Experts ONLY: comment on whether vision+forward radar is sufficient for Level 5?

[Sir Guacamolaf]

Tesla, a publicly traded company, which carries a great deal of weight as it relates to being truthful and honest about what they know at the time they say it. They are yelling as loudly as possibly that they have all but solved FSD and they solution is different then everyone else's. The demo will be Nov or Dec of this year. My guess is that they don't even mention that unless they could do it today, because of the weight their words carry great consequence

............I will happily eat a giant bucket of fried Crow if I have been bamboozled by the Musk that is Elon.

Well you'd better start frying then, Elon has bamboozled numerous times, starting with the world's best service experience, AP1 will summon on private property, and yes FSD - it isn't happening. What will happen is much better autopilot on freeways, and cleaner inside streets - and that's 90% of driving anyway, so it'll all be fine, but it isn't FSD.

Then you'll have people arguing in favor and against, technically he promised LA to NY (which could be freeway exit to freeway exit). He never promised driveway to parking garage.

Anyway, start frying.

 

[calisnow]

c'mon everybody this is one thread that has a lot of good content - let's try to keep the bickering, taunting and ad hominem on this one to a minimum. It makes for a much more pleasant experience for someone coming through later reading for content.

 

[Reciprocity]

Well you'd better start frying then, Elon has bamboozled numerous times, starting with the world's best service experience, AP1 will summon on private property, and yes FSD - it isn't happening. What will happen is much better autopilot on freeways, and cleaner inside streets - and that's 90% of driving anyway, so it'll all be fine, but it isn't FSD.

Then you'll have people arguing in favor and against, technically he promised LA to NY (which could be freeway exit to freeway exit). He never promised driveway to parking garage.

Anyway, start frying.

Parking lot to parking lot was promised during the TED talk. Also, the route and destination could change dynamically. I think the example given was LA to NY could change miss trip to Toronto.

 

[dknisely]

Elon said in the earnings Q&A today that Model 3 will have the same sensor suite as Model S with AP2 no changes and the only possible upgrade it would ever need is a CPU/GPU processor swap.

Great news. I haven't seen that specific statement before.

And making sure that I get the processor upgrade "for free" was the reason that I pre-paid for FSD.

 

[EVie'sDad]

Trump’s transportation secretary made a huge mistake about self-driving cars
13
Definitely do not leave your seat while using driver assistance features
by Sean O'Kane@sokane1 May 3, 2017, 3:59pm EDT
Trump’s transportation secretary made a huge mistake about self-driving cars

 

[J1mbo]

Trump’s transportation secretary made a huge mistake about self-driving cars

Wrong thread for this.

 

[EVie'sDad]

But it does communicate quite well that unequivocally we are not yet beyond Level 2 autonomy at this time and should not be under any misconceptions that it is anything more than a driver assistance feature at this time. Although the sec's comments lead me to believe she is either poorly or ill-informed about the subject.

 

[malcolm]

Parking lot to parking lot was promised during the TED talk. Also, the route and destination could change dynamically. I think the example given was LA to NY could change miss trip to Toronto.

What he actually said was "I think we're still on track for being able to go cross-country from LA to New York, by the end of the year, fully autonomous" etc

"We should be able to go all the way from a parking lot in LA to a parking lot in NY"

Scroll to 15:07 mark:
The future we're building -- and boring

I think it will happen, but less confident with the November/December timeframe

 

[Reciprocity]

What he actually said was "I think we're still on track for being able to go cross-country from LA to New York, by the end of the year, fully autonomous" etc

"We should be able to go all the way from a parking lot in LA to a parking lot in NY"

Scroll to 15:07 mark:
The future we're building -- and boring

I think it will happen, but less confident with the November/December timeframe

Keep listening he says they could change the destination mid trip with no issues.

 

[Bladerskb]

Keep listening he says they could change the destination mid trip with no issues.

How is it that you can't understand that freeway driving is a magnitude easier than urban driving?
It doesn't matter if he changed it to DC, Minnesota or Florida. Or went from LA to NY to Miami to DC to Houston without a touch. Its still all freeway driving. Driving on an empty road in one lane is not hard at all. Its EASY. If Elon can't do that in a "demo", which has been done to death years ago. Then you should be concerned.

