Originally Posted by
MountainBear
I think the trend towards automation will continue not because it's safer but because it's cheaper.
.....
The net effect is that "improvements" are more often than not simply rearranging the furniture. We might just be at that point in aviation today.
...
.
Disagree.
The next generation of flight control systems may not just be rigidly programmed, but will contain increasing amounts of machine-learning / pattern-recognition systems - although this technology is very much in it's infancy. There is probably scope in applying such systems to failure workload analysis and prioritisation, providing flight-crews with real-time assistance. For example, providing flight-crews with a situtation analysis and possible root causes with an indication of the degree of recognition between the observed inputs and the trained pattern:
e.g
95% match, given current ADC inputs, of an airspeed probe failure.
62% {some other related pattern}
At the very least such indications would take the flight crew direct from a collection of "low-level" warnings to a small number of possible explanations.
As a software engineer myself, I am amazed we don't see more accidents directly related to FBW; given the inherent complexity in building safety-critical software. The fact we don't is a credit to the specialists involved in these highly complex projects.
In terms of reliability and correctness of function the field is light years ahead of other applications of computer programming.
NASA - NASA Dryden Fact Sheet - Intelligent Flight Control System