Yeah, I think this is going to be a long one. Rodney Brooks, AI researcher and Roomba cofounder, says he doesn't expect a real robot taxi service before 2035: https://rodneybrooks.com/my-dated-predictions/
> When Waymo can get rid of the safety drivers [...] they're getting close to something useable.
I suspect that they can cheat this one a little. As long as they're in an area with good connectivity and the cars are smart enough to pull over if they don't get human guidance, I expect they'll move the backup drivers to a central location. Call it SAE level 3.5. It wouldn't be good enough to sell cars, but it would be workable for a ride service, and would allow them to undercut Uber, etc, quite handily.
I dunno, it's hard to imagine any safe way to drive a car in traffic via cell. I don't see how you could get all-round vision with acceptable resolution, frame rate, and latency. VR might solve the viewing angle, at least, but would greatly magnify the effect of frame rate/latency drops. You don't want your driver getting VR-sick five minutes in.
And that's leaving aside reliability. A 99.9% solid connection is not nearly good enough.
I think you misunderstand. I agree that a safety driver being expected to be able to pay attention and being ready to take over at any moment isn't viable.
But consider a point where the cars are good enough that the remaining risks are ones where they can show that they can tune the hazard detection so that it doesn't miss any potential hazard situations, but may occasionally be overly cautious, and that it can reliably and safely slow down or stop short of potential hazards it doesn't know how to handle.
In that case you might get to a point where it's ok (safety wise, if not in terms of customer satisfaction) if the car stops for 30 seconds until a human safety driver reviews the data and confirms that what the car "sees" is not a dangerous situation.
In that case you might have e.g. 10 cars per safety driver, or more, and most of the time the car might not even stop - if a driver is available to respond immediately it may be sufficient for it to slow down until it gets a response. And you can simply slowly reduce the number of safety drivers as the cars get better. For a fleet service you might well never stop having some people monitoring to respond to unexpected conditions.
Of course, for this to be viable, the car needs to be possible to be made safe without human intervention, but that safety may be achieved by opting to stop or slow down the car in situations where continuing might be perfectly safe (and with the caveat that this may e.g. restrict where it may be possible to let it drive etc.), but where the car can't yet tell by itself.
This of course presupposes specific types of failure scenarios where the car can safely find a way to come to a stop but can't safely determine if it can continue forwards. It's not a given that's achievable with low enough effort (relative to solving the issues that might cause it to fail to spot a hazard) to be worth it.
Exactly. For example, imagine a situation where a tree is down, blocking one direction of travel. Humans would very cautiously share the remaining road space. But a robot taxi would just stop and wait for the tree to move. At that point it summons a human who tells it what to do in broad terms (e.g., "the new lane is here" or "do a u-turn and follow this other route").
The remote assist isn’t to directly drive the vehicle like some sort of video game.
It would be to annotate something the algorithm flagged as impassable in some way such that the car can continue driving itself.
If the car entirely fails to identify a driveable lane, I don’t think you can remote in and actually active steer a human-occupied 2 ton vehicle over 4G.
Right. What happens when a construction crew accidentally cuts the backhaul fiber for the local cellular tower? Or if the autonomous vehicle failure occurs in a tunnel with no cell service?
You'll note that Waymo is very much operating in a limited area. So I think the answer is: they don't offer service in an area that lacks good cell coverage.
Of course, it's still possible that the vehicle will somehow be out of communication when it encounters something it doesn't understand. A cell jammer, say. In which case, it'll do what it will do if it, say, detects engine trouble: it'll pull over and wait.
Run the live video feeds from the vehicles through a style transfer AI that has been trained on Mario Kart, then secretly stream the result to a mobile app disguised as a free taxi driving game. All the app needs is a feature where the app turns off adverts if the player keeps completing taxi journeys without crashing the vehicle and you are good to go.
> When Waymo can get rid of the safety drivers [...] they're getting close to something useable.
I suspect that they can cheat this one a little. As long as they're in an area with good connectivity and the cars are smart enough to pull over if they don't get human guidance, I expect they'll move the backup drivers to a central location. Call it SAE level 3.5. It wouldn't be good enough to sell cars, but it would be workable for a ride service, and would allow them to undercut Uber, etc, quite handily.