Autonomous Car Progress

[diplomat33]

SuperCruise seems to be worse than Autopilot at crash avoidance, at 7:20 in the video (where a car veers into his lane and he has to take evasive action while the SuperCruise just goes merrily along). Autopilot would likely have braked for that, as I had similar situations happen to me. Also a good reason to keep your hands on or near the wheel with SuperCruise:

I don't think braking would necessarily have prevented that accident. It looks like the car was going to side hit the other car. I would not trust AP to handle that. I would have taken over on AP just to be safe.

Supercruise does handle the merging lane better than AP earlier in the video. Supercruise does not try to center in the wider lane like AP does.

Also keep in mind that Supercruise is L2 so the driver is still expected to monitor and be ready to take over. It just let's the driver keep their hands near the wheel rather than require physical hands on wheel.

I don't think AP handles every case better than SC or SC handles every case better than AP. I am sure each handle some things better and some things worse. They are different systems with pros and cons.

 

[diplomat33]

"I'm going to engage another automatic lane change"

I do not think automatic means what he thinks it means.

I think it means the same as when we use the stalk to do an auto lane change. We initiate the lane change with the stalk but the car performs the lane change automatically.

 

[powertoold]

Supercruise does not try to center in the wider lane like AP does.

Ya, because it's all premapped beforehand. AP's centering with widening lanes has been improved over the last 6 months and is a non-issue for me, as it's not human-like but definitely not dangerous or confusing.

It looks like the car was going to side hit the other car. I would not trust AP to handle that. I would have taken over on AP just to be safe.

I wouldn't "trust" autopilot to handle that, but it would have at least taken some action or beeped. In that SuperCruise video, it seems the system didn't take any action whatsoever. It was not until later that the driver swerved.

 

[Bladerskb]

SuperCruise seems to be worse than Autopilot at crash avoidance, at 7:20 in the video (where a car veers into his lane and he has to take evasive action while

Also another problem with SuperCruise as mentioned in the video is that he differentiates two different SuperCruise "systems" (at 7:40), when it likely is just one system with two different software versions. Another reason Autopilot is vastly superior IMO:

Actually there is supercruise 1 and supercruise 2.0. They are different systems.
There is also two forward cameras not one. New one can auto lane change the other can’t. New one works on interchange and ramps but the last gen doesn’t.

I believe this also uses *EyeQ3 (2013 tech) but can’t confirm till a tear down happens.

 

[daniel]

I don't mind having to keep my hands on the wheel as long as it's L2 and I'm responsible. If they offered me no-nag hands-off-the-wheel but still Level 2 for, say $500 I would not take it. If they offered me Level 3 (car is responsible for alerting me ahead of time when it needs me to take over, and I can take my attention off the road) I'd pay $5,000 for that in an eye blink if it demonstrated that it doesn't need to do it very often.

No-hands Super Cruise has no value for me as long as it's Level 2.

For Level 4 sleep-in-the-back-seat I'll pay whatever they ask.

I would never dream of hiring my car out as a robotaxi. Letting strangers into my car alone for the small amount they'd pay, not worth it. Even if I could monitor them with the camera (which I could not do if the car was working at night while I'm sleeping) it would not be worth it. Ten or fifteen bucks for a ride and maybe the spill a soda on the seat?! No way.

 

[Bladerskb]

Yup, Wholemars went from disengaging every few miles to driving from Silicon Valley to LA and back within a month. It's possible.

People will say OHHH highway driving is so easy though! Well, show us some other company doing that.

As i said in my other post. Other companies ARE doing it. They just don't have the hype Tesla has. There's like 1 million (?) Tesla AP video on youtube and there is like under 250 for other ADAS systems combined which clearly outnumber AP (guesstimating).

However EV startups are striding ahead. 42Marks does one of the best independent testing of ADAS in china and they did one for Navigate on Autopilot, Navigated Guided Pilot and Navigate on NIO Pilot. The results show these companies have caught up and over took Tesla's NOA in china in alot of scenarios. If they were an American company, it would be the same also. Its also important to note that NIO's ES6 and ES8 is doing this with only forward camera and no surround camera unlike Tesla and Xpeng.

Also by the end of 2021, 2 more companies will not only have on-ramp to off-ramp (Geely's Lync & Co - Zero and Allibaba's Zhiji Auto - IM).
They will both likely also have point to point (door to door) system, with Geely most likely because its using SuperVision and Zhiji saying its pending regulation.

Its about to get heated.

