Re: Re: Sydney Metro Partial Shutdown
  Matthew Geier


On 19/6/19 7:34 am, Prescott wrote:
> When I rode it immediately after riding in Perth with humans at the

> controls, I couldn't detect any quality difference in the "driving".

> Exactly the same.

>

In which case the Perth drivers need more simulator time.

If I drove a tram like the ATO drives the metro trains when I was
examined, id probably have been marked down with 'needs more tutoring'.

The M1 in Paris was doing the same thing, diving at speed towards the
'signal', braking, the train ahead thus got further away, so it would
power again, close the gap, brake, increase the gap, power again, etc.

Now human drivers who blindly follow the signal indications will do that
too - but some one who cares about their craft will adjust their speed
to keep a constant speed.

(I've watched tram drivers who obviously don't care about the 'craft'
wheel spin the TFS trams on starting EVERY time. They operate them like
they operate the Citadis, who's computer moderates the drivers inputs.
The TFS tram controller tries to follow that command to immediately go
to full power.)


Several engineers at work (robotics specialists, so know control
algorithms) commented on how the Metro train set charges up the grade at
full power, crests the grade, overshoots the speed limit, goes into full
electric service brake, undershoots, so powers again, over shoots,
brakes again, under shoots and it never stabilises before it brakes for
the station - which they do too early, but given the overshoot problems,
this isn't surprising. I suspect they may have the load sensors better
calibrated by now and over/undershoots are less common than when I rode it.

The technology they are using is mature. It should 'just work'.

The ATO should aspire to be a good driver, not a merely adequate one.
The technology certainly exists. I work in a building full of PHD
students who could implement it with ease.