This hints that Uber's mindset is focused on "obstacles", not "flat road". The first job of automated driving is to drive only on flat surfaces. Doesn't matter why it's not flat. Then you worry about where to go. That's how the off-road DARPA Grand Challenge vehicles had to work. "Flat road" is a pure geometry thing. Not much AI needed.
On road, non-flat road is rare, and you don't have to worry much about going off cliffs, rocks in the road, and such. So it's tempting to focus on "obstacles" to be tallied and classified. Tesla definitely has an "obstacle" focus, and a limited class of obstacles considered. Waymo profiles the ground. Looks like Uber had the "obstacle" focus.
If you have a "flat road" detection system, things the obstacle detector doesn't understand get stopped for or bypassed. So the vehicle isn't vulnerable to this flaw. There's a higher false alarm rate, though. And a non false alarm problem. A piece of trash on the road is likely to result in a slowdown and careful avoidance, not a bump as the car rolls over it. Still, better to take the safe approach unless you're on a freeway and the cars ahead of you just got past the obstacle successfully.
The Udacity self driving car "nanodegree" trains people to build obstacle detectors and classifiers. That may be a problem.