Re: robot race

On Thu, 27 Oct 2005 12:04:39 +0200, Borek
<borek@xxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:

>On Fri, 21 Oct 2005 15:13:07 +0200, <makc.the.great@xxxxxxxxx> wrote:
>
>>> The funny thing is that if ONLY robots are on the road,
>>
>> Imagine computer network with some standard protocol bots would be
>> required to log in, before start their engine. Then, bots send request
>> for the route over the network; powerfull mainframes deal that request,
>> shedule this bot route and everything to ensure no-traffic-jams
>> constraints, etc, and get back to our bot. So, typical sequence would
>
>That will mean (at least partially) that the 'intelligence' was moved
> from the bot to the supervisor. Still, it will work only as long as only
>bots are allowed to drive. That will be cool idea, as bots:
>
>- don't listen to the sport events on the radio
>- don't check their lipstick in the back mirror
>- don't drive drunk
>- don't quarrel with their second half before driving
>- don't get upset due to the presidency vote results
>- don't have a PMS

YET ... :-)

There are two problems with moving some of the navigation
tasks to some central supercomputer :

1) For military purposes a network link is both
insecure and it means the vehicle must emit
radio signals that can be detected and traced
by the foe. MIGHT be OK for autonomously piloting
civilian vehicles ... but there are still issues
about hackers, jammers and Big Brotherism.

2) Seems like MOST of the computing task has to do with
navigating obstacles and rapid-response is a must.
Therefore, it's unclear how much of the job COULD
be offloaded to a central computer - maybe not
enough to be worth it. Very general stuff like the
starting and destination points ... doesn't use
much memory or CPU time anyway.

Hmmm ... cockroaches can navigate just fine using a
handful of brain cells. How come we can't get digital
logic to even properly emulate a handful of brain cells ?

.