No, he specifically said regulators have to agree to FSD to be used with human oversight. So, even release of BETA depends on regulators.
Listen @10:53
Last edited:
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable
- Thread starter AudubonB
- Start date
Listen @10:53
Sean Wagner
Member
Just the right spot to bring the following vid up [so few views!] [also, the ark interview was informative]:
Cruise Automation: San Francisco Maneuvers | 3,116 views | Published on Jan 25, 2019
Our driverless cars constantly encounter challenging situations on the streets in San Francisco. The driving seen in this video is 100% autonomous. This video is sped up approximately 2.5x. Do you have what it takes to solve challenging problems like this?
Cruise Automation: San Francisco Maneuvers | 3,116 views | Published on Jan 25, 2019
Our driverless cars constantly encounter challenging situations on the streets in San Francisco. The driving seen in this video is 100% autonomous. This video is sped up approximately 2.5x. Do you have what it takes to solve challenging problems like this?
No, this is using the new Back To The Future mode discovered in the app.
That's pretty impressive...but..those weird pauses seem sure to get a car rear ended.
It like a person behind would assume the car was going to go right away yet it pauses for that extra second.
In the ARK podcast with Elon, I was pleased to hear that AutoPilot "disengagement" events are summarily passed on to Tesla. Thinking of our own personal "disengagements" while driving the Model 3 with AutoPilot activated, this should indeed be a rich source of training data. It's also nice to think that our abortive attempts to engage AutoPilot on our local mountain roads truly are contributing to future improvements.
Personally, my guess regarding the SR is that, while there will be a new pack architecture, the SR will have at least 240 miles of range and may replace the MR. I'd like to see the SR at least equal the EPA range of all of today's lower-end, longer-range EVs from the legacy OEMs.
I find this very impressive! My enthusiasm for TSLA in the vehicle autonomy space is moderated by the fact that autonomy is a very hard problem and other companies are also working hard at it. That said, when it comes to autonomous, full-speed highway driving, it appears that Tesla has the most advanced tech.
Actually if you followed things he made that story up to see the reaction he would get and to see who his friends were.
Perhaps the guys who interview the guy who interviewed Elon will be interviewed by Elon thus completing the circle.
Yes, seems like he is saying that by the end of next year it is pretty likely that FSD will meet Tesla's idea of safety. Regulators will take longer. He used the term the "march of 9s" which seems to me to be the argument of how safe is safe enough for the regulators.
He indicated that the new NN board will support full frame at full resolution on the cameras with room to spare. Chip/board development has been in the works for 3 yrs. Interesting that he said that any exit from autopilot resulted in the video being uploaded and he was careful to include that there was no attribution with the video so what car or what driver was not part of the upload.
Sean Wagner
Member
The complexity of the situations mastered by Cruise's system really should get people in the "never gonna happen" brigade thinking. I don't understand how this isn't more "popular". Allows some inference concerning people's perceptions in general, which would also apply to Tesla?
As @anthonyj once pointed out, he rolled his position to long term calls and is playing in Supper Bulls thread Super Bulls Only
Right and just like voice recognition is miles ahead of what it was a few years ago the tech will keep getting better.
humbaba
sleeping until $7000
The sped up video rate makes the cars seem much more confident by reducing the perceived time. And yet, it still is slow going by construction as it is apparently hunting for the first opportunity to return to lane. I admit, I didn't watch the whole video -- I stopped as soon as they cut scenes. Apparently the car can't drive without disengagement for long enough to make a video to demonstrate a complete trip.
Cruise, like Waymo, is unimpressive. Other than the large amount of money being burned on it. I've noticed GM likes to include future payments into Cruise from investor companies that are contingent on meeting milestones that they are apparently having difficulty completing. Sorry, I don't have citations handy, but the overall impression is that GM wants to present how much money might be invested into Cruise by outside companies as if it was already invested.
I really would like to see progress from a company other than Tesla, whether that is EV, energy or autonomy, but despite splashy press releases there just doesn't seem to be much in the way of substance.
Nice. I like ARK, I like Elon, etc. I thought the podcast was worth a listen. But the home-team softballitudinalness of that interview was a little cringe-inducing even for me.
Sean Wagner
Member
I have a different take: they compiled diverse difficult scenes into one youtube video to demonstrate their system's current capabilities. Watched the whole thing twice, once at speeds roughly matching real time. I'll gladly repeat I find it impressive. My point is less about singling out Cruise than being able to get an idea of what's possible today.
Last edited:
It seemed to me that is was running:
10 Move toward the gap *
20 Stop
30 If Something_Moving == 1 goto 20
40 Goto 10
*Lidar's strength...
humbaba
sleeping until $7000
I agree both that the MR is likely to be replaced and that the range is likely going to be greater than the initial estimate. About the only way I can see the MR being retained is if the SR is introduced at prices below $35k, or the LR and MR both get significant range increases. Which is possible, but while giving people many options (SR, MR, LR; PUP, non-PUP, performance) is popular in some quarters, overall sales tend to be higher when consumers have less choice because it eases the decision making. Which is one reason I doubt that the MR has legs -- AFAICT, Tesla has kept a trim selection for vehicles.
humbaba
sleeping until $7000
I've probably watched that video before, though I don't have a specific memory of it. I've read material put out by GM on how great it is and why they think so. @neroden, who is at least as much of a skeptic about FSD as I am, has made positive comments about Cruise. So, yeah, I get that some people find their demonstrations impressive. But GM has consistently had to resort to cherry picking and careful framing to give those positive impressions.
Now, I think GM Cruise is better than this, but if you put a million monkeys to typing and got a novel out of it should someone view the utility of having a million monkeys banging on keyboards from that one novel, or should the entirety of the garbage that was output be considered?
When the best GM can show is a hesitant autonomy that works on its own over short stretches then I'm left concluding that they can't manage anything longer or they would be bragging about it. When they talk about all the issues facing autonomous driving, I'd be more impressed if they demonstrated that they had overcome them rather than trying to leave the reader with that impression.
It has every appearance of a lot of hot air with managers desperate to hit their milestones to keep the money coming in. And investors getting a bit skittish doesn't help that impression either.
Similar threads
- Locked
- Locked
- Locked
