Why German Gridguide is better than American Gridguide Options

[suoq]

http://www.tgdaily.com/trendwatch-features...reets-of-berlin

This could end sobriety as my gaming group knows it.

[Seerow]

I want one of these, but capable of operating anywhere not just in a specific city (IMG:style_emoticons/default/frown.gif)

[hobgoblin]

Funny how the requirements are more strict for the computer then for man, yet they are less likely to make mistakes once they have the skills in place.

[Seerow]

Funny how the requirements are more strict for the computer then for man, yet they are less likely to make mistakes once they have the skills in place.

It's because a human who makes a mistake is simply being human. A machine that makes a mistake leads to the company that made it getting sued for a lot more due to releasing a defective product.

[Paul]

If I thought Europe would take me I'd consider moving just for this. I hate driving. Unfortunately I am almost positive in a game of Red Rover I'd rank somewhere behind Carrot Top and Gary Busey.

[Draco18s]

Pfft, only 20km driven.

Try 140,000 miles.

[pbangarth]

Funny how the requirements are more strict for the computer then for man, yet they are less likely to make mistakes once they have the skills in place.

As yet humankind is more able to learn from mistakes. A guy makes a boo boo, he doesn't do it again (OK, OK, able to, not guaranteed to). A program makes a mistake and the program will repeat it ad infinitum, until the car gets back to the shop.

Two quotes from the Google article above:

"But of course, to be truly safer, the cars must be far more reliable than, say, today’s personal computers, which crash on occasion and are frequently infected. " Imagine what a virus could do on a busy highway.

"The car can be programmed for different driving personalities — from cautious, in which it is more likely to yield to another car, to aggressive, where it is more likely to go first. " The 'Transporter' virus, anyone?

[Sengir]

The German Gridguide system is called ALI, by the way (IMG:style_emoticons/default/wink.gif)

[Irion]

Funny how the requirements are more strict for the computer then for man, yet they are less likely to make mistakes once they have the skills in place.

Well, we now people for a long time. So it is quite a safe guess to assume what they do and what they do not. (Not considering crazy people, but crazy people will bea problem for trafic in any case...)

With a robot it is not that easy. The car getting stuck in a loop, the computer crashing or anything and bang it goes...

[Traul]

That, or nobody wants to tell the truckers union they're redundant.

[hermit]

Well, Rojas has been working on this since the late 90s (ever since he came to the FU back when they brushed up their Compiter Sciences), and the robotics lab at FU is one of the best worldwide.

However, the track they're talking about mostly is 6-lane streets and Autobahn, if they take the obvious route. Starting at Messedamm going onto the Autobahn, exiting next sortie at Kaiserdamm, then left through Kaiserdamm, staright over Ernst Reuther-Platz, following it to the Victoria Collumn, and then on through 17th June. There's three really difficult points: the slip road at Messedamm (bloody maze and lots of traffic of confused people coming in via AVUS from the southern outer ring highway), Ernst Reuther-Platz, and the Great Star, otherwise the car can go straight, stop at red lights and be overtaken. Little chance of spontaneously crossing pedestrians like on smaller streets or cyclists - most of these aren't suicidal enough to try their luck on these streets. If it manages that track entirely going through side roads, then we're talking.

Funny how the requirements are more strict for the computer then for man, yet they are less likely to make mistakes once they have the skills in place.

Based on which data?

Also, there must be a human driver present because in case of autonomous control, it's the company producing the controls who's at fault if the car runs over a kid, pedestrian, idiot cyclist or wildlife sich as boar or foxes (which are a lot more common in the city than you might think).

[Draco18s]

wildlife sich as boar or foxes (which are a lot more common in the city than you might think).

Having had a mangy fox take up residence in our barn* a few times, I am not surprised in the least (we're not in the boonies, either. We've got suburbia on all sides). Now an elephant is something you won't expect to see in a city.

*I had so much trouble typing this word, it's not funny.
bo...
br...
bw...
ban...

[Irion]

Also, there must be a human driver present because in case of autonomous control, it's the company producing the controls who's at fault if the car runs over a kid, pedestrian, idiot cyclist or wildlife sich as boar or foxes (which are a lot more common in the city than you might think).

You are acutally not allowed to break for small wildlife, such as foxes.

[Draco18s]

You are acutally not allowed to break for small wildlife, such as foxes.

Depending on the load, flat bed trucks aren't allowed to stop for anything short of "human." Too many and the load can break loose and kill the driver (I've been told it's good for "one emergency screeching halt").

[HunterHerne]

You are acutally not allowed to break for small wildlife, such as foxes.

It's true. It's seen as a safety hazard to other drivers.

[LurkerOutThere]

I've often wodnered why so much time is spent working on makign individual cars drive autonomously and why more effort isn't put into researching how to automate the roadbed. By designing millions of cars to drive independantly we're not really solving any problems or improving the current systems. In fact we could be making them worse.