But anything other than software in customer's hands of FSD that's better than human and doesn't require intervention is meanless.

@malcolm

 

[Reciprocity]

How is it that you can't understand that freeway driving is a magnitude easier than urban driving?

Odd, I havnt seen any parking lots on the freeway.

But anything other than software in customer's hand of FSD that's better than human and doesn't require intervention is meanless.

This is an enormously false and a really short sided statement. Once the software in any form is pushed to the car, the car could take over in any emergency situation as defined by the software. This would allow it to help improve safety before it is fully released. This shows why some people just dont get it and definitely the auto industry doesnt get it. The real world isnt full of absolutes and FSD will only be 10x better then humans when it is released to the public, It will not be perfect in every situation. Sorry, that is not how the real world works.

I continue to believe Tesla has a massive lead and is the only company that has hardware in the hands of customers that is ready. Tesla has showed their hand by putting all this extra hardware in the car. There is no reason to do that if it wont be ready for HW2, they would have just waited for HW3 for full autonomy and HW2 is not required for AP2, they could have gone with a much simpler hardware solution.

I mean its really simple to me. I have someone on a message board telling me that they know more then 100's of engineers and Elon Musk who regularly does the impossible and I am supposed to believe the forum troll. I mean really? Who does that. I guess they could be lying to us, but if you look at the downside to them lying vs the downside of forum troll assuming they know more then everyone else, its clear to me who to trust.

Again, I could be wrong, and I will happily admit it in Jan, 2018. I am sure this post will still be here.

 

[Bladerskb]

Odd, I havnt seen any parking lots on the freeway.



This is an enormously false and a really short sided statement. Once the software in any form is pushed to the car, the car could take over in any emergency situation as defined by the software. This would allow it to help improve safety before it is fully released. This shows why some people just dont get it and definitely the auto industry doesnt get it. The real world isnt full of absolutes and FSD will only be 10x better then humans when it is released to the public, It will not be perfect in every situation. Sorry, that is not how the real world works.

I continue to believe Tesla has a massive lead and is the only company that has hardware in the hands of customers that is ready. Tesla has showed their hand by putting all this extra hardware in the car. There is no reason to do that if it wont be ready for HW2, they would have just waited for HW3 for full autonomy and HW2 is not required for AP2, they could have gone with a much simpler hardware solution.

I mean its really simple to me. I have someone on a message board telling me that they know more then 100's of engineers and Elon Musk who regularly does the impossible and I am supposed to believe the forum troll. I mean really? Who does that. I guess they could be lying to us, but if you look at the downside to them lying vs the downside of forum troll assuming they know more then everyone else, its clear to me who to trust.

Again, I could be wrong, and I will happily admit it in Jan, 2018. I am sure this post will still be here.

No you have a software engineer with knowledge of machine learning and computer vision plus other software engineers and engineers in various of specialization telling you the same thing. You can't have a massive lead without software. Hardware is mean-less, anyone can put cameras in cars today and a NVidia chip and call it done.

I have explained to you the inner workings of self driving technology and you have refused to learn.

Do you care about anything factual then? Tesla HQ to NYC Time Square has about 2 miles of surface street.

Do you even understand that humans drive accurately 99.9999% of the time? Even if you take into account that some accidents are not reported (50%?) and double or triple the accident rate. You are still looking at 99.9995%. To be 10x better than human you have to be somewhere around 99.99999%

For a L4/L5 car
99% = 1 (disengagement/accident) in 100 miles
99.9% = 1 (disengagement/accident) in 1000 miles

And as Elon said himself. You don't want to fall asleep in a car that gets into an accident every 1,000 miles (99.9%).
Obviously if you know someone that gets into accident every 1,000 miles (which is every other week). They need their drivers license ripped away from them forever. if you were to know someone who has an accident every 10,000 miles. They need to go back to driver's ed.


