Fully automated vehicles - 'not in our lifetime'?

Cash-Strapped.T32

GreatApe wrote: »

But we already have what appears to be the first true AIs

Western's point however war larger & more interesting than that - He was putting forward the idea that in technological progress, there comes a plateau point, where further gains come at greater & greater cost (almost akin to the energy requirements to accelerate a body, etc..).

In short, that early gains are relatively "cheap", and so technological progress in a given field seems rapid, but there comes a point where this plateaus and so you need to take different approaches to a problem in order to keep up the same rate of improvement.

Eventually, you will run out of alternative angles to take, and thus your rate of improvement in a given field will slow.

That as a concept is an interesting one to explore, and to be fair deserves more thorough consideration than it's likely to get in this thread.

kabayiri

Cash-Strapped.T32 wrote: »

....
That as a concept is an interesting one to explore, and to be fair deserves more thorough consideration than it's likely to get in this thread.

Sometimes, emergent technology allows you to redefine the problem space though.

One of the more famous computing companies had a tagline years ago : "the network *is* the computer". This was pre-internet days.

Now, we take for granted collaborative; distributed computing.

Our road network only works successfully because of the protocols of use we have introduced.

There is no reason to suggest these protocols remain fixed once more machine involvement becomes commonplace.

Cash-Strapped.T32

kabayiri wrote: »

There is no reason to suggest these protocols remain fixed once more machine involvement becomes commonplace.

For sure - I mean in relation to the overall thread, in my opinion entirely self-driving cars & HGVs are a relatively easy problem to solve - it's only human behavior, based on us voraciously buying-up packaged dreams of the open road that makes any of it difficult.

kabayiri

Cash-Strapped.T32 wrote: »

For sure - I mean in relation to the overall thread, in my opinion entirely self-driving cars & HGVs are a relatively easy problem to solve - it's only human behavior, based on us voraciously buying-up packaged dreams of the open road that makes any of it difficult.

Which ties into my earlier point that there are many other species on the planet which manage to act in cooperative behaviour. We are only just beginning to learn just how elaborate this can be with insects and marine life.

If machines lead to greater cooperation between vehicles, it changes what protocols we need.

GreatApe

Cash-Strapped.T32 wrote: »

Western's point however war larger & more interesting than that - He was putting forward the idea that in technological progress, there comes a plateau point, where further gains come at greater & greater cost (almost akin to the energy requirements to accelerate a body, etc..).

In short, that early gains are relatively "cheap", and so technological progress in a given field seems rapid, but there comes a point where this plateaus and so you need to take different approaches to a problem in order to keep up the same rate of improvement.

Eventually, you will run out of alternative angles to take, and thus your rate of improvement in a given field will slow.

That as a concept is an interesting one to explore, and to be fair deserves more thorough consideration than it's likely to get in this thread.

I don't think we are yet close to that point.
Transistors can still get smaller until around 2025 after that they can be packed in tighter

Beyond that we could just produce larger chips. The apple iPhone chip is a lot faster than the best from android partly because its almost twice the physical size.

While I accept there could be some point where further progress is difficult I doubt it is going to be within the next 20 years which means we have at least a 1000x from today. That would take us towards 5 trillion transistors per chip. After that point if nothing else we could just go from 100 cubic millimeters to 1 million cubic millimetres volume for chips (still smaller than a hunan skull). That gets us another 10,000x to take us towards 50 quadrillion transistors. That would be 50 x more transistors than the Hunan Brian has synapses.

kinger101

kabayiri wrote: »

And all of this has precisely nothing to do with automated vehicles.

I think we are well off point here.

What did you expect? These threads always end up in pseudo-intellectual gobbledygook.

boliston

Herzlos wrote: »

......

3. The cross-over would be horrendous. You'd probably need to spend 10 years with overlapping human/robot drivers before you can start optimizing the infrastructure.

In terms of collision space; the cars would hopefully identify most mechanical problems and warn their neighbours in plenty of time to allow them to back off until it can park up and await recovery. For external issues like a jaywalker, all cars on that street would be able to brake more or less in parallel so should be able to avoid running into each other.

Even if human driven cars were phased out you are unlikely ever to see roads in cities that are dedicated to autonomous-only vehicles as there will always be pedestrians and cyclists using the roads. The scenes like you see in minority report and total recall with computer driven cars racing through cities at speed on dedicated elevated roads would cause too much blight to be accepted from a planning point of view.

antrobus

kinger101 wrote: »

What did you expect? These threads always end up in pseudo-intellectual gobbledygook.

Trouble is, a lot of it doesn't even reach the standard of 'pseudo-intellectual gobbledygook'.

kabayiri

boliston wrote: »

Even if human driven cars were phased out you are unlikely ever to see roads in cities that are dedicated to autonomous-only vehicles as there will always be pedestrians and cyclists using the roads. The scenes like you see in minority report and total recall with computer driven cars racing through cities at speed on dedicated elevated roads would cause too much blight to be accepted from a planning point of view.

This is why it would make more sense to develop and refine the technology in places where there is very little infrastructure, or little resistance to change.

The cost to make the road network ready in somewhere like London might be too high to be in the first wave.

HAMISH_MCTAVISH

kabayiri wrote: »

The cost to make the road network ready in somewhere like London might be too high to be in the first wave.

Or ever.....