DPRG
DPRG List  



[DPRG] harder than we think

Subject: [DPRG] harder than we think
From: David Anderson davida at smu.edu
Date: Mon Oct 27 13:14:16 CDT 2014

Ah Paul, good reply.  And spoken like someone who might have a vested 
interest (wink-wink-nudge-nudge!)

And as you say, all well and good from the vantage of anyone actually in 
the field.   But that's not the popular perception, the one that is 
being sold to the driving public, to investors, regulators, and gov't 
officials.   For example, check out this article just published yesterday:

<http://www.fool.com/investing/general/2014/10/26/self-driving-cars-what-will-it-take-to-make-automa.aspx>

It's an article about all the annoying legal barriers to self-driving 
cars.  Here's an excerpt from the first paragraph:

"If Google can drive a car without human intervention /today/, then why 
should we have to wait another nine years before these vehicles hit the 
mainstream?"

See?  We already have self-driving cars!  Not just fancy cruise control 
or radar-activated braking.  Not just human-assisted steering.  Cars 
that can drive *"**without human intervention." ///T//oday/!
*

Both the articles linked in the first post make the same point, with 
which I think you agree.  We are far, far away from these goals 
(self-driving cars, strong artificial intelligence) that appear to 
casual observers to be so tantalizingly close.  My point as well.

cheers!
dpa


On 10/26/2014 12:50 PM, Paul Bouchier wrote:
> All very true David, and problems anyone in the field is acutely aware
> of. Nevertheless, I think it's worth drawing a distinction between fully
> autonomous driving as defined in the article where even a steering wheel
> is unnecessary, and the many possible levels of assisted driving.
>
> By the standard defined in this article, the humble wall thermostat is a
> failure, because it doesn't know if the door is left open, or if I feel
> too cold for whatever reason. The end goal: comfort control - is not
> fully automatic and sometimes I have to turn the thermostat off or up or
> down. Yet despite this failing, it does a good job 99% of the time and
> I'm glad enough to have it that I use it.
>
> I expect autonomous driving to exist in this middle state for a very
> long time. Increasing levels of autonomy and capability will become
> available, but possibly never 100% for all the reasons stated.
>
> My own personal vision: the autonomous UPS truck that drives up to your
> house and unloads a little robot that brings the package the last 50
> feet to your door, will always, I expect, need banks of teleoperators in
> a far-away place ready to step in and exercise human judgement and
> teloperate the system as needed. A simple example of the need: someone
> parks right behind it and boxes it in so it can't move. But even with
> such limitations, such a system could bring substantial economic benefit
> to its owner.
>
> In this kind of a world, the most valuable attribute of an autonomous
> vehicle is recognizing situations it can't handle, making itself safe
> (stopping), and calling for help. We may well see the day when an
> autonomous vehicle stops in the middle of I35 because it thinks a
> tumbleweed might be a rock, and we have to wait behind it for a
> teloperator to take over and drive it over the tumbleweed.
>
> Nice to get grounded in reality once in a while though.
>
> Regards
>
> Paul
>
> On 10/23/2014 10:47 AM, David Anderson wrote:
>> ...as most robot builders learn on their own:
>>
>> <http://www.slate.com/articles/technology/technology/2014/10/google_self_driving_car_it_may_never_actually_happen.single.html>
>>
>> among other reasons:
>>
>> <http://spectrum.ieee.org/robotics/artificial-intelligence/machinelearning-maestro-michael-jordan-on-the-delusions-of-big-data-and-other-huge-engineering-efforts>
>>
>>
>> onward!
>> dpa
>>
>> _______________________________________________
>> DPRGlist mailing list
>> DPRGlist at dprg.org
>> http://list.dprg.org/mailman/listinfo/dprglist
>>
> _______________________________________________
> DPRGlist mailing list
> DPRGlist at dprg.org
> http://list.dprg.org/mailman/listinfo/dprglist

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.dprg.org/pipermail/dprglist/attachments/20141027/d55a8ee7/attachment.html 

More information about the DPRG mailing list