Being in a bike lane sandwiched between northbound vehicular traffic; and, a vehicular traffic right-turn lane does not make a biker a pedestrian. Consider that the area of the accident was lighted with overhead street lamps. The car system has a prior known issue with acknowledging a bike lane or any occupants of a bike lane.
The vehicle system LIDAR sensor likely painted the person & bicycle combination near the car's path before the collision, but the software failed to properly determine what was being seen. Unless it can accurately categorize confusing sensor data, it apparently does not default to "brakes" to slow down, or com to a halt. It just merrily chugged along--and upon hitting an obstruction, required the human driver to override the A.I. and stop the vehicle. It never noted the impact as significant--just as some drunk drivers with a pedestrian lodged in the windshield continue about their travels.
And this is a problem. Made worse by a felon behind the wheel. I looked at the visual area, someone had a link to the location. There is a walkway in the median that she likely used. There is lighting. And nothing to obstruct the view. I figure the ‘coming out of shadows’ was not her, it was the fact that the headlights fell upon her. The did not slow, did not brake. The human was oblivious, not paying attention.