefront-shaping.

Computational imaging benefits from untrained neural network
In a recent study, investigators from the Chinese Academy of Sciences described how they combined an untrained neural network and physics knowledge to eliminate the limitations of deep-learning-based CI methods.

Neural hardware for image recognition in nanoseconds
Usually, artificial intelligence is based on software. Scientists at TU Wien (Vienna) created intelligent hardware, which is much faster.

The neural basis of sensory hypersensitivity
A study from MIT and Brown University reveals a neural circuit that appears to underlie sensory hypersensitivity in a mouse model of autism, offering a possible strategy for developing new treatments.

Read More: Neural Network News and Neural Network Current Events is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to