Watch Video using closed caption:

 

[diplomat33]

As i said in my other post. Other companies ARE doing it. They just don't have the hype Tesla has. There's like 1 million (?) Tesla AP video on youtube and there is like under 250 for other ADAS systems combined which clearly outnumber AP (guesstimating).

However EV startups are striding ahead. 42Marks does one of the best independent testing of ADAS in china and they did one for Navigate on Autopilot, Navigated Guided Pilot and Navigate on NIO Pilot. The results show these companies have caught up and over took Tesla's NOA in china in alot of scenarios. If they were an American company, it would be the same also. Its also important to note that NIO's ES6 and ES8 is doing this with only forward camera and no surround camera unlike Tesla and Xpeng.

Also by the end of 2021, 2 more companies will not only have on-ramp to off-ramp (Geely's Lync & Co - Zero and Allibaba's Zhiji Auto - IM).
They will both likely also have point to point (door to door) system, with Geely most likely because its using SuperVision and Zhiji saying its pending regulation.

Its about to get heated.

Watch Video using closed caption:

Thanks.

Also, GM has Super Cruise. Nissan has Pro Pilot 2.0, Mercedes has Drive Pilot. Lucid has Dream Drive. So there are plenty of automakers now with highway L2-L3 systems, most of them are hands-free too, unlike AP. Tesla definitely does not have a monopoly on "autopilot" systems.

 

[powertoold]

Show some stress testing by real owners. That's all I ask and have been asking for.

Don't show some well marked highways with mostly straight lines.

Tesla is well ahead once you go into real world use. These marketing type videos are misleading.

 

[Bladerskb]

Show some stress testing by real owners. That's all I ask and have been asking for.

Don't show some well marked highways with mostly straight lines.

Tesla is well ahead once you go into real world use. These marketing type videos are misleading.

That's not a comprehensive independently verifiable comparable test of several systems. NIO, Tesla, Xpeng has highway autopilot systems. They were all tested extensively in similar conditions, scenarios and environment.

Its a fact that Xpeng and NIO systems are activatable in more roads than Tesla. That's a fact, period. A fact you will never get from the above video you posted. Henceforth the video you posted is an antidote and not comparable nor is it a benchmark.

Its funny how you call them marketing hype videos when 42Mark would actually be classified as Tesla fanboys. Half their videos is promoting Tesla. Did you even watch the video with CC? Why only accept something if it paints the company you like in favorable manner?

Show me any one or company doing extensive comprehensive side by side tests with dozens of cars like these..
You can't. Period.

 

[powertoold]

Show me any one or company doing extensive comprehensive side by side tests with dozens of cars like these..
You can't. Period.

Most of these are comparing user experience type features. We are talking about performance. Again, show me the stress tests. I want to see where the system fails, not some 2 minute clips of straight lines.

 

[Bladerskb]

Most of these are comparing user experience type features. We are talking about performance. Again, show me the stress tests. I want to see where the system fails, not some 2 minute clips of straight lines.

Litterally everything they test is performance based. You are literally refusing it because you don't like the results.

They test:

1. Cut-ins
2. Vehicle Partial in ego lane
3. Longitudinal Control smoothness
4. Lateral Control smoothness
5. Different degrees of curves
6. Automated Lane Changes
7. Lane Keeping
8. Ability to take exits
9. On-Ramp Lane Keeping
10. Off-Ramp Lane Keeping
11. Merging Capability
12. Auto Parking
13. Seeing stopped car from various distance
14. Stopping for stopped car
15. Response to Forward Car Hard braking
16. Automated Routing to exit Efficiency
17. Weather Conditions
18. Day/Night
19. Lane Keeping through Intersection with and without forward car
20. Handling blinding light out of tunnels
21. Stopping for a line of construction cones in ego lane
22. etc

 

[powertoold]

Litterally everything they test is performance based. You are literally refusing it because you don't like the results.

They test:

1. Cut-ins
2. Vehicle Partial in ego lane
3. Longitudinal Control smoothness
4. Lateral Control smoothness
5. Different degrees of curves
6. Automated Lane Changes
7. Lane Keeping
8. Ability to take exits
9. On-Ramp Lane Keeping
10. Off-Ramp Lane Keeping
11. Merging Capability
12. Auto Parking
13. Seeing stopped car from various distance
14. Stopping for stopped car
15. Response to Forward Car Hard braking
16. Automated Routing to exit Efficiency
17. Weather Conditions
18. Day/Night
19. Lane Keeping through Intersection with and without forward car
20. Handling blinding light out of tunnels
21. Stopping for a line of construction cones in ego lane
22. etc

Again, show me owners’ stress tests. We’ve been here before with the bmw traffic light videos. None of your videos are tests, they’re highlight summary advertising-type videos with questionable funding sources.

