Full autonomy in 2-3 years

[paulj]

Interesting video posted today of Elon talking about Tesla. Amongst lots of interesting points, says the car will be fully autonomous in 2 (or worst case 3) years. See from 8 mins:
https://www.youtube.com/watch?v=K1Zvatbdydk

 

[Todd Burch]

I think the general consensus on this is that Elon is a bit optimistic.

 

[EcoHeliGuy]

No there is a difference between capable and legal. 2-3 years is very probable

 

[Max*]

No there is a difference between capable and legal. 2-3 years is very probable

I don't think it is, not for "full" autonomy, aka Level 4.

Level 3 (highway autonomy), sure 2-3 years is very probable.

 

[Todd Burch]

No there is a difference between capable and legal. 2-3 years is very probable

Capable as in take a nap and let the car do the work? Not a chance. I'm all for Elon and am amazed at what he's been able to do, but he's a bit off the deep end on this one. Even the Google car which has much more hardware and is much further along with software still fails in many situations--and that's not even allowing it to travel at higher speeds.

IMHO a safe autonomous car is even harder than spaceflight, just as autopilot in a car is much harder than autopilot in an aircraft.

 

[dhanson865]

Interesting video posted today of Elon talking about Tesla. Amongst lots of interesting points, says the car will be fully autonomous in 2 (or worst case 3) years. See from 8 mins:
https://www.youtube.com/watch?v=K1Zvatbdydk

that's an old video already posted on this forum twice before.

 

[S4WRXTTCS]

There is no difference between elon 2-3 year time frame, and a decade time frame.

He's being very consistent with his time estimates.

 

[Max*]

Capable as in take a nap and let the car do the work? Not a chance. I'm all for Elon and am amazed at what he's been able to do, but he's a bit off the deep end on this one. Even the Google car which has much more hardware and is much further along with software still fails in many situations--and that's not even allowing it to travel at higher speeds.

IMHO a safe autonomous car is even harder than spaceflight, just as autopilot in a car is much harder than autopilot in an aircraft.

Google is aiming to get straight to Level 4, full autonomous even in the city. It's a big difference. I think it's possible to have highway autonomous (get on the highway, press a button, take a nap) in 2-3 years.

 

[Todd Burch]

I think it's possible to have highway autonomous (get on the highway, press a button, take a nap) in 2-3 years.

Maybe, if all the stars completely align...but of course that's not full-autonomous. Even on the highway, I think 2-3 years is really stretching it. Tons of situations to consider. Suppose the car enters a construction zone where there are no lane markings and traffic is being directed by a flagger, for instance? Good luck with that one!

 

[Ryan H]

I predict separate divided highways for autonomous vehicles. It will keep them away from human drivers, which will be good for both parties. I think most human drivers will be skeptical of autonomous vehicles for quite a while.

 

[EcoHeliGuy]

I predict separate divided highways for autonomous vehicles. It will keep them away from human drivers, which will be good for both parties. I think most human drivers will be skeptical of autonomous vehicles for quite a while.

I predict no government is ever going to finance that.

Won't happen and no legitimate reason to.

 

[Ryan H]

I predict no government is ever going to finance that.

Won't happen and no legitimate reason to.

I think there's a legitimate reason. Autonomous vehicles will perform differently from human drivers. As Todd pointed out, what will happen when a fully autonomous vehicle enters a construction zone? The infrastructure for autonomous vehicles will be different. There is enough skepticism around autonomous vehicles that they will function best isolated from human drivers. What will happen when a person in an autonomous vehicle is sleeping behind the wheel and someone pulls up next to them and sees them sleeping? The other driver will most likely panic and call the police. The point I'm trying to make is there will be people that will be resistant to this, and I think there's going to need to be separation for them to adjust to it. They will not want to drive next to other vehicles where the person inside the vehicle is sleeping.

 

[Max*]

I think there's a legitimate reason. Autonomous vehicles will perform differently from human drivers. As Todd pointed out, what will happen when a fully autonomous vehicle enters a construction zone? The infrastructure for autonomous vehicles will be different. There is enough skepticism around autonomous vehicles that they will function best isolated from human drivers. What will happen when a person in an autonomous vehicle is sleeping behind the wheel and someone pulls up next to them and sees them sleeping? The other driver will most likely panic and call the police. The point I'm trying to make is there will be people that will be resistant to this, and I think there's going to need to be separation for them to adjust to it. They will not want to drive next to other vehicles where the person inside the vehicle is sleeping.

That's the silliest reason I've ever heard, and I've been on TMC for a while now...

A more legitimate reason might be that if you have a row of autonomous vehicles you can allow them to travel at a higher rate of speed keeping a distance of 6" between cars. And that would give incentives (think HOV lanes) for people to buy autonomous vehicles.

 

[Ryan H]

A more legitimate reason might be that if you have a row of autonomous vehicles you can allow them to travel at a higher rate of speed keeping a distance of 6" between cars. And that would give incentives (think HOV lanes) for people to buy autonomous vehicles.

I concede that would be a better reason.

 

[calisnow]

I predict separate divided highways for autonomous vehicles. It will keep them away from human drivers, which will be good for both parties. I think most human drivers will be skeptical of autonomous vehicles for quite a while.

I don't think the U.S. government has the kind of money to spend going about building new roads for autonomous vehicles - however I *do* think that one thing which would really help avoid accidents would be freshly painted lane markings so AP can know with a higher degree of confidence where it is. Don't know how it is for you in Ohio, but here in California we have many terrible roads with very poor markings - Autopilot's job would be much easier with a simple re-paint of the stripes.

 

[EcoHeliGuy]

I don't think the U.S. government has the kind of money to spend going about building new roads for autonomous vehicles - however I *do* think that one thing which would really help avoid accidents would be freshly painted lane markings so AP can know with a higher degree of confidence where it is. Don't know how it is for you in Ohio, but here in California we have many terrible roads with very poor markings - Autopilot's job would be much easier with a simple re-paint of the stripes.

Couldn't agree more

 

[Ryan H]

I don't think the U.S. government has the kind of money to spend going about building new roads for autonomous vehicles - however I *do* think that one thing which would really help avoid accidents would be freshly painted lane markings so AP can know with a higher degree of confidence where it is. Don't know how it is for you in Ohio, but here in California we have many terrible roads with very poor markings - Autopilot's job would be much easier with a simple re-paint of the stripes.

At the risk of digging myself into a deeper hole, fresh lines won't help when it comes to snow-covered roads. That's why I'm leaning toward something different. I'm sure there's something out there that would be very good for hardware to detect regardless of (human) visibility.

 

[EcoHeliGuy]

At the risk of digging myself into a deeper hole, fresh lines won't help when it comes to snow-covered roads. That's why I'm leaning toward something different. I'm sure there's something out there that would be very good for hardware to detect regardless of (human) visibility.

I suggest you watch the videos on YouTube of a Tesla driving snow covered roads. For a Beta you have nothing to worry about

 

[Ryan H]

I suggest you watch the videos on YouTube of a Tesla driving snow covered roads. For a Beta you have nothing to worry about

Do you have any specific suggestions? The only ones I've seen are ones where the car follows the vehicle in front of it, which has nothing to do with lane markings.

 

[EcoHeliGuy]

Do you have any specific suggestions? The only ones I've seen are ones where the car follows the vehicle in front of it, which has nothing to do with lane markings.

There is one which I will try and find, near the end there is no other traffic and the car keeps going without "indicating" road marking on the display. The car does some larger corrections but for a Beta is was damn impressive.