By 2030, commercial passengers will routinely fly in pilotlessplanes.

Peter Duniho wrote:
"George Patterson" wrote in message
news:mle_e.11361$L15.4226@trndny01...
I agree that a computer can do a great job when everything goes more or
less according to plan, but what about when it doesn't?

Actually, a computer can do a great job of anything you can think of. It
has a problem if something comes up that nobody thought of

The real question is whether pilots on average are able to come up with
inspired solutions to problems more often than they create problems with
perfectly good airplanes.

Looking at major air accidents in the US over the past 5 years I'd say
humans are doing awfully well. Aside from the AA airbus right after
9/11 (which has lots of question marks) it's not at all clear to me
that well trained pilots in modern airliners don't save more than they
cause. A fairly large chunk of Part 121 accidents involve maintenance
or systemic causes that a computer pilot would not presumably make any
difference with.

OTOH, fully-automated aircraft would probably make a huge difference
for GA safety, where pilot failure is the primary cause of accidents.

This is the same reason that autopilot cars are a good idea, no matter how
offensive they may seem to some people. Yes, there will be failures of the
equipment. But that will happen MUCH less often than the failures of the
humans, and will improve the reliability and efficiency of our
transportation infrastructure at the same time.

Look at this for an idea of the state-of-the-art in robot cars. It's
pretty pathetic.

http://en.wikipedia.org/wiki/2004_DARPA_Grand_Challenge

OTOH, ABS and stability control, etc. have unquestionabaly made driving
much safer. Some high-end cars use forward-looking radar to sound an
alarm if you start closing in on the car ahead of you very quickly and
even cruise control which maintains a following distance rather than
fixed speed. Presumably this trend will continue much as an Airbus
today is a largely automated plane but with big decisions still made by
pilots.

-cwk.

wrote in message
oups.com...
Looking at major air accidents in the US over the past 5 years I'd say
humans are doing awfully well.

I'm not talking about restricting one's view to "major air accidents". In
any case, if you have actual statistics to refute my intuition, I'm all
ears. Otherwise, your intuitive view is no more compelling than my own (no
less either, granted).

[...]
Look at this for an idea of the state-of-the-art in robot cars. It's
pretty pathetic.

http://en.wikipedia.org/wiki/2004_DARPA_Grand_Challenge

The DARPA event is a completely different scenario from a general
autopiloted transportation infrastructure. For you to use it as a
comparison is laughable. Instead, try the many successful demonstrations of
computer-driven cars on paved roadways with appropriate guidance technology.

Pete

Peter Duniho wrote:
wrote in message
oups.com...
Looking at major air accidents in the US over the past 5 years I'd say
humans are doing awfully well.

I'm not talking about restricting one's view to "major air accidents". In
any case, if you have actual statistics to refute my intuition, I'm all
ears. Otherwise, your intuitive view is no more compelling than my own (no
less either, granted).

I assumed Part 121 and did mention that it would clearly be an
improvement for GA. But, let's look at just 121 for a minute:

Searching for domestic accidents since 1/1/2000, Part 121, with
fatalities, I find 14 NTSB records, 4 of which are from 9/11. The other
10:

6/05: Belt loader truck crash kills driver
10/04: 13 pax killed on a regional crash during approach in IMC.
Awaiting final report.
8/04: Convair 580 (freight) crash on approach kills 1 of 2 crew,
awaiting final report
9/03: Tug driver crashes into DC-9, is killed
1/03: US Air Beech 1900 crashes in Charlotte, 21 dead, maintenance
error
11/01: AA Airbus 300 crash due to rudder failure, pilot error (insert
alt. theory here)
8/01: Ramp agent walks into propeller, 1 dead
11/00: 1 FA killed when cabin door opened on ground before
depressurizing during evacuation, FA opened door and was blown out and
fell to the ground
2/00: Emery DC-8 lost with all crew (3) due to "A loss of pitch control
resulting from the disconnection of the right elevator control tab. The
disconnection was caused by the failure to properly secure and inspect
the attachment bolt"
1/00: Alaska Airlines MD-83 lost with all on board (88) after
stabilizer trim problem caused by improper maintenance

This leaves us with 6 actual aviation accidents, 3 of which are due to
mechanical/maintenance issues. 2 are awaiting the final report, but
pilot error looks like a safe bet. The AA crash is open-and-shut except
that I recall some debate that the pilot was in fact following the book
as written by AA. Still, I'll give that one to the computer.