99.99% = 1 (disengagement/accident) in 10,000 miles
99.999% = 1 (disengagement/accident) in 100,000 miles
99.9999% = 1 (disengagement/accident) in 1,000,000 miles

Multiple that last one times 10 is your "FSD will only be 10x better then humans when it is released to the public".
Again I have relayed these facts to you before but seems like you don't care about truth

 

[JeffK]

99% = 1 (disengagement/accident) in 100 miles
99.9% = 1 (disengagement/accident) in 1000 miles

Keep in mind a disengagement is not an accident and shouldn't be treated as one in statistics. A disengagement doesn't necessarily mean the system did anything wrong. For example, using AP 1 or EAP a disengagement is whenever you get off the highway and turn it off.

NHTSA in 2015 estimated 47% of traffic accidents go unreported so you were about right with 50%.

It can be shown statistically even in shadow mode, that the system is safer than a human. In the future even to be twice as safe as a human you'd only have reduce the accident rate to the current reported number haha.

 

[Bladerskb]

Keep in mind a disengagement is not an accident and shouldn't be treated as one in statistics. A disengagement doesn't necessarily mean the system did anything wrong. For example, using AP 1 or EAP a disengagement is whenever you get off the highway and turn it off.

NHTSA in 2015 estimated 47% of traffic accidents go unreported so you were about right with 50%.

It can be shown statistically even in shadow mode, that the system is safer than a human. In the future even to be twice as safe as a human you'd only have reduce the accident rate to the current reported number haha.

A disengagement is not a accident for a L1, L2, and L3 if the human were to take over.
But an disengagement in an L4 or L5 car which both do not require a human driver to be in it will result in an accident.
L4/L5 doesn't need a human driver awake, or in it or in the driver seat.
L4/L5 cars cannot disengage. L4 can ask you to intervene if you were around but it will handle the situation safely.
For example an L4 car that is going out of its geo-fenced zone or can't handle construction area will alert you and then pull over if you don't respond.

"using AP 1 or EAP a disengagement is whenever you get off the highway and turn it off."
Lastly voluntary take over is not an disengagement.
That is also why its not reported in the California DMV. An disengagement is when the system fails and immediately hands back the control to the driver or the driver takes over immediately to avert a collision or wrong action by the car.

For example a request from a L3 car to take over with sufficient time (10-15) is not a disengagement, but rather a operational domain limit.

 

[JeffK]

L4/L5 cars cannot disengage.

This is a false assumption as long as cars keep being made with steering wheels and pedals. You will still be able to manually take control any time you want.

One might say in a perfect world an L5 car wouldn't need a steering wheel but that doesn't mean an L5 car can't have a steering wheel. People still like to manually drive in a little more sporty fashion when they aren't travelling to and from work.

Here's a cool quote from Elon on the topic:

To be clear, Tesla is strongly in favor of people being allowed to drive their cars and always will be. Hopefully, that is obvious

 

[bob_p]

When Tesla takes the official coast-to-coast FSD trip, it's likely they'll do that only after they've made multiple unofficial trips on that route - and will only make the official trip after they've proven the software can handle the roads and conditions likely to be encountered on that trip.

Tesla could also adapt the route to avoid those areas where the software could encounter challenges.

And, that's OK - it'll be another demo of what is possible - even if it's not an indication that FSD is ready for prime time.

We'll know FSD is ready - when Tesla can do a coast-to-coast trip, using the onboard navigation software routing, to an arbitrary coast-to-coast destination, and drive the route without any human interaction. Seems unlikely we'll see that this year (or next year)...

 

[SlyWombat]

For starters, SAE autonomous driving grades need to be stated and understood up front:
SAE J3016 Automated Driving Standards
Level: 5
Name: Full Automation
Narrative Definition: the full-time performance by an automated driving system of all aspects of the dynamic driving task under all roadway and environmental conditions that can be managed by a human driver.
Execution of Steering and Acceleration/ Deceleration: System
Monitoring of Driving Environment : System
Fallback Performance of Dynamic Driving Task : System
System Capability (Driving Modes) : System

Definition of driving mode: Driving mode is a type of driving scenario with characteristic dynamic driving task requirements (e.g., expressway merging, high speed cruising, low speed traffic jam, closed-campus operations, etc.).

The above are all SAE definitions, verbatim.

In short, L5 requires full 360 degree situational awareness without requiring human intervention. For example, lane changes cannot be executed without keeping track of cars, bikes, bicycles or pedestrians in the rear flanks. Front visibility is not sufficient.