We need raw videos of reasonable length like the ones from JJRicks or pretty much any AP video.

 

[mikes_fsd]

Lucid has Dream Drive.

Oh, this one is rich!
Lucid - one that has sold 0 cars to consumers - has "Dream Drive"..
Yup, they're dreaming! Just proving @powertoold 's point

 

[powertoold]

My intuition tells me that actual owners don't use the ADAS from many of these companies because they aren't consistent or comfortable in real world use (jerky movements or turning off for no reason or ping ponging). That's the main reason why there isn't as much YouTuber excitement or content for them.

Tesla doesn't actually advertise Autopilot. Most of the public perception and opinion of Autopilot is from viral events or YouTuber testing videos. And the reason why there have been so many Autopilot videos is because people actually use it often.

 

[Bladerskb]

Again, show me owners’ stress tests. We’ve been here before with the bmw traffic light videos. None of your videos are tests, they’re highlight summary advertising-type videos with questionable funding sources.

We need raw videos of reasonable length like the ones from JJRicks or pretty much any AP video.

I too would be vehemently against independent and repeatable tests if it will prove that the system I preach saying its 5 years ahead and developed using "billions of miles of data" is actually inferior to other systems....oh wait i'm not biased or emotionally clinging to a company.

 

[diplomat33]

Oh, this one is rich!
Lucid - one that has sold 0 cars to consumers - has "Dream Drive"..
Yup, they're dreaming! Just proving @powertoold 's point

Laugh all you want. Lucid Air is entering production this year and "Dream Drive" is real. The cars will also have OTA updates so Dream Drive will get updates just like AP gets.

@powertoold is wrong. There are many auto makers from Nissan, Toyota, NIO, GM and many others with systems as good as AP.

You and @powertoold can live in your little Tesla fantasy bubble where Tesla is the only company with "autopilot" if that makes you feel better.

 

[powertoold]

many others with systems as good as AP.

No, there aren't. Unless you define good as having some similar features.

Again, people keep saying this or that is good. Well, why aren't people using them and testing them? Why are all the videos about these other companys' ADAS systems highlight / marketing type videos that don't allow you to make your own conclusions. Why are they always making the conclusion for us?

I question any fsd or ADAS video that tells me what they want me to know.

 

[Bladerskb]

No, there aren't. Unless you define good as having some similar features.

Again, people keep saying this or that is good. Well, why aren't people using them and testing them? Why are all the videos about these other companys' ADAS systems highlight / marketing type videos that don't allow you to make your own conclusions. Why are they always making the conclusion for us?

I question any fsd or ADAS video that tells me what they want me to know.

Yeah people with their entire life savings, college tuition and every paycheck in the TSLA stock.
Making daily/weekly video to either absolve Tesla of wrong doing, to prop up anything tesla does, promote the stock and to attack other automakers.

Clearly an unbiased source of independent information.

And..

Claiming every month that you basically already have and will have L5 autonomous driving and hyping up each update on twitter to over 45 million people clearly is NOT marketing or advertisement.

 

[Bladerskb]

My intuition tells me that actual owners don't use the ADAS from many of these companies because they aren't consistent or comfortable in real world use (jerky movements or turning off for no reason or ping ponging). That's the main reason why there isn't as much YouTuber excitement or content for them.

Oh wait i forgot. This is a fake marketing material and is an obvious lie. No one uses NIO Pilot. I forgot that every company lie except ofcourse the angel Tesla.
Even Jesus is like a fithy rag compared to Tesla. All praise Tesla. The Holy One. The one that lieth not!

 

[powertoold]

Oh wait i forgot. This is a fake marketing material and is an obvious lie. No one uses NIO Pilot. I forgot that every company lie except ofcourse the angel Tesla.
Even Jesus is like a fithy rag compared to Tesla. All praise Tesla. The Holy One. The one that lieth not!

Lol, geez, I guess people don't own NIO stock or GM stock or Nissan stock or IBM stock. Why is it so hard to find videos from actual owners?

Yeah people with their entire life savings, college tuition and every paycheck in the TSLA stock.
Making daily/weekly video to either absolve Tesla of wrong doing, to prop up anything tesla does, promote the stock and to attack other automakers.

That's funny because people were making Autopilot videos when Tesla stock wasn't worth 1/20 of what it is now.