> Which seems unlikely.

No kidding.

https://www.cdc.gov/niosh/docs/2001-128/pdfs/2001-128.pdf?id...: The two most dangerous things in road work, accounting for half of all of the industry's fatalities, are industrial vehicles working on the site and passenger vehicles passing it.

So, yeah, you're right - there aren't any alternative systems that manage to avoid this problem. Even humans suck at it. And if even humans can't handle construction areas safely, how can you possibly expect computers to be perfect?

* It depends on why the humans are failing. You can design a computer to not speed and not get distracted.

* Hitting a worker is very different from hitting a truck.

* Many of those fatalities are not in clear daylight.

There are. Use two or three alternative AI implementations, which will watch each other. Kind of famous "Predator" algo, but for driving. If one will fail, second will pick up. If one makes mistakes, second will teach it.