Well, well, well, what do you know, I actually drove AP2 half an hour (out of roughly one hour's drive) today! Here's my report. This was the .42 version. Unfortunately these roads did not include my usual testing motorway, which has its issues in the past, but still a divere set of roads.
First the bad:
- IC did not show blue cars or cars on other lanes (and I did not use lane change). Also there does not seem to be proper visual recognition of where cars are turning (or it is a poor one), since some cars were turning left, yet their graphic on the IC was showing a car moving towards the left, but tilted towards the opposite direction (right).
- Local roads (I guess some call them surface streets) were just as unreliable as before, but to give it a fair chance, I did try it as many times as possible. So no, where I drive, these are out of the question. I wouldn't have tried either, unless I felt pressured by peer-pressure on TMC to do stupid things.
Then the average:
- Wide lanes caused disengagements. It just lost the sight of the lane when it was wider than normal, even though markings were clear and the road unambiguous. But these are rare, so.
- Bias on the lane was more uneven than before, perhaps 40-60 split instead of 50-50 split. Oddly this changed from side to side. Also this was correctly reflected on the IC, so the car did know it was biased towards one side or the other. I did not find any logic to this.
- There was ever so slight zig-zagging going on, nothing like the ping-ponging of early AP2 versions, but you could feel the steering wheel swaying a little.
Finally the good:
- Other than those minor niggles,
it was really good on divided roads in a wide range of speeds. I found it pleasant even in stop-and-go traffic, usually a sore spot that has caused too many disengagements to be worth it before. All in all, very useful on divided roads.
This adds to my pleasant experiences with .36, .40 and now .42. I am cautiously optimistic things are finally going somewhere.