[nfb-talk] The Google Car, what do you think?

Wed Oct 20 18:35:04 UTC 2010

That is exactly what I meant -- of course it is obvious that if the computer 
could recognize the woman with the baby, they would be assigned a higher 
score than a garbage can -- but image recognition is a complex process and 
even humans make mistakes in recognition.
But back to the car, it would also need to have separate processors doing 
image recognition at multiple points of the car -- the rear view mirrors, 
the front and rear and sides.  In a sense, travelling driverless might be 
safer if the image recognition were perfected, since you could have it 
processing its surroundings more closely and from more vantage points than a 
driver with a rear view mirror.  It also wouldn't be distracted by the 
passengers or the cell phone and would be alert, unlike a sleepy or drunk 
driver.
Interesting.
--le

----- Original Message ----- 
From: "John Heim" <john at johnheim.net>
To: "qubit" <lauraeaves at yahoo.com>; "NFB Talk Mailing List" 
<nfb-talk at nfbnet.org>
Sent: Wednesday, October 20, 2010 8:46 AM
Subject: Re: [nfb-talk] The Google Car, what do you think?

Yes, a computer  can be programmed to make value judgements .  In fact,
that's the easy part.  You create a list of everything you could possibly
run into and assign them a score. A woman carrying a baby would have a
pretty high score. A garbage can would have a very low score. The car simply
runs into the thing with the lowest score. The car would make that decision
quicker and be more likely to make the correct decision than a human.

The hard part is making it so a car can tell the difference between a woman
carrying a baby and a garbage can.

----- Original Message ----- 
From: "qubit" <lauraeaves at yahoo.com>
To: "NFB Talk Mailing List" <nfb-talk at nfbnet.org>
Sent: Tuesday, October 19, 2010 3:30 PM
Subject: Re: [nfb-talk] The Google Car, what do you think?

