-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Autonomous Car Progress
- Thread starter stopcrazypp
- Start date
-
- Tags
- Autonomous Vehicles
Apology if this has been discussed before, but has any autonomous system demonstrated ability to use traffic curve/blind corner mirrors to detect incoming traffic? Or would the mirror confuses the car instead?
Something like this:
Something like this:
Daniel in SD
(supervised)
That would be amazing. I doubt any of them can do it. I have noticed that many prototypes have sensors all the way at the corners of the vehicle so they can see around corners better than a human in the driver’s seat.Apology if this has been discussed before, but has any autonomous system demonstrated ability to use traffic curve/blind corner mirrors to detect incoming traffic? Or would the mirror confuses the car instead?
Something like this:
Doggydogworld
Active Member
Robos have two possible models - Uber replacement and personal car replacement. Cruise aimed at Uber replacement from day one. That's why they focused in Uber hotspot SanFran.
IMHO Waymo originally wanted to go after the big fish -- car replacement. They wanted to start by replacing 2nd and 3rd cars in suburbia. If half of Phoenix metro's 1.5 million households give up their 2rd or 3rd car that's 750k cars. Say each Robo replaces 10 cars, so you need 75k Robos. Add 10% spares to get 82k. That's why they ordered 82k cars from FCA and Jaguar.
As business models go it was pretty daring. But they got lost somewhere along the way, without getting far enough along to figure out if the model could work or not. Now they are pivoting to Uber replacement, and following Cruise into SanFran.
But their recent moves with their trucking arm suggests they want to instead just sell their hardware to the trucking industry and not focus on having their own fleet. They seem to believe that has a stronger business case sooner. But whatever it is, they are still in the experimental stage. It doesn't appear they are running things like they really expect to make a profit anytime soon (nor do they seem to have a plan to do so).
There was a recent article talking about how they were struggling to scale the installation of their hardware (much of it hand built) and that their huge order that you mentioned above never materialized as a result. Of course Waymo denies it and just says the agreement was "fluid and subject to change", but that seems to at least indirectly confirm those targets may now be lower.
Bloomberg - Are you a robot?
Ahhh... Tesla's UNprotected lefts are equated to Waymo's lefts at signaled intersections.
That sounds about right coming from our resident Waymo booster.
That sounds about right coming from our resident Waymo booster.
diplomat33
Average guy who loves autonomous vehicles
I am not equating the two. I did not even mention Tesla.
I guess I made a mistake calling it an unprotected left turn (the Waymo blog called it an unprotected left turn). But that does not change my conclusion about it being a good turn unto a very tight street where there was no margin for error.
diplomat33
Average guy who loves autonomous vehicles
I guess I made a mistake calling it an unprotected left turn. The Waymo blog called it an unprotected left turn.
But, out of everything the blog said about Waymo's AI, ML, perception, planning, sensor fusion etc... the only thing you got out of it is that I made a mistake calling it an unprotected left turn. WOW.
jimm01
Member
This could get interesting (or at least fun to watch), although I doubt much will come of it.
www.cnbc.com
Section 5(a) of the Federal Trade Commission Act (FTC Act) (15 USC §45) prohibits “unfair or deceptive acts or practices in or affecting commerce.”
Deceptive Practices An act or practice is deceptive
• a representation, omission, or practice misleads or is likely to mislead the consumer;
• a consumer’s interpretation of the representation, omission, or practice is considered reasonable under the circumstances; and
• the misleading representation, omission, or practice is material.
Unfair Practices An act or practice is unfair where it
• causes or is likely to cause substantial injury to consumers;
• cannot be reasonably avoided by consumers; and
• is not outweighed by countervailing benefits to consumers or to competition.
Tesla misleads customers about self-driving features, senators allege in request for FTC probe
Sens. Richard Blumenthal and Ed Markey claim Tesla may not be fully representing Autopilot's capabilities in its marketing messages.
Section 5(a) of the Federal Trade Commission Act (FTC Act) (15 USC §45) prohibits “unfair or deceptive acts or practices in or affecting commerce.”
