I think it's crazy for reasons other people state, especially latency and connection reliability issues. But I think a related idea can work pretty well.

Self-driving cars are classified into 6 SAE levels, 0-5. 5 is full autonomy, 4 is autonomy in limited conditions, 3 is self driving but there has to be a driver ready to take over.

Getting to 5 is really hard. It could take decades, because there are a zillion edge cases. But what if you could turn cars loose at the "can handle 95% of driving" level and then let it call a human to give guidance when the car is nervous or confused?

Obviously, the phone-a-friend approach can't work for things requiring quick action. But if you bias the car to just stop safely and wait for orders, I think it's doable. E.g., the car comes upon an accident where the lanes in one direction are blocked. Humans know to go around the accident by slowly and carefully sharing the remainder of the road. So perhaps the car stops before the obstruction and asks for help. A human looks at it and draws in the path it should follow to get back to normal road.

So I think remote "robot coaches" could plug the gap between levels 4 and 5. And maybe some of the level 3 gap, too. And from what I hear of the Waymo operation in Arizona, I suspect Google's doing something akin to this.

I've thought something similar for a while, and it's been long enough that I think the problem is the failure mode. It might be that getting a car to recognize when it's encountered an edge case and needs to kick it to a human is about as challenging as just getting the car to come to the right conclusion and drive itself anyway. Intuitively we think of the problem as what the car needs to do, but the main problem is ascertaining the state of the road accurately. In the autonomous crashes we've seen from Tesla and Uber, it looks like the cars never even noticed enough to have an ambiguous condition they could deactivate autopilot from. They just had a completely erroneous perception of the outside world.

And I think that's due to a "let's keep driving" bias. In Uber's crash, for example, it detected an object moving in the road, but didn't understand what it was. Any reasonable human driver would have quickly stopped until it was clear what was going on. But the Uber car just kept going at speed until it was sure that it was a pedestrian, at which point a collision was inevitable.

I think the right thing to do is to invert the bias. If the car isn't sure it understands what's going on and that it's safe to proceed, it should stop. In which case, development becomes a long series of teaching about rarer and rarer circumstances, so it needs less and less human input to understand the road.

In the case of the Uber crash in Tempe, the car did kick the decision over to a human operator. However, it did so 1.3 seconds before impact. This wasn't enough time for the human to assess the situation and decide what to do.

We're talking about two different things. Uber's approach was horrific in many ways, and I agree that's bad. I'm talking about a car that is advanced enough to drive and stop safely, but may sometimes not be sure what to do, especially in situations that require social cognition.

There are many places where cars have to go with no cellular data coverage, even inside major cities. Tunnels for example.

Starlink won't grant Internet coverage inside tunnels or underground, but it plans to blanket all of the US with Internet coverage making that far more possible than with terrestrial cellular data networks.

I'm sure Tesla has been keeping that in mind as their self-driving technology progresses.

Sure. And you'll note that Waymo doesn't go anywhere like that.

The easy solution to this idea is to have the guy inside the car (you know, the passenger) pick up in this case. He's right there - no latency problems. The difficulty is in correctly identifying an unresolved edge case with enough margin to brake safely and draw the attention of a distracted passenger.

Sure, but then you lose the self-driving illusion, which is a big part of what's being sold. By all means have that as a backup mode if it can't phone home. But trained remote operators will not only preserve the fiction, but they'll be much more effective and they'll be the source of a lot of good feedback for the dev teams.

I would not get into a car driven by someone in a "low cost" country whose only incentive for not getting into an accident is getting fired.

No, they hire 3 drivers in "low cost" countries and the car controls respond to the majority vote, for built-in redundancy. If you buy the premium subscription it's 7 drivers instead.

They might also care about not killing people, but fair enough I guess.

Not as much as if they were actually in the car.

Having been in a taxi with a driver who bragged about staying up all night on stimulants, I disagree.

What about DRUI?

This is an interesting case - probably the only time your lawyer would RECOMMEND blowing! :)

Latency is literally a killer problem in remote operation. For example, the US Air Force operates Predator drones all around the world remotely from a base in Nevada. But for takeoff and landing (the most hazardous phases) they usually hand over control to another pilot located at that airfield.

