![]() | |
A lot to unpack here, I also used to work in self driving and still have friends who do Full autonomy with some edge cases is here.... with Waymo/Cruise. Waymo/Cruise/MB take liability, Tesla does not. Type of data matters - Tesla will never hit unsupervised city driving because of sensor placement and type of data. FSD does very well in Vancouver/DC where layout is reasonable but even 12.5 shits the bed in SF/LA/NY/Boston Predictions 1) Level 3: MB Drive Pilot & GM Super Cruise release 80mph hands off, eyes off highway and take liability by end of 2026. 2) Level 4: Cruise partners with Lyft, Waymo with Uber for rollout across USA by end of 2025. 3) Level 5: Robotaxi does not get Cali DOT approval for testing until mid 2025, will have steering wheel & pedals to avoid FMVSS recert, will have 4 seats, and new sensors on corner of car and either radar/lidar for fog/rain/low light. Probably a modified Model 3. |
Quote:
|
It's not trolling, they genuinely crash into things like ambulances and airplanes because of how they're set up. |
^ everything constructive or critical about Tesla ... Hehe thinks you must be trolling. |
Quote:
|
Quote:
LA: Waymo just launched. Tesla struggles cuz no sensors on side of car, can't reliably creep for an unprotected left onto main stroad. Boston/Seattle: I think Waymo will launch here to test inclement weather. NY: No robotaxis for now, maybe limited testing to improve response to environment. I have a hypothesis that FSD shits the bed with city path planning because it runs out of compute. |
Slow speed city driving seems easier, or at least safer. SF has a ton of driverless cars roaming around downtown for some brand I can't remember the name of. It's a car-sized object getting confused at highway speeds that's sketchiest. Quote:
|
It's rly just radar that's bad at detecting stopped objects; Tesla's radar in older models was especially low res and short distance. Both LIDAR and vision are better at this but both also really struggle in bad weather. The LIDAR point clouds in rain/snow/fog are really noisy, needs work. |
The SF/LA ones are Waymo which are available to general public now. Usually a couple bux cheaper than Uber and Lyft. Really awesome for taking a nap in. https://waymo.com/intl/es/waymo-one-san-francisco/ https://waymo.com/intl/es/waymo-one-los-angeles/ |
I took a WAYMO in SF in June 2024. It was pretty safe. We were picking up at Fishman Wharf. Alot of traffic and pedestrians. https://i.postimg.cc/7YsggsLt/Whats-...8-6a03b770.jpg https://i.postimg.cc/vBTzRJDn/Whats-...0-1e7bb9ba.jpg |
$1900/month to insure a CT with $5000 deductible |
Quote:
|
Private insurance for thew win!!! |
Tesla earnings report yesterday was quite impressive. Stock went up almost 22%,but im sure it will drop soon like it always does. |
On my prediction: Level 5: Robotaxi does not get Cali DOT approval for testing until mid 2025, will have steering wheel & pedals to avoid FMVSS recert, will have 4 seats, and new sensors on corner of car and either radar/lidar for fog/rain/low light. Probably a modified Model 3. Elon has confirmed a lot of this 1) ✅ Steering Wheel and Pedals: Cybercab "volume production" after 2026 means they gotta start in late 2025; they promised but didn't show roadmap for Unsupervised FSD in 2025 so imo they're not gonna deliver 2) ✅ New Sensors: Elon said they were having trouble identifying errors / validating the training data they collected, implying they need new categories of data to help. HW4 has added a radar. 3) ✅ New Hardware: Per my friend who worked on HW3, it will have a very different power structure to HW4. There is 0 chance HW3 cars get retrofitted for HW4 and there is likely going to be a HW5. 4) Timeline: IMO they're going to need HW5, some more new sensors, we'll see some limited rollout of unsupervised FSD in late 2026. |
Quote:
|
Robo taxi = model 2? Throw a hatch shell on it, same platform. Sell for $3xxxx? Profit |
They also announced the model 2 is cancelled My best guess is they will use Robotaxi to prototype some new sensors and changes based on Model 3. HW5 model 3 + new sensors will finally have FSD unsupervised in 2027 No one is going to release a new EV at 30k when Equinox is already there and a used Taycan is 45 already |
who makes the batteries for taycan? i know quoted range is already very average i wonder if degradation is comparable to tesla (which is pretty good?) |
Not my area of expertise, but AFAIK there's diff types of failures/degradation we see: - Lithium plating: if you charge too fast or at too low of a temp, this reduces lithium but can also cause short circuits (TBH the GM/Hyundai/Porsche packs have me worried with 350kW) - Normal degradation: cathode and anode degradation caused during charging and whatever - Overcharging / discharging / whatever: something to do with cell balancing and the BMW |
We also seeing motor/drive unit failures - the Valeo? motors in EQB, Q8 eTron, and some others have issues with these, as do every pre 2022 Model S with the large rear drive unit |
Quote:
|
Quote:
|
tbh yeah but if Tesla is smart they'll split highway and city. By end of 2025, I think they can do Level 3 highway on HW3/4. Maybe around 2027-28, they'll get Level 3 city on HW5 then backport. They never said you can take a nap / defined which level of autonomy, which kinda feels more level 3 than 4. Now with Mercedes testing Level 4 on city roads China, I'm pretty sure i'll nap in a MB/Waymo in DC/NYC/Boston before Tesla allows me to nap in any city. https://electrek.co/2024/08/06/merce...tech-in-china/ |
Quote:
|
All times are GMT -8. The time now is 11:53 AM. | |
Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2025, vBulletin Solutions Inc.
SEO by vBSEO ©2011, Crawlability, Inc.
Revscene.net cannot be held accountable for the actions of its members nor does the opinions of the members represent that of Revscene.net