Re: [aclug-L] Smart Cars

[Bob Deep <bobd@xxxxxxxxxxxx>]

Ahhh, but one slight rub.

The success of linux, is in some strang way related to the success of
Microsoft.  Without MS's serious control of the computing hardware
market, and the standardization of the Personal Computer hardware, the
bulk of linux's current users and developers would still be in the dark
ages (ie B.P.C).  Linux then benifits from a standard and stable
hardware platform that's cheep and reliable. (Yes, it runs on various
platforms, but the bulk of development is being done on PC's).  It is
the availablity of the hardware that makes linux as successful as it is,
and we owe a lot of thanks to the M$oft/Intel pair for the situation we
find our selves in.

We are not yet close to having the same situation for "automatic"
driving cars.

In order to develop the system you describe, you will need to develop
hardware systems to take care of the mechnical functions, as well as the
software to run the whole thing.  I see three major peices to be
developed...
        
1.  Electro/Mechnical interface which allows a car to be controlled via
some sort of digital interface.  You must steer, change speeds, apply
the breaks etc, all from some sort of digital interface. (Don't
underestimate the complexity of this!)

2.  Input and data collection interfaces... (Hey, how do you intend to
have the computer figure out how to drive? Video, Audio, Radar, etc?)

3.  The Software system to link the input (be it a video camera, and
other data sources) to the outputs to drive the car (and of course the
hardware platform this runs on).

Most of the things necessary for development of such a "automatic"
driving system do not yet exist in even hobyist form.  Ths cost of
development of such systems is high, and the avarage hobyist is not able
to spend such large sums, nor actually go out and install the stuff on
his families mode of transportation.

When the costs come down, look for development.  But I doubt it will
happen until then.

In the beginning, Software and Hardware where both very expensive to
obtain or develop.  Then with the advent of higher level languages and
compilers, software became cheeper and cheeper, initially much faster
than the hardware all these programs ran on.  With the standardization
of the PC over the last few years (say 10) and the massive jump in
performance, now hardware is *really* cheep... So cheep that computer
out sold TV's at Christmass.  Now, with linux, anybody can develop
software at home that would put most 10 year old programs to shame, all
on a $1000 computer!

Now for the cost of a computer and $2 for a CD you can have some of the
best development tools anywhere....  No wonder linux is so well
supported.

-= bob =-

"Wilner, Alden" wrote:
> 
> How far away from reality are smart cars? Technically, they're already here.
> Economically they're a long way off. And it has to do with the Linux
> philosophy.
> 
> Why isn't Linux accepted more readily at major corporations? I'll tell you
> what one company's answer is: There's no one to sue if things go horribly
> wrong. See, with Windows, you have this behemoth with piles of cash in
> Redmond, and if their mail server crashes and costs you a week's
> productivity, you can haul them into court and recover some of the money
> your company lost. (At least, that's the theory.  I've not heard of anyone
> actually _doing_ that).  With Linux there's no one to sue.  You could sue
> individual developers, but there's no way you're gonna get near the money
> you could from a giant corporation.
> 
> But one person's problem is another person's opportunity....
> 
> Given the tremendous safety improvements that could be realized with smart
> cars, why isn't more being done to develop them? Because everyone's afraid
> of getting sued. Existing smart cars are really pretty stupid.  They require
> an attentive driver behind the wheel at all times. The ideal smart car won't
> require any human intervention at all.  In the middle of the development
> curve will be cars that can almost drive themselves, but not quite. With
> these cars, the failure modes are likely to be spectacular, and dangerous.
> No corporation is going to risk the lawsuits that would result from such
> accidents.
> 
> Enter the Open Source Philosophy of Linux (et al).
> 
> Suppose a group of developers were to decide that massive improvements in
> highway safety and convenience were a laudable goal. Suppose these
> developers got together and started creating an open-source smart car.
> Suppose smart cars were developed, in effect, by hobbyists, not
> corporations.
> 
> In this environment, if a single individual has a catastrophic failure of
> their smart car, they are still liable for prosecution (assuming they
> survive!), but there's no corporation to attack. The liability is strictly
> contained. Development can proceed and improvements can continue to be made.
> And eventually smart cars will become so smart that nobody will be able to
> deny that they are better drivers than the humans inside them, and they will
> become, not just accepted, but commonplace, and perhaps even required.
> 
> Consider the case of the stereotypical drunk driver. He has been out
> partying until the wee hours of the morning, and must now find his way home
> through dark, empty streets, probably on surface roads the whole way, with
> speed limits never in excess of 30 mph, etc. etc. This is the perfect
> situation for smart car development. If our smart car drives home at 20 mph
> in something approaching a straight line, it will be at least as good as the
> drunk driver behind the wheel. And, unlike the drunk driver, our smart car
> is extremely unlikely to get overconfident and start driving at 60 mph. Now,
> to me that means that smart cars could start saving lives inside a year,
> with enough hardware & software developers.
>
---
This is the Air Capital Linux Users Group discussion list.  If you
want to unsubscribe, send the word "unsubscribe" to
aclug-L-request@xxxxxxxxxxxx.  If you want to post to the list, send your
message to aclug-L@xxxxxxxxxxxx.