[Sengir]

Funny how the requirements are more strict for the computer then for man, yet they are less likely to make mistakes once they have the skills in place.

...and as long as they are used in precisely the same environment for which their skills were developed. RL example from RoboCup Germany (two years ago I think), one team had obviously tested their robots in light conditions different from those at the competition. So on the field, the robots misclassified their own shadows as obstacles and did evasive maneuvers, until the team simply switched off obstacle detection...

[Kirk]

I've often wodnered why so much time is spent working on makign individual cars drive autonomously and why more effort isn't put into researching how to automate the roadbed. By designing millions of cars to drive independantly we're not really solving any problems or improving the current systems. In fact we could be making them worse.

Several reasons, some good some bad.

A good one is deployment cost. An automated roadbed is essentially an all-or-nothing expense, and is not useful if done below critical sizes.

Another good one is deployment flexibility. An automated roadbed is a fixed route and is less useful for those not on the route. Autonomous vehicles, on the other hand, have significantly greater ranges of useful origins and destinations. Rephrased, the roadbed may be ideal in getting from A to B, but users have to get from X to A and then from B to Z. With autonomous vehicles they can get from X to Z without worrying about A and B.

A bad reason is selfishness. People like to believe they're unique, and have their own transport on demand without needing to share or be significantly concerned with others. The autonomous vehicle plays upon that belief.

A lesson. There are three feasibility tests, and a successful project must pass all three.
Engineering: Can it be built?
Economic: Can we afford it?
Political: Do we want it/will we use it?

The roadbed can pass the political, eventually. The hurdle is a lot lower for autonomous vehicles.

[LurkerOutThere]

I would disagree that there is more of an engineering hurdle in the centralized system then in the car based one. RFID tagging embedded in license plates could go a long long way to enabling such a project. There would be a matter of gettign the sensors on the roadbed but their already looking at sensors in thhat vein to catch speeders and weigh trucks in motion. Lojack and other systems can already track cars fairly granularly. I know these arn't directly applicable but I just can never figure out why so much time and attention in making a car drive itself around all the plausible risks and vagaries of traffic when the answer is to work to eliminate the vulgarities of traffic leaving the cars autopilot to worry about non vehicle related items.

[Kirk]

I didn't say it had a lower engineering hurdle. I said lower political hurdle.

[hermit]

Having had a mangy fox take up residence in our barn* a few times, I am not surprised in the least (we're not in the boonies, either. We've got suburbia on all sides). Now an elephant is something you won't expect to see in a city.

Yeah, but have you ever seen a fox using the sidewalk of a mid city street carrying a neatly folded Mcdonald's takeaway bag in it's mouth? Because I have.

You are acutally not allowed to break for small wildlife, such as foxes.

You sure? I knowyou're not allowed to run over the larger, hunt-heavy animals, but have to brake and try to evade, at least that's what I was taught. So in the end they ask you to commit suicide so treefuckers and hunters are happy.

I've often wodnered why so much time is spent working on makign individual cars drive autonomously and why more effort isn't put into researching how to automate the roadbed. By designing millions of cars to drive independantly we're not really solving any problems or improving the current systems. In fact we could be making them worse.

An automated lane on the highway is among the things being planned here, and one of the few things about this e-mobility bullshit that can be salvaged. However, this makes sense for scenarios where there are few diversions save other cars and moving fast and packing cars tight is imperative. It will not help much in sidestreets or streets wehre there are routinely non-automated trafficants (pedestrians, cyclists, horse wagons....) to be expected. Also, in that case, whoever operates the system is liable for any accident that happens, which may not be what a company or government agency would want. Not to mention what kind of tempting target the wireless infrastructure necessary would be for hackers. And making wireless systems reliably secure is next to impossible.

RFID tagging embedded in license plates could go a long long way to enabling such a project.

It would also go a long way to total surveillance of the population.

[Sengir]

You sure? I knowyou're not allowed to run over the larger, hunt-heavy animals, but have to brake and try to evade, at least that's what I was taught. So in the end they ask you to commit suicide so treefuckers and hunters are happy.

There is no legal requirement for either behavior. However, the insurance will usually not pay for damage you cause while evading something that wouldn't damage your car.

[KarmaInferno]

I've often wodnered why so much time is spent working on makign individual cars drive autonomously and why more effort isn't put into researching how to automate the roadbed. By designing millions of cars to drive independantly we're not really solving any problems or improving the current systems. In fact we could be making them worse.

You mean like a train?

(IMG:style_emoticons/default/smile.gif)




-k

[CanRay]

Oh the way rail lines could be improved! The Canadian rail lines used to be something to be enjoyed, now? Horrible.

Still better than flying, however.

[HunterHerne]

Oh the way rail lines could be improved! The Canadian rail lines used to be something to be enjoyed, now? Horrible.

Still better than flying, however.

Especially to America?