> can a computer be programmed to make value judgements like a human, for
> example, swerving to collide with a dog or phone pole rather than a child
> when those are the only choices?
> The visual recognition of what is around the car would have to be pretty
> sophisticated.
> Note that I'm not being a "naysayer", I'm just pointing out there is more
> to
> driving than making one's way through an obstacle course.
> I am doubtful this technology will be ready for common use within the next
> few years, and even when the powers that be judge it is ready, you have
> the
> problem of the inevitable mixing of human drivers in old fashioned cars
> and
> auto driving machines.  This I think will make the road a scarier place
> than
> it is now.
> --le
>
>
> ----- Original Message ----- 
> From: "John Heim" <john at johnheim.net>
> To: "NFB Talk Mailing List" <nfb-talk at nfbnet.org>
> Sent: Tuesday, October 19, 2010 8:48 AM
> Subject: Re: [nfb-talk] The Google Car, what do you think?
>
>
> Well, your point is the same thing I said when I first heard about the
> Blind
> Driver Challenge. Folks with a lot more resources than the NFB are pushing
> the technology in a different direction. But to be fair to the NFB, there
> is
> a significant difference between a car that drives itself and a blind
> driver. The two technologies are complimentary. Well, we can hope they
> will
> be.  There are actually 3 different technologies converging here, drive by
> wire, driverless technology, and blind driver technology.
>
> Jets are already equipped with something called "fly by wire". A jet pilot
> does not directly control a plane these days. He's essentially operating a
> computer program which in turn controls the moving parts of the plane.
> This
> kind of technology will almost certainly be integrated more and more into
> cars in the near future.
>
> What google is doing is something slightly different. In the technology
> google is developing, the computer is making the decisions. But drive by
> wire and driverless technology can work together by having the computer
> correct mistakes by the human or taking over in emergencies. For example,
> the driverless technology might correct a driver's mistake by preventing
> them from shortening a right turn and having the back wheels go up on the
> curb. Or it might apply the breaks if a pedestrian steps out in front of
> the
> car. Sooner than most people think, driverless technology will be required
> in caars the way seat belts and airbags are today.
>
> The way the NFB's blind driver technology fits in is that it would allow a
> blind person to do the limited things a sighted person will have to do
> once
> drive by wire and driverless technologies become commonplace. I have
> little
> doubt that it will be practical for a blind person to drive a car in the
> fairly near future. Fairly soon, it will be safer for a blind person to
> drive a car than it is for a sighted person to drive a car today. Of
> course,
> that will be mainly because the computer is mainly driving the car instead
> of a human. Its an open question as to how society will react to that. I
> could imagine there will be quite a bit of resistance to blind drivers
> even
> if it is fairly safe by today's standards.
>
> ----- Original Message ----- 
> From: "Sherri" <flmom2006 at gmail.com>
> To: "NFB Talk Mailing List" <nfb-talk at nfbnet.org>
> Sent: Monday, October 18, 2010 4:32 PM
> Subject: Re: [nfb-talk] The Google Car, what do you think?
>
>
>>I have people in my life who drive. I know how much they have to
>>concentrate on to get from point A to point B. I just don't understand how
>>tactile and auditory feedback is going to help us focus on maybe 15
>>different things at once. A car that is programmed and takes you from
>>point
>>A to point B makes a lot more sense to me. It is a car that could appeal
>>to
>>all! Again, I think working with Google would be a real plus for NFB. But
>>what do I know and if being first is really the most important thing then
>>....
>> ----- Original Message ----- 
>> From: "Bryan Schulz" <b.schulz at sbcglobal.net>
>> To: "NFB Talk Mailing List" <nfb-talk at nfbnet.org>
>> Sent: Monday, October 18, 2010 5:24 PM
>> Subject: Re: [nfb-talk] The Google Car, what do you think?
>>
>>
>>> because the nfb wants to be first and wants the credit
>>> Bryan Schulz
>>>
>>> ----- Original Message ----- 
>>> From: "Sherri" <flmom2006 at gmail.com>
>>> To: "NFB Talk Mailing List" <nfb-talk at nfbnet.org>
>>> Sent: Monday, October 18, 2010 3:56 PM
>>> Subject: [nfb-talk] The Google Car, what do you think?
>>>
>>>
>>>>I really think this is more what we are looking for. This is a car
>>>>"anyone" could "drive". Google has lots of resources. Why don't we jump
>>>>on their bandwagon?
>>>>
>>>> Sherri
>>>>
>>>> Google is testing cars that drive themselves  Google announced Sunday
>>>>   that it has developed cars that drive themselves
>>>> automatically in traffic, and that it has been testing them on the
>>>> streets of California for months. It might seem like an unusual project
>>>> for Google, but it  could actually have big benefits.
>>>> We're not just talking about cars running Google Android.
>>>> This is the stuff of science fiction. The only accident that has
>>>> occurred so far: One of the cars was rear-ended by a driver at a stop
>>>> light. Human error!
>>>>
>>>> The vehicles have been tested on 140,000 miles of California road, from
>>>> Silicon Valley to Santa Monica.
>>>>
>>>> Each car is manned during the tests. One person sits in the driver's
>>>> seat, ready to take control of the vehicle instantly by grabbing the
>>>> wheel or touch the brake should something go wrong with the system. The
>>>> person in the
>>>> passenger's seat is an engineer who monitors the software operations on
>>>> a
>>>> computer.
>>>>
>>>> Google (Google) hired engineers who previously participated in
>>>> competitions and races involving automated cars -- important turning
>>>> points in the development of the technology, which has been coming into
>>>> its own since around 2005 according to The New York Times.
>>>>
>>>> If your first concern is one of safety, Google would argue that you're
>>>> going about it all wrong.
>>>>
>>>> Safety is one of the the project's purposes. Google believes that the
>>>> technology could nearly half the number of automobile-related deaths
>>>> because computers are supposedly
>>>> better  at driving than humans in the right circumstances.
>>>>
>>>> There are other hypothetical pluses, too. The vehicles' instant
>>>> reaction
>>>> time and 360-degree awareness would allow them to drive closer together
>>>> on the highway than humans can, reducing traffic congestion. They could
>>>> be more
>>>> careful when operating the gas, reducing fuel consumption.
>>>>
>>>> But the biggest benefit for Google would be the hour or so of daily
>>>> commute time the car owner would save. Instead of driving, he or she
>>>> could either be productive or entertained in the vehicle, doing work on
>>>> a wireless
>>>> Internet (Internet) connection or watching television.
>>>>
>>>> Google doesn't say it explicitly, but TechCrunch was
>>>> quick to note that this time could be spent using Google products and
>>>> absorbing
>>>> Google-run advertising.
>>>>
>>>> The most optimistic projections put this technology at least eight
>>>> years
>>>> away from market, though. Legal hassles are among the myriad problems;
>>>> all of the current traffic laws assume that a human driver is present
>>>> in
>>>> the vehicle
>>>> _______________________________________________
>>>> nfb-talk mailing list
>>>> nfb-talk at nfbnet.org
>>>> http://www.nfbnet.org/mailman/listinfo/nfb-talk_nfbnet.org
>>>
>>>
>>> _______________________________________________
>>> nfb-talk mailing list
>>> nfb-talk at nfbnet.org
>>> http://www.nfbnet.org/mailman/listinfo/nfb-talk_nfbnet.org
>>
>>
>> _______________________________________________
>> nfb-talk mailing list
>> nfb-talk at nfbnet.org
>> http://www.nfbnet.org/mailman/listinfo/nfb-talk_nfbnet.org
>>
>
>
> _______________________________________________
> nfb-talk mailing list
> nfb-talk at nfbnet.org
> http://www.nfbnet.org/mailman/listinfo/nfb-talk_nfbnet.org
>
>
> _______________________________________________
> nfb-talk mailing list
> nfb-talk at nfbnet.org
> http://www.nfbnet.org/mailman/listinfo/nfb-talk_nfbnet.org
>