Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Autonomous Car Progress

This site may earn commission on affiliate links.
Cool, give us some numbers. Tells us why a car that is approaching a construction zone and isn't sure what to do, and slowing coming to a stop wouldn't be able to be remotely controlled. The remote operator would just have to give the software confidence which path to take.
Latency, unreliable connectivity, etc.
The internet and this forum are full of people with conviction and who are completely wrong.

Latency is not a problem when the vehicle is stopped. Sending a single command also makes latency a non issue. Unreliable connection is not an issue because communication protocol can verify the command was received and resent if not.

Another example: A tree has fallen across the road. The right thing to do is a u-turn, but the local software doesn't have that capability. The remove operator sends a command to chose an alternate path that involves a u-turn and helps the car make a slow u-turn if necessary.
Latency is not an issue because things are done slowly. Unreliable connectivity is not an issue because previous command can be verified received before new ones are sent.
 
Last edited:
  • Like
Reactions: alsetym and CarlK
The internet and this forum are full of people with conviction and who are completely wrong.

Latency is not a problem when the vehicle is stopped. Sending a single command also makes latency a non issue. Unreliable connection is not an issue because communication protocol can verify the command was received and resent if not.

Another example: A tree has fallen across the road. The right thing to do is a u-turn, but the local software doesn't have that capability. The remove operator sends a command to chose an alternate path that involves a u-turn and helps the car make a slow u-turn if necessary.
Latency is not an issue because things are done slowly. Unreliable connectivity is not an issue because previous command can be verified received before new ones are sent.

This isnt what's being discussed or proposed by @CarlK you ppl were proposing direct control at 80 mph
 
A day in the life of a Waymo self-driving taxi
If the vehicle encounters a complex driving scenario that it struggles to interpret, it automatically calls in the problem to the response team to weigh in with a solution, which is then shared with the rest of the fleet so Waymo’s vehicles can avoid the area if necessary. Those remote operators are based both in Phoenix and Austin, Texas, but they have no direct control over the vehicle’s operations, Perez said; they just serve as an extra set of eyes for difficult-to-navigate scenarios. “The car might see cones up ahead and could ask for context,” she said. “Should I move to another lane? Should I turn ahead? Should I reroute myself?”
 
@DanCar

I don’t think anyone here is disputing the fact that AVs like Waymo have a remote control mechanism? After all it is a very publicly known fact that they do.

What was or is at dispute is whether this is some kind of real-time functionality that the whole system depends on (ie someone is watching every car all the time remotely) or an issue solving mechanism, which the autonomous system falls back on in rare cases (a predicament at the parking lot for example).

If you are advocating for something in-between, it may have been lost in a conversation about these two extremes.
 
2) There are a bunch of little startup companies like Phantom Auto who make remote control services and try to sell them to AV companies. They made their remote control systems very quickly, maybe it under a year. Waymo took ten years to put journalists in empty cars. Is Waymo just a really slow and incompetent remote control company?

@DanCar You do realize that is literally the company (Phantom Auto) that @SandiaGrunt mentioned and called out here.
 
Self-Driving Cars Have a Secret Weapon: Remote Control

... provide the ability to have a human briefly take control of the car whenever the robot encounters what engineering types call “edge cases.”

This shows it's pretty much mandatory since none of them, Waymo and everyone else except Tesla, has implemented real deep learning which requires huge fleet data to learn all edge cases the car might encounter.
 
Why hasn’t anyone tried to address my questions?

I think most (who have throught this through) of us understand they are not remote control in the typical sense of the word, i.e. they are not saying left/right, gas/brake. At this point, Waymo appears to be 'safe' enough that it won't do something dangerous without knowing it. However, I would expect that at this point in their implementation an engineer can log-in to approve a suggested route IF the car gets 'stuck' somewhere. That sounds like the next logical step to me after removing the safety driver. That also matches with reports over the last couple of months that the car would sit there for a minute or two and then start driving again, presumably this could be an engineer helping sort out a new path.

It is also plausible that part of their roll out plan would be to pull the safety driver and have them monitor (not drive) them remotely. Looking at planned routes, strange circumstances or whatever and then they will start having a safety driver monitor two and then more vehicles. Googles pretty smart, they can likely highlight higher risk maneuvers and pull them up queue for remote monitoring if they thought it was necessary.

Or maybe they just let them be, with no remove access or monitoring whatsoever. This seems like a big step immediately after removing the safety driver. If you have insight into what they actually do, rather than simple what would be stupid, I would be interested in knowing.
 
  • Like
Reactions: DKPowers

nrdarz8h6lv31.jpg

To me the best thing about that video is how obnoxious the driver filming it was being.

Like the point in the video where he got to the left of the vehicle as the vehicle was turning left. Yet, the Waymo vehicle still performed admirably without missing a beat.

As to it being remotely controlled? I think that's borderline conspiracy theory stuff. It makes way more sense to me that it's remotely monitored, and can remotely assisted if some non-real time situation arises. Like if a bunch of people start to block the vehicle on purpose.

As time goes on I imagine those who ride in Waymo vehicles will want more details on the remote control capabilities. My biggest concern as a rider is I'd want to make sure that someone was watching over in some capacity, and I'd want to know what that capacity was.

The best approach to this whole question of remote control is to simply address what they're likely doing, and the limitations of it.

I don't think it's healthy to discuss full realtime remote controlled because that's just dismissing Waymo's accomplishments. I also don't think it's healthy to assume that Waymo doesn't have any remote control capability as I don't think that's realistic either given all the possible edge cases.
 
Last edited:
Right, Waymo has the ability to remotely direct its vehicles.

It does not simply remotely control them 1:1 with human drivers in sweatshops, or whatever the hell else has been argued here.

That would be stupid.

To me it only seem like one person was trying to argue that, and I think we can move on from that.

To me it's not a question that they can remotely monitor, and control them at least to some degree. But, I'm curious about how far that extends.

Like I don't see them having remotely operated safety drivers that are watching multiple screens where they can intervene at any time. I think they've even acknowledged that they don't have that capability. Where instead they can intervene in a stopped situations or when a customer has some concern.
 
  • Like
Reactions: alsetym