What this analysis doesn't include is how many non-accidents we had due
to humans acting intelligently and non-computerish. Also, others might
protest that restricting this to US accidents in the past 5 years (an
unprecedentedly safe period) is cherry-picking my data. Fair 'nuff.
Still, it suggests that human flight crews properly trained can achieve
extremely high levels of safety.

[...]
Look at this for an idea of the state-of-the-art in robot cars. It's
pretty pathetic.

http://en.wikipedia.org/wiki/2004_DARPA_Grand_Challenge

The DARPA event is a completely different scenario from a general
autopiloted transportation infrastructure. For you to use it as a
comparison is laughable. Instead, try the many successful demonstrations of
computer-driven cars on paved roadways with appropriate guidance technology.

I gave a link for my argument. Now you try.

At best the things you speak of are capable of handling traffic flow on
the Interstate, and could make a difference. Preventing rear-endings,
lane drift, asleep at the wheel would be good. Of course, we could get
most of this benefit a lot more cheaply if we assumed the human was
still in control. Radar could be used to warn of cars slowing ahead,
and a guidance stripe painted on the highway could be used to provide
directional "assistance" and to alarm for instance if you started to
drift off the centerline without using your turn signal. (Ha! What
chaos that would cause in Boston...) This is something we could roughly
do with today's technology and automobiles and would not cost a
gogoobillion dollars to rewire our highways.

Once you get off the highway, the problem becomes pretty gnarly what
with pedestrians, interchanges of every kind, etc. Don't forget
generational problems where you have autopilot and non-autopilot
vehicles. We're having a hard enough time switching to HDTV so don't
try to tell me this would be straightforward. Actually, aerial
navigation is a much simpler problem. In any case, this just
underscores my point that "assistance" systems are a far cheaper and
more effective path to enhanced safety.

-cwk.

wrote in message
oups.com...

Peter Duniho wrote:
"George Patterson" wrote in message
news:mle_e.11361$L15.4226@trndny01...
I agree that a computer can do a great job when everything goes more
or
less according to plan, but what about when it doesn't?

Actually, a computer can do a great job of anything you can think of.
It
has a problem if something comes up that nobody thought of


How about:

Pilots Battle Computer For Control Of 777

Stanley Kubrick couldn't have scripted anything more eerie than the
real-life odyssey of a Malaysian Airlines Boeing 777 that seemed hell-bent
on crashing itself on a trip from Perth to Kuala Lumpur last Aug. 1.
According to The Australian newspaper, the Malaysian flight crew had to
literally battle for control of the aircraft after something went wonky with
the computerized controls. The plane was about an hour into the flight when
it suddenly climbed 3,000 feet and almost stalled. The Australian Transport
Safety Bureau
http://www.atsb.gov.au/aviation/occurs/occurs_detail.cfm?ID=767 report
said the pilot was able to disconnect the autopilot and lower the nose to
prevent the stall but the autothrottles refused to disengage and when the
nose pitched down they increased power. Even pushing the throttles to idle
didn't deter the silicon brains and the plane pitched up again and climbed
2,000 feet the second time. The pilot was able to fly manually back to Perth
but the autothrottles wouldn't turn off. As he was landing, the primary
flight display gave a false low airspeed warning and the throttles
firewalled again. The display also warned of a non-existent wind shear.
Boeing spokesman Ken Morton said it was the only such problem ever
experienced on the 777 but airlines have been told via an emergency AD
http://www.airweb.faa.gov/Regulatory_and_Guidance_Library/rgad.nsf/0/25F9233FE09B613F8625706C005D0C53?OpenDocument
to load an earlier software version just in case. The investigation is
focusing on the air data inertial data reference unit (HAL for short?),
which apparently supplied false acceleration figures to the primary flight
computer.