The only surprising thing in your statement is the fact that the drones can’t land themselves. Even large commercial passenger aircraft can do that. Nobody dies if a drone cocks up on landing.

Very few commercial passenger aircraft have autoland capability, and even those that do have two human pilots ready to take over instantly in case of system failure.

The first part of your statement is not true. Autoland as in "press a button and go get a coffee" sure. Almost all commercial aircraft have systems that can land the aircraft themselves (with the pilot ready to take over at any point), but it is not often used in practice because:

- pilots still need to know how to land an aircraft themselves, so they may as well practice it (it's not like they are driving to work and can read Facebook)

- to prevent interference with the radio signals used for the ILS, ground aircraft are routed away from transmitters which reduces capacity of the airport when it has to be used

- the system isn't perfect, so from a safety perspective it's safer to use it as guidance rather than rely on it

Autoland is a requirement during severe weather conditions - when a pilot cannot obtain enough visual information to perform the landing themselves.

1) Ground aircraft are always kept out of the ILS critical zone when there is IFR traffic landing, which happens to be all commercial flights and almost all business aircraft. Autoland being used has nothing to do with this, besides, even GA planes can autoland today with RNAV GPS LPV approaches.

2) Autoland is never a requirement — usually, deciding to try an autoland landing or diverting to an alternate airport is the pilot’s game time decision.

3) Not “almost all” commercial aircraft have autoland capability. It requires specific certifications for the pilot and crew as well as the aircraft itself.

>Not “almost all” commercial aircraft have autoland capability. It requires specific certifications for the pilot and crew as well as the aircraft itself.

The 737NG, 737 MAX, 747, 757, 767, 777, 787 and every Airbus model have autoland. It's a requirement to be able to use autoland in order to be certified as a pilot for a particular type. Garmin even has a system for newer GA planes so they can autoland nowadays. Autoland is used rarely because its more work for the pilots to enable and they all think they can do a softer landing, but in very low-vis weather conditions its used pretty often.

They have had the capability for decades back at my first job on campus at Cranfield University (which has its own airport)

BAE where doing tests of this lading passenger jets on wet and slippery runways for example

So the drones has a few strains, Predator is a frontline ones akin to A-10, reportedly uses some Cisco VPN gears in it. There are Global Hawk, which is basically modern Air Force U-2, or X-37B or X-47B, controlled like stah-tegic wargame piece, and how each ones fly are different.

For whatever purpose they fly Predators manually.

The primary issues here are the latency, throughput, and availability of information to this remote driver.

* The remote driver is seeing things 10s-100s of milliseconds after they've happened * And somehow we're getting all of the data coming out of sensors on the car at high enough fidelity over the air * And we're not having connectivity blips that would result in loss of remote control for seconds

There was. They are doomed. It's a bad idea.

https://www.freightwaves.com/news/exclusive-mass-layoffs-rep...

Among many other issues is connectivity, in the bay area on 280 or 101 there are dead zones and when there's heavy traffic all those cars/people/phones connected to the towers and swapping as they move along overload the connection at each transition point. It wouldn't be possible today to reliably maintain a connection for such a human critical situation. Even with 5G a small interruption would likely be too risky for consumer projects.

Welcome to https://phantom.auto

They have a clever market strategy but those vehicles don't operate on public roads.

That's incorrect; Postmates, for one, uses Phantom Auto (https://www.forbes.com/sites/alanohnsman/2019/09/23/postmate...), and they definitely operate on public roads.

There's a bit of difference between a 100lb Postmates vehicle moving at 3mph and a 10-ton truck moving at 65mph.

The former could annoy you, the latter could kill your entire extended family.

They sure do, or at least have operated on public roads:

https://spectrum.ieee.org/cars-that-think/transportation/sel...

Yes, this is called remote operation or teleoperation and many people have attempted it. But as far as I can see, they all hit against the latency/bandwidth/availability barriers. The hope is that 5G will help.

Currently, everyone is vaporware. Including Phantom.

Bit scary due to latency and connection reliability. If the connection breaks you’re stuck, or worse. If it breaks when you’re on the highway at speed, or on a twisty road, or in a tunnel...

Latency.