[ExI] self driving cars

Kelly Anderson kellycoinguy at gmail.com
Thu May 17 04:57:46 UTC 2012


On Wed, May 16, 2012 at 10:31 PM, spike <spike66 at att.net> wrote:
>>... On Behalf Of Kelly Anderson
> Subject: Re: [ExI] self driving cars
>
> On Wed, May 16, 2012 at 7:26 PM, Mike Dougherty <msd001 at gmail.com> wrote:
>>> I think spike was suggesting the speeds go down for engineering reasons
> (mostly fuel economy)
>
>>...Economics will push speeds faster if safety isn't an issue. Unless Al
> Gore is in the white house, or OPEC starts throwing their weight around ala
> 1973, I can't see a speed limit for fuel economy because the overall
> economics is that the time of the car's occupant is more valuable to the
> overall economy than the gasoline.  -Kelly
>
> _______________________________________________
>
> I do mean we have plenty of applications where slower speeds will not hurt
> us much.  Plenty of us have shorty commutes, such as my own, less than ten
> miles.  We can imagine software guided cars that have no need to go faster
> than the speed limit, and if so, they might not have the ability to go
> faster.  If so, the potential weight savings and increased fuel economy is
> astonishing.  Assume a car that only needs to go the local top speed limit,
> and give it 15 seconds to get to that speed.  Work backwards from just those
> requirements, and assume carrying only two apes.  See how light this vehicle
> can be?

Perhaps you can pay to go faster if you want to. Autonomous vehicles
should be better at getting out of the way of people that want to go
faster... there are legitimate reasons to hurry sometimes.

On the other hand, delivering stuff in trucks could benefit
significantly from slower speeds, going 24 hours a day, they would
still get there faster than they do now. The balance there is the cost
of gasoline vs. the cost of owning the truck per hour. I suspect that
going slower would make some sense there.

> But your comments and Mike's gave me an idea.  My intuition and every
> instinct says that if any two or more devices have motors, there is some law
> of nature which requires they must be raced against each other.  I can see
> this as a terrific new breakthrough sport: robot racing.

I'm a big fan of robotic fights to the death... :-)  I would really
love to see autonomous fighting to the death some day. I have a great
design for same.

> I grew up an hour drive south of Daytona International Speedway, and I
> loooved car racing as a kid, motorcycle racing even more, but if you look at
> the cars now vs then, the sport hasn't changed much in the last 40 yrs.
> Evolutionary changes have pushed the cars faster, sure, but from the
> spectator's point of view, not so different than it was in my own misspent
> youth.  I am past the half century mark now.

But would it be exciting to watch robots race? I dunno. It's the human
aspect that makes it interesting.

> I would buy tickets to watch several identical Priuses (Prii?) racing around
> a city-street track.  That would be a kick!  Several years ago, they made a
> loop around several downtown San Jose streets for formula 1 cars, which was
> a hell of a lotta fun, but it would be even more cool if they raced
> robo-cars.  Let the Prii be identical engine and drive train, but let the
> hackers mess with the software all they want, and the instrumentation.  It
> would be like any other car race except with this one oddball rule: if any
> car hits or even touches any other car, both cars are automatically
> disqualified regardless of who hit who.  That motivates the programmers to
> do something useful: avoid collisions and contact as top priority, higher
> than getting there first.

Racing has always been a great way to develop car technologies, so
maybe you have something here. I personally don't get off on Nascar...
but lots of people do, so maybe there is something to be gained by
implementing your suggestion. I can only imagine the outcry as the
first robotic car tries to get into a Nascar race.

> I was present at the first DARPA challenge, robo-cars in the desert, the one
> which was a total flop in 2004.

Sweet. Wish I had been there, I had to watch reruns on the Internet.

> That was a time trial, with cars starting
> every five minutes, the fastest cars going out first.  What I have in mind
> is a dozen cars starting on a city street loop of a couple miles.  Once the
> robo-cars demonstrated they are safe, we could introduce a humans vs
> software races.  Would that be wicked cool or what?
>
> When software demonstrates that it can race competently, it can be trusted
> to drive.

Works for me. But what happens to injured humans until then? :-)

Seriously, the greatest hurdles to face in autonomous vehicle land are
legal, not technical at this point. We haven't discussed those legal
issues at all, but you can imagine the class action lawsuits if people
die in significant numbers while the bugs are being worked out of the
system.

-Kelly




More information about the extropy-chat mailing list