DanCar
Active Member
Volvo agreement is just to design Waymo vehicles for ride sharing.
Waypoint - The official Waymo blog: Partnering with Volvo Car Group to scale the Waymo Driver
Waymo said:
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FSD rewrite will go out on Oct 20 to limited beta
- Thread starter diplomat33
- Start date
Waypoint - The official Waymo blog: Partnering with Volvo Car Group to scale the Waymo Driver
Wow that was a lot of posts to read through. For myself, I don't see Tesla's "HD" maps anywhere equivalent to Waymo's level of detail in their maps. I don't imagine Tesla is running down every road mapping out features using vision as someone mentioned.
That said, I have no problem with Tesla using some sort of enhanced map to improve the FSD experience.
Green, what do you think happens for visualizations for MCU1/HW3 cars?
That said, I have no problem with Tesla using some sort of enhanced map to improve the FSD experience.
Green, what do you think happens for visualizations for MCU1/HW3 cars?
Once the car starts doing most of the driving, you're entering L3 territory. Tesla, from what I can see, is clearly aiming for L3 at the very least. If they're serious about robotaxis, then that would require L4 at a minimum. Most of the ground work for what would become Waymo started way before with the Google self driving cars, which IMO, were pretty much L3 cars until they switched to geofenced driverless operation (L4), after which Waymo was born. So technically, they did not skip L3. As long as a driver was needed to intervene as necessary in their cross country or city drives, it was not L4...yet.
Now, don't get me wrong, in terms of autonomous miles driven vs disengagements, Google / Waymo is still far ahead. However, Google was already working on self driving tech since 2008.They had a six or more year head start on Tesla.
no. Common definition of L3 is when you can remove your hands from the steering wheel and stop paying attention, but be available to help the car when it asks (in other words, the car knows when it needs help, not like right now where you need to babysit it with undivided attention).
Mo City
Active Member
En route to L4, we don't know yet if Tesla will ever give drivers permission to not pay attention at all times. They may decide L3 is too risky for them.
AlanSubie4Life
Efficiency Obsessed Member
I guess I am trying to envision a situation in city driving where L3 would work. Can you describe it?
I think there is probably a reason why people don’t talk much about L3 in this use case.
I don’t want to derail this thread by talking about about the differences between L3 and L4, though, since it is EXTREMELY boring. Maybe you meant we are getting into L4 territory at the point you were describing?
they are shipping the new godot-based viz for mcu1 as well, so at least in theory it's going to work too. Not that I have an mcu1 car to actually try it and see how it performs.
I wish people would be a bit more clear with what they mean rather than argue that their semantics was somehow not technically incorrect, wasting energy trying to defend self conceptions. Listen to this guy at 18:00
Back to the question. I would be very surprised if in version 2020.48.8.11 that any form of map is input to the same neural network that takes the cameras and radar and later has its outputs visualized on the display. I believe some neural network might be using some map data and outputs from that NN to decide which lane to change to.
Back to the question. I would be very surprised if in version 2020.48.8.11 that any form of map is input to the same neural network that takes the cameras and radar and later has its outputs visualized on the display. I believe some neural network might be using some map data and outputs from that NN to decide which lane to change to.
But, how do we know where Tesla is actually going with this?
L2, L3, L4, and L5 are all very different directions. We like to see them as stepping stones, but how you develop for one isn't necessarily how you would develop for the other.
L2 is not autonomous driving, but failing is sometimes the goal. Sometimes people purposely fail to win the long game.
L3 is clearly not the goal. You certainly wouldn't be messing around with city streets if freeways were the goal (what L3 is meant for)
L4 is my choice of a goal, and it seems pretty popular on in this section of TMC. But, I don't seen any evidence from Elon/Tesla that this is even on their radar.
L5 is actually what Elon mentions when he talks about FSD, and the way Tesla determines path planning from vision is an L5 requirement. As in if you don't have that you don't have an L5 system. An L5 system vehicle needs to be able to go on roads that the system hasn't seen yet.
L5 is often rejected as a fairytale, and I agree that it is. But, that doesn't mean Tesla won't use the failure to achieve that to leave the system on L2.
With the FSD beta Tesla knows regulatory agencies are watching. They know at anytime between now, and the wide release of city NoA that the regulatory agencies can put a stop to it. Without billions of miles of testing there is no L3, L4 or L5. Without the fleet doing the testing for you then you lose a key advantage Tesla has.
It's going to be an exciting, and dramatic 6-12 months that will decide the fate of FSD on HW3
Tesla is working towards L5. Elon Musk is a bullshitter, but not much of a liar.
They just work on it knowing that any improvement increases the value, rather than targeting L3. L3 is kind of pointless, when good L2 is the same as L3 if nobody's looking. Hence the addition of code for the internal camera, trying to be in a position better to satisfy regulators.
L4 is where the value really increases, but L5 is the ultimate goal, since that means cheaper and more profit.
mikes_fsd
Banned
It seems " murky " is how you prefer it.
This way you can flip-flop on the definition to your heart's content and then be flabbergasted and somebody calls you out on the BS.
For me, this conversation is very enlightening to your thought process. I hope it was helpful for others as well when taking your "revelations" into account.
Have a good life!
The technology choice tells you everything you need to know about their intention.
The fact that they don't use Lidar/HD Maps, shows that they are all in for Level 5 based on vision. If the goal at any point was level 4 , then you would have HD maps and a geofenced area.
Level 2-5 is nothing more than who has liability. You could have a car with zero disengagements for 5 million miles, and if the company won't take liability if something happens, it will still be classified as a level 2 system.
The fact that they don't use Lidar/HD Maps, shows that they are all in for Level 5 based on vision. If the goal at any point was level 4 , then you would have HD maps and a geofenced area.
Level 2-5 is nothing more than who has liability. You could have a car with zero disengagements for 5 million miles, and if the company won't take liability if something happens, it will still be classified as a level 2 system.
Even better than the George Hotz one, this is the best interview on Lex channel about the topic. Jim Keller was very close to the effort at Tesla and is one of the best engineers on the planet. Especially his comments about things in the real world being "really damped in terms of rates of change" compared to the calculations of computers and NN algorithms is a confidence builder.
I hope Lex has Karpathy next. The next few months are going to be so interesting..., the rate of change in improvement of the FSD build seems already mighty quick. I wonder were we will stand in a year, this might just really be the big one
clydeiii
Member
lol @ "acolytes" but it seems far more likely that you could procrastinate far more effectively by letting good be the enemy of perfect, no? "We can't take action until we have ALL the data." "We can't allow FSD on public roads until Tesla has proven it to be 100% safe." We could procrastinate for decades with that attitude.
I don't have the FSD Beta but on 2020.40.8, I've noticed this behavior too. My car stopped well short of the stop line and then lunged forward twice to reach the stop line.
As I replied to someone else, in this context we need to proceed as even beta autopilot saves lives
The parallel is to vaccines - we require guinea pig humans to be infected & tested on to ultimately save lives
Similar threads
- Question