Pilots Battle Computer For Control Of 777

I also remember when it took two people to fire up the Tandem mainframe
computer. Today PCs are more powerful than that old tandem. Don't fall
into the belief that things can't change a lot in 25 years.

-Robert

Peter Duniho wrote:

I admit, I don't have the statistics in front of me, but I suspect that
human error in the cockpit causes more accidents than human novelty recovers
from.

I doubt that anyone has good statistics. People investigating a accident in
which the pilots don't survive are (or at least were) likely to declare it
"pilot error" anytime they couldn't figure out what went wrong. And if the pilot
survives, he's probably going to try very hard to hide any mistakes he might
have made.

There's also the tendency of the NTSB to blame the pilot for *something*, even
if the basic cause was beyond anyone's control. If the engine fell off, one
"cause" of the accident is likely to be "failure to maintain adequate clearance
from terrain."

George Patterson
Give a person a fish and you feed him for a day; teach a person to
use the Internet and he won't bother you for weeks.

Peter Duniho wrote:
"George Patterson" wrote in message
news:mle_e.11361$L15.4226@trndny01...
I agree that a computer can do a great job when everything goes more or
less according to plan, but what about when it doesn't?

Actually, a computer can do a great job of anything you can think of. It
has a problem if something comes up that nobody thought of

The real question is whether pilots on average are able to come up with
inspired solutions to problems more often than they create problems with
perfectly good airplanes.

I admit, I don't have the statistics in front of me, but I suspect that
human error in the cockpit causes more accidents than human novelty recovers
from.

This is the same reason that autopilot cars are a good idea, no matter how
offensive they may seem to some people. Yes, there will be failures of the
equipment. But that will happen MUCH less often than the failures of the
humans, and will improve the reliability and efficiency of our
transportation infrastructure at the same time.

The trouble is that you never hear of the thousands of 'pilot skill'
saves a year.
And in an accident the first claim by the accident inspectors is that
it's 'pilot error' and, sadly, they can maintain that position in
spite of other factors.

"george" wrote in message
oups.com...
The trouble is that you never hear of the thousands of 'pilot skill'
saves a year.

You also never hear of the thousands of "pilot skill" failures that require
"pilot skill" saves, either. So what?

And in an accident the first claim by the accident inspectors is that
it's 'pilot error' and, sadly, they can maintain that position in
spite of other factors.

Yes, it IS unfortunate that so many accidents turn out to be attributable to
"pilot error", and that in spite of other factors, the inspectors CAN still
attribute the accidents to "pilot error". Seems to me you're just making
the point that more automation would be good.

Pete

In article ,
"Peter Duniho" wrote:

And in an accident the first claim by the accident inspectors is that
it's 'pilot error' and, sadly, they can maintain that position in
spite of other factors.

Yes, it IS unfortunate that so many accidents turn out to be attributable to
"pilot error", and that in spite of other factors, the inspectors CAN still
attribute the accidents to "pilot error". Seems to me you're just making
the point that more automation would be good.

That is not at all what George said.

--
Bob Noel
no one likes an educated mule

In article , Peter Duniho
wrote:

The trouble is that you never hear of the thousands of 'pilot skill'
saves a year.

You also never hear of the thousands of "pilot skill" failures that require
"pilot skill" saves, either. So what?

I'll give you an example: We had an electrical short a few months ago,
causing smoke in the cockpit and cabin. First checklist item for us,
after putting the oxygen masks, is to shut off all electric power.

Had that been a "pilotless airliner," you *couldn't* shut off all
electric power, and the wire would have continued to burn. I doubt it
would have been as uneventful as it turned out with humans at the
controls.