"Jevan" <jp@jevan1.com> wrote in message

"Harry Conover" <hhc314@yahoo.com> wrote in message

While working at General Railway Signal Company, I was the >senior
system engineer in charge of control system design for >Washington's
WMATA rail system, so I'm still pretty familiar with the technical
details of its operation. I also performed the same function for
Atlanta's MARTA system and parts of the Frisco's mainline rail system.

That's cool! I never knew what your job was. Neat.

I wonder what went wrong in the train collision situations. Seen on the

in the last few years - incidents of one train immobile while second

runs into it.

In one situation, the people on the immobile train could see the moving
train coming toward it from afar, but couldn't understand why it kept

toward them. Eventually the became alarmed and stood up (but hadn't

off the train). Also this worsened injuries because then you had

people during the collision (apparently worse than if they were

they had known it was going to collide, they actually had plenty of time

jump off their immobile train, but they hadn't thought that this would
happen (I suppose once it got too close they must have known).

The psychology is easy to understand, at least. Jumping off a train
because a train coming from afar doesn't seem likely to stop, although
it still has plenty of space to do so, is unpleasant, embarassing, and
dangerous (you could break your leg). In fact, if a significant body
of passengers had jumped off the stationary train (following a few
souls less motivated by conformity, a "panic"), resulting in avoidable
injuries, and the second train _had_ stopped in plenty of time, then
we might focus on the identical passenger behavior receiving identical
inputs as sub-optimal, irrational, and so forth.

In truth all complex, dangerous and largely unprecedented situations
are fraught with potential false positives and negatives. In fact, I
sometimes find myself thinking of "what-if" scenarios in daily life,
which would present me with the same choice between potentially
embarassing and seemingly irrational flight in the case of a false
positive, and standing in the face of oncoming danger: a person who
continually jumps off stationary trains in the face of potential but
largely unrealized collisions accrues negative consequences.

As you have correctly stated, this system requires only
that a train operator depress his "ATO Start" button at the begining
of a run, then simply sit back and enjoy the trip

But you wonder because presumably then, maybe not all countries track

their trains positions. One would assume that a train in the system you
described, would still have its position tracked, even if the "ATO

button had not yet been pressed. Say for example if something failed on

railway track causing it to stop and then it deactivated overnight, and
another train came along then the next day.

Ohmygod ... that sounds frighteningly plausible. You mean the system
tracked only active trains, but not inactivated trains which became
immobile hazards? Tsk, tsk, tsk. It presumably had no capability to
deal with sleeping cows either: other than a fence.

The cute old "cow-catchers" must have made a major amount of messy
road kill, and a boon for coyotes.

Maybe there could be a simple & inexpensive technology you can sell to

countries, to install in trains to avoid such situation of a collision

trains. Even a simple distance sensor (laser position sensor) or

that point out in front & behind the train, to stop the train if it see
something ahead in front (with sufficient time to stop) or warn if

behind it getting too close or accelerating toward it from afar. Has to
handle bends in the tracks, but presumably curvature must be limited

in order for the train to be able to attain speed?...

Is it possible this visual recognition of obstacles was the function
of the pinocle playing train operator? It's hard to maintain
alertness in jobs where 99.99% of the time alertness is useless
overhead, but it can be done: failure to do so amounts to a management
failure ... whatever the invdividual culpability of the employee ...
because management is supposed to have the experience to know that
0.01% will eventually occur with certainty, and devise drill systems
to raise the rate of effective positives.

But as for cars - yes, well, if something is possible (which it is),

will do it (fully computer control of car, with sensing/ awareness for
things like road signs/ pedestrians/ etc). Indeed it is true now cheap

cameras are available and the only real expense is development cost, but

research organisations they may already have need for projects anyway

government funded, presumably). I suppose, major car manufacturer could

such projects. I agree though, got to have a good implementation. No
bugs!! - or at least, must be failsafe.

Causing the car to "stop" is not really enough, if something goes worng,

it to be failsafe. Well, if the car in front brakes, because their

video gets clogged up with dirt and it can't see properly, that could

an accident too. I suppose you'd also need redundant mechanisms that
automatically take over if need be (for example if you have three

if one doesn't match the other two then the nonmatching one gets shut

Aircraft already have redundant systems, apparently (for example).

No system is completely failsafe ... you can only multiply redundancy
to the point of diminishing returns, the criteria for which will
always be second guessed by bureaucrats, law makers, and tort lawyers.

Any major technology shift also faces an unfairly skewed awareness of
its fatalities -- 30,000 killed by people? No problem. God help us
when the first ten are killed by computer failure.

Incidently, I think one of the big challenges of this technology is
the determination in real time of the characteristics of the tire-road
interface, i.e. how does the system know if it's on a dry road or
slick ice?

ABS systems do this but you have to be braking at the time. Is
there technology to measure this continuously?