In my opinion, upto L3 is quite straightforward to implement. AP2 should be able to accomplish it. The jump from L3 to L4 is very tricky, not just technologically, but in terms of regulatory oversight. L4 to L5 is even harder.

Note that a number of humans do not qualify for this as well (how many accidents per year?). But agree that depth based vision not just forward, but like a human driver is all that is needed for hardware. Software advances will give us level 4 fairly quickly (2 years), edge cases to properly qualify for L5 will be subjective, just like human drivers cannot cope with weather they have not seen. Fun watching people who have never driven in snow

BSc Computer Science, CTO, neural networks and AI (not driving related).

 

[JeffK]

When Tesla takes the official coast-to-coast FSD trip, it's likely they'll do that only after they've made multiple unofficial trips on that route - and will only make the official trip after they've proven the software can handle the roads and conditions likely to be encountered on that trip.

Tesla could also adapt the route to avoid those areas where the software could encounter challenges.

And, that's OK - it'll be another demo of what is possible - even if it's not an indication that FSD is ready for prime time.

We'll know FSD is ready - when Tesla can do a coast-to-coast trip, using the onboard navigation software routing, to an arbitrary coast-to-coast destination, and drive the route without any human interaction. Seems unlikely we'll see that this year (or next year)...

Elon seems to disagree and thinks it'll happen toward the end of this year still

Chris: Won’t have to ever touch the wheel. By the end of 2017.

Elon: Yeah essentially November or December of this year we should be able to go from all the way from a parking lot in California to a parking lot in New York. No controls touched at any point during the entire journey.

Applause

Chris: Amazing. Part of that is possible because you’ve already got a fleet of Teslas driving all these roads. You’re accumulating a huge amount of data of national road system.

Elon: Yes. But the thing that was interesting is that I’m actually fairly confident it will be able to do that route even if you change the route dynamically. So it gets it’s fairly easy… If you say I’m going to be really good at one specific route, that’s one thing, but it should be able to go, really be very good, so once you enter a highway, to go anywhere on the highway system in a given country. It’s not sort of limited to LA or New York. We could we could change and make it. Seattle-Florida, that day, or you know, in real time. So you are going from L.A. to New York now go from L.A. to Toronto.

 

[bob_p]

The coast-to-coast demo will very likely happen this year, even if it means Tesla will carefully select the route being used to avoid areas where the software could encounter unexpected conditions, and Tesla has to do some software modifications to ensure the software handles the highway intersection transitions and is able to drive to/from each of the superchargers along the route. This isn't unusual when software teams are preparing a major demonstration.

Musk isn't claiming this demo will show the FSD software is ready for use on any route - it is only a demonstration that FSD could go coast-to-coast (including charger stops) without human interaction.

This type of demo is typical for major software projects - a "proof of concept" demonstration of what the software will eventually be able to do, but under controlled conditions that allows the developers to simplify the software without spending the bulk of the development time required to handle all expected situations. [90% of the software development can be spent in getting the critical last 10% of functionality needed to get the software to achieve goals.]

Now, if Musk states that after the coast-to-coast demo, the software is ready to handle any arbitrary route, without human interaction, then FSD really would be ready for prime time by the end of this year, even if it required human monitoring (driver assistance mode) until receiving regulatory approval.

But I don't believe Musk has claimed the demo is an indication FSD is fully ready - I believe he's projected that is two years out (2019).

 

[shurst]

So I was reading some articles at work today and came across this. I'm not saying it's impossible or possible, just figured I'd contribute some insights. This is an article published by NVIDIA about using neural nets and just images with driver input. I'd say I also meet the requirements of the OP, PhD, head of data science, machine learning and some computer vision. I do like the debate and I know this is highly contested in the field. I think people are trying many things and any combination has the possibility of working. Largely depends on the approach. Likewise, to invest this heavily in hardware, they must have seen a proof of concept that convinced them they can do it when they were doing R&D. Not usually a decision that's made on the fly. Anyway, happy reading! Hope this article sheds a bit of light on maybe how they're thinking! Also they're not using a tesla in this article...

https://arxiv.org/pdf/1704.07911.pdf