Originally Posted by TopTup
In the words of Mr Earl Weiner [sic] (late 80's, early 90's pioneer of CRM), "Automation is dutiful yet dumb". Meaning if programmed incorrectly that same automation will quite happily and extremely successfully plough that aircraft into the side of a mountain. So, give me a pilot who has the knowledge, experience, hours, and training to know when the automation has failed the crew by either human error or software/hardware error....and then to competently know how to handle the situation to achieve the most successful possible outcome.
The answer to Prof. Wiener's quote today would be "that's so late 80's, early 90's".
Recall the first FBW commercial airliner came into service in 1988, not yet thirty years after jets were introduced to airline flying, and we are 22+years on from that.
There is no doubt that the QF 32 incident, with which this thread began, was handled in part by superior airmanship. There is equally no doubt, if you read
the RAeS interview with David Evans, the check-check captain on the flight, that this superior airmanship was aided incomparably by the automation, which amongst other things gave the crew a detailed list of all that was awry, which they used to significant effect, in particular to calculate landing performance.
The USAF is installing a terrain-collision avoidance system in its fighter aircraft, and the people who developed that system are short listed for an Aviation Week Laureate Award this year. This was first mooted nearly thirty years ago, but it was clear then that knowledge of automation was not up to the task. Now, the USAF agrees it is.
Many of the significant airline accidents in 2010 could have been avoided had the standard manoeuvres which were being performed been automated, and the proverbial dog had bitten the pilots had they touched anything, as I argued in
this blog post in September 2010, which was itself written to address a similar issue raised in another thread on this forum.
I think there is no doubt that automation will take up many more of the routine as well as non-routine tasks in flying a commercial aircraft. As we see these systems becoming more reliable in a wider variety of flight situations, we will see the requirements for successfully monitoring them correspondingly reduced. To put it bluntly, 200 hours programming a veridical simulator will be preferred over 1500 hours teaching people to fly C152s by most airlines, for good reasons. Not tomorrow necessarily, but in ten to twenty years.
PBL