Deceptive Practices An act or practice is deceptive
• a representation, omission, or practice misleads or is likely to mislead the consumer;
• a consumer’s interpretation of the representation, omission, or practice is considered reasonable under the circumstances; and
• the misleading representation, omission, or practice is material.
Unfair Practices An act or practice is unfair where it
• causes or is likely to cause substantial injury to consumers;
• cannot be reasonably avoided by consumers; and
• is not outweighed by countervailing benefits to consumers or to competition.
Daniel in SD
(supervised)
You didn't make a mistake. It is an unprotected left turn.
You literally only repeated what Waymo wrote. Even the sim shows green on the rendered traffic light. That traffic light means a great deal to human drivers and ADAS software alike.
We wouldn't even know about Chuck's testing with his unprotected lefts if he had a traffic light holding up all cross traffic.
We wouldn't even know about Chuck's testing with his unprotected lefts if he had a traffic light holding up all cross traffic.
Daniel in SD
(supervised)
Is it a green arrow? If not it's an unprotected left. Note the pedestrians (not that pedestrians don't violate crosswalk rules).
It's the same as Chuck's lefts off the highway in front of oncoming traffic (though not as difficult as his left on to the highway).
All this discussion is somewhat ridiculous since a video of a self-driving car doing a maneuver once is pretty useless. The question is, what is the failure rate is and is it better than a human?
Last edited:
Doggydogworld
Active Member
Yeah, trucking is pre-business model so they're experimenting. Regulation will be the gating issue there. That's co-CEO Mawakana's domain.
I'm 100% certain the plan was for FCA and Magna (who builds the i-Pace) to pre-install all the interior wiring and such once Waymo was ready to take delivery in the thousands. They can handle that kind of stuff in their sleep. They don't do it for onsies and twosies, though. So as long as Waymo buys tiny volumes they have to do the teardown/retrofit described in the article.
TL;DR, the cumbersome retrofit process is a result of failure to scale, not the cause.
rxlawdude
Active Member
So, again, where is the "unprotected left" in a scenario even remotely resembling the Chuck Cook videos? Here, unprotected meaning a non-signalled intersection.
Oh, I guess that's why they geofence and dumb the system down for potentially dangerous left turns? That's the "so." You've consistently ignored the fact that Waymo has not demonstrated it can safely perform this manuever, even with its superior sensors.
Or, show us the video. Scenario: a non-signalled intersection, vehicle with a stop sign on cross street attempting to make a left turn on a divided, two-lane-in-each direction road with traffic moving at >40mph.
Definitely it's still an unprotected left turn, but it's not the same level as Chuck's intersection, even with pedestrian traffic. Pedestrians are within a few feet of the car, aren't travelling 35mph so can stop relatively quickly even if cut off, and there is essentially only one relatively narrow pedestrian "lane".
Does it really never make this turn?
Scenario: a non-signalled intersection, vehicle with a stop sign on cross street attempting to make a left turn on a divided, two-lane-in-each direction road with traffic moving at >40mph.
I have randomly watched a couple dozen of JJ's videos in fast forward and the closest I found was a single lane-in-each-direction left turn, not yet a double lane. If it's programmed out of taking them for safety then I suppose we'll never know. I'll keep looking.
I imagine there may be some example somewhere, but given Waymo seems to be programmed to avoid left turns in general, it's also possible the programming makes it so the car practically never has to make such a turn (given usually you can replace it with u-turns, right turns, or a protected intersection. The places Waymo is testing at (at least publicly) also may not have many roads with this type of intersection (a fact discussed upthread).
There's E Warner Rd
S Rural Rd
Doggydogworld
Active Member
I don't think those are in the service area. But plenty others are. The famous conegate video shows one. I've only seen Waymo turn right in such cases.
Waymo does perform unprotected lefts all the time in AZ, just not the particular type of unprotected left that @rxlawdude wants to see. And he's wrong about the left turn in the SanFran video, which was unprotected by definition.
rxlawdude
Active Member
I explained what the definition was in context in at least two posts.
But your point is taken. Let's call it an "unprotected crossing with an unprotected left turn" for the Chuck Cook example.
I'm still waiting for that Waymo video.
Similar threads
- Article
