If I accelerate into traffic and get close to a car while accelerating, it will start flashing a warning light and if I still don't stop accelerating or braking, it will intervene, but if the other car accelerates it doesn't do anything. So the radar setup is quite well done and it gives very few false positives.
The lane assist system is a bit bad though because once it identifies the lane markings it does not take into account potential hazards that want me to change a lane or exit it for a few seconds to avoid road hazard lights or something similar.
The point I am trying to make is that there should be a high standard to what drives the AEB and it should not rely on cameras, which seems to be what automakers want to implement driving assisted features to save on more advanced tech. And somehow I don't think they're gonna pony up for high-speed cameras that can replace a radar at accuracy.
Maybe the intention is good, but most likely the implementation will suck.
Right, if I had to condense my issues into something more constructive it is just that: if regulators want to pursue a mandate for these systems, they should also work on establishing testing criteria and specifications that cover the many different conditions drivers face and reduces variation between manufacturers as possible. Not an easy task.
To give a quick example: another commenter in this thread states that lane assist only kicks in at 35mph. That is true in some models but not all which could lead to some nice surprises. Further, the speed limit in NYC is 25mph so it's not unusual if the road surface is decent to be cruising closer to 30. Unless there's a threshold for activation, you might be walking a fine line between "safety on"/"safety off" and any resulting corrections it might make.