I think it is a good idea for them to be able to navigate without HD mapping. Suppose a road is damaged or there is an accident and the AV has to take a route that it has never been on before. Or Storm damage causes routes to be blocked and the vehicle has to travel 10 miles out of the way just to get back on the right route. Since these vehicles are supposed to replace personally owned cars, they are going to need to handle finding their way when losing the connection to the Mothership or navigating after suffering damage due to Hail. It will be interesting to see how these cars handle severe weather. Dallas Texas gets quite a bit more severe weather than San Fransisco or Phoenix.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Waymo
- Thread starter Daniel in SD
- Start date
EVNow
Well-Known Member
EVNow
Well-Known Member
*ignores videos of humans driving in the rain and heavy fog and claims camera alone can't do it.
Its the same old argument. Move on.
Doggydogworld
Active Member
They'd be much less safe in these situations if they blindly drove the map "on rails" as the cult claims.
But you and I both know that's not what happens. The map/perception mismatch triggers these cars to slow way down and proceed much more cautiously. Just as you do when entering a construction zone, an accident site or a downed tree. At least I hope that's what you do.
My Tesla copped some sun glare yesterday & ran across the white line to the other side of the road. Simple simple stuff.
Cameras do not equal eyes. Tesla will never have robotaxis with current setup
bkp_duke
Well-Known Member
You are comparing the stale Autopilot codebase (circa 2017 with VERY few updates since then - all dev efforts have been on FSD) to solutions that are in constant development (FSD and Waymo). Australia currently is in the FSD group, so you are stuck on a 5 year old system.
I know, I have cars with both (AP on the 3, FSD on the Y). The AP codebase is very outdated and needs to be replaced, requires constant interventions when we drive the 3, it is not representative of what a vision-based system can do. The Y we have with FSD 11.4.2 is doing extremely well. Sunny day here, just completed a 90 mile round-trip on FSD from San Diego to Orange County, no interventions. Plenty of "glare" available today.
EVNow
Well-Known Member
A 2 year old has the same vision as a 18 year old - but can't drive.
The state of AP years ago or FSD today do not and can't tell us what camera only can eventually do.
I'm confident camera only will be capable of FSD - when that happens and who will make that happen is the open question.
diplomat33
Average guy who loves autonomous vehicles
EVNow
Well-Known Member
After 13 years - you'd think they have seen enough rain. Apparently not. Its like a teenager figuring out how to walk while chewing gum for the first time
diplomat33
Average guy who loves autonomous vehicles
Key topics in this conversation include:
- The absence of unreasonable risk for automated driving systems
- Waymo’s layered approach to safety, including hardware, the ADS behavioral layer, and operations
- The dynamic approach to safety
- Waymo’s credible approach to safety
- The importance of collaboration and a common language amongst the automated driving community
- IEEE Std 2846 and reasonably foreseeable behaviors of other road users
EVNow
Well-Known Member
Don’t worry we have 9 months of rain here and rarely does the FSD not work.
Your analogy makes no sense. Seeing enough rain =/= having the capability to do so "reliably" and safe enough to allow it to do sans a safety personnel.After 13 years - you'd think they have seen enough rain. Apparently not. Its like a teenager figuring out how to walk while chewing gum for the first time
2 years ago, the 4th gen Waymo would not drive in the rain at all.
Now the 5th gen Waymo will drive in rain without a safety personnel.
That does not mean you stop testing in different regions with different climate and precipitation pattern.
7 Years ago, Tesla released their "Self-Driving" Demo video.
Tesla still cannot drive without someone supervising. In 7 years, a teenager (now an adult after 7 years) should be proficient in driving by themselves in clear weather. Has Tesla with the billions of miles driven and enormous data collected not seen enough driving to be able to do so safely without someone supervising?
Last edited:
EVNow
Well-Known Member
Then you are welcome to ignore my posts. You can keep thinking Waymo is great and progressing at a terrific speed - I've not drunk the Waymo Koolaid and can see what's happening.
You're making a strawman argument. I have never made the argument that autonomous driving will be solved completely in this decade. Elon Musk said it was already solved 4 years ago; you drank the Kool-Aid.
They are progressing as fast as the available technology allows them to in the process inventing new technologies as they go. Tesla is not progressing any faster than Waymo, in fact they are behind Waymo in timeline while using much of technology Google invented in ML.
We've had this discussion before.
So again, according to your numbers Teslas has 360 million miles of driving data per year vs 2.5 million waymo miles yet is still at a level of teenager still learning to drive with supervision.
bkp_duke
Well-Known Member
It's the rate of advancement, the absolutely most critical point, that you left out.
6 months ago, FSDb drove like a 6 yo. Now, arguably a 16 yo after a few lesions. It's the rate of advancement that is crucial, and as the training of the neural nets advances, that will accelerate at a far greater pace than other solutions can keep up with.
EVNow
Well-Known Member
But Waymo / Google execs did. Search my old posts for reference.
Similar threads
- Article
