I never really understood how the industry was able to ensure, from a legal standpoint, the failsafeness of autonomous cars. There are bound to be errors, such as the incident last year with running over a homeless woman who was clearly visible to a real eye, but not detected.

It’s simply impossible to account for all the corner cases, especially when you have to factor in manual drivers who can behave arbitrarily (what if they purposely drive weird just to mess with the autonomous cars?).

In fact, autonomous cars may increase danger rates as human drivers become quite peeved at sharing the road with them, and try to derail them at all cost.

Maybe these are the reasons they gave up.

> what if they purposely drive weird just to mess with the autonomous cars?

I got a car about 6 months ago with Level 2 Autonomy and I'd say it's even more than just trying to mess with them. Now that I know the subtleties of how my car behaves, it's easy to spot other drivers who have theirs engaged too. I can use that to my advantage when appropriate and it takes some ethics not to abuse them. For example, if I have to merge I'll merge in front of one of those cars, purely from a safety standpoint -- I know that car is going to respond and give itself lots of space. Of course, people could use that much more maliciously to take advantage of those cars "niceness".

I've never had anyone mess with my car just for the sake of it, but you can see the seasoned Tesla owners taking the opportunities that they spot. It changes the way other drivers drive if they can identify the behavior.

A number of researchers without vested interests have been saying this all along. But they’ve been drowned out by those many who so wanted to believe, not least of which the young urbanites who so desperately wanted to believe they would never need to buy a car and would be driven everywhere all their lives.

It’s true. I watched a panel discussion from maybe three years ago and those who were closest to the research, even those who were actively working on the technology, estimated 2022-2025 for a general debut and longer for the difficult cases. We may see large pilots sooner but the timing given by the experts seems properly prescient compared to all the excited people who thought we’d have the tech widely this year.

Well, it really really looks like Waymo has something decent. TBD because they are damn secretive and there is probably ten years or more before the tech is perfect, but I think there’s opportunity to start taxi services in good weather in the coming years.

It's already starting https://techcrunch.com/2019/01/22/waymo-plans-to-open-a-self...

There are too many cars already.