[OT] Hit and Run

Nicholas Leippe nick at leippe.com
Thu Aug 23 11:53:24 MDT 2007


On Thursday 23 August 2007, Shane Hathaway wrote:
> Although it works smoothly most of the time, the CA system still can't
> avoid the fact that humans have a minimum response time of nearly 1
> second, and that's only when they're very alert.  However, if we could
> wirelessly and securely transmit a brake signal between cars, we could
> fix that hole and the freeways could behave even more like trains.  I'm
> not talking about cars that drive themselves, I'm just talking about a
> single enhancement to human response time that could save lives in
> congested areas.
>
> I've been tempted to set up computer simulations of such a system.  The
> simulation would include many kinds of drivers.  The simulation might
> even be complex enough to require a serious Linux cluster.  (Ha!  I'm on
> topic!)

Just a few years ago I read an article saying they were testing some automated 
cars down near LA. They were applying similar ideas--that if cars would 
communicate to each other they could manage their distance far more 
accurately than a human--they were talking just feet apart at highway speeds.

You can easily see how it could produce a more efficient system--far more 
throughput. It would also get rid of the transient waves or 'bubbles' that 
you observe hours after an incident. The problem is that drivers care about 
their individual latency ('late' pun not intended, but fully applicable) 
rather than the net throughput of the roadway system.  But it really is just 
a packet network, but where every packet is demanding QoS.




More information about the PLUG mailing list