One under appreciated benefit they have is the "eyes" on a Waymo car are mounted at the corners and high above the roof, giving it a much better vantage point that just sitting in the driver seat.

It is demonstrably better. They see more and react faster. They can see around obstacles that humans can't and they never get tired or distracted.

See more in the rain y dark? Sure.

But can they look at a driver's face and realise that they are distracted with their phone and haven't seen you?

This is an important point. Humans have evolved a way to intuit the thoughts and behaviors of other drivers. It’s why people who are mentally ill are unsettling; it’s very hard to pick up on their intentions and course of action. That’s why in think the hurdle isn’t “better than the average human driver.” Because humans won’t trust a machine in the same way as another human, the level of safety for AVs may have to be much, much greater before the public as a whole trusts them with safety critical decisions.

Yes they can. They are absolutely capable of looking at gestures. That matters a lot more when you're trying to determine the intent of a pedestrian than a driver. If it's another driver, the system can also do real time physics calculations to avoid a vehicle straying into the av's trajectory. And they might even do some kind of computer vision with other drivers to determine intent. Also if another driver just hits the car because they were distracted, why would the AV be at fault?

These are really sophisticated systems and they've really thought of a lot of issues. All of these companies have hundreds, if not thousands of engineers who have been working on this problem for over a decade, as well as an army of lawyers trying to get things into compliance. I don't think "well it can't do x as well as a human" is going to hold up in 10 or 20 years when these systems are clearly much safer. Human driven cars will eventually be uninsurable.

> All of these companies have hundreds, if not thousands of engineers who have been working on this problem for over a decade,

So do a lot of tech companies that make a lot of dangerous, fraudulant, or just bad products, or that try to develop technology that never comes to fruition.

The tech is impressive, but that doesn't lead to an assumption that it must be able to do whatever is needed. I've seen enough tech hype cycles to know them when I see them. In the end it will have great strengths and great weaknesses, and we should learn what they are.

>They are absolutely capable of looking at gestures.

But 'looking at gestures' isn't the same thing as having a real-time interpretation of the sensors. This, to the point of another comment above, may be conflating 'perception' with the larger problems of 'self-driving.'

Humans see a gesture and understand the meaning very fast because we process a lot of communication on the subconscious level. If you read the safety report of the Uber incident, it shows that the self-driving system could 'see' the pedestrian, but it struggled to effectively classify them. To be fair, that was years ago and another company, but we (as the public) don't have much insight into how good (or bad) the software is.

So 'seeing gestures,' 'correctly classifying gestures', and 'correctly classifying gestures in a time-sensitive manner' are varying levels of difficulty. The last one is the one that matters for safety-critical operations, and the requirements change with the velocity of the vehicle. IMO, there should be a regulatory framework that addresses a baseline of performance on these systems.

That comparison is a very common sales technique and fallacy: Take two products, list the things product B does better than A, and therefore B is better!

I'm not sure I get what you're saying. I'm saying the waymo is probably better than me at detecting and avoiding obstacles. That is the point of this thread.

I understand what you are saying. I'm saying that argument presented doesn't support that conclusion (though it may be true for other reasons).