Another recent tweet on Summon:
"Parking lots are a remarkably hard problem. Doing an in-depth engineering review of Enhanced Summon later today."
Elon Musk on Twitter
Also, last week:
"Intersections with complex traffic lights & shopping mall parking lots are the two biggest software challenges. Developer branch already mostly works in these scenarios, but massive effort required to get to 99.9999% safety."
Elon Musk on Twitter
At first glance, Summon seems really easy. And it is, for simple uncrowded parking lots.
But in a very busy parking lot, with many moving cars and pedestrians, where there are not even lanes or traffic lights, Summon is probably the hardest of all Self-Driving problems. Human drivers may rely on non-standard or subtle cues (for example, direct human to human communication), which are not possible in a car without a human driver (of course, Tesla can train its NN to recognize, for example, human hand gestures from other vehicles and pedestrians).
So this is going to be very interesting. What will Tesla do here? Looks like Summon is a much harder problem than initially thought. It could be quite a while till it is fully solved. Or does Tesla limit summon to non-busy parking lots (and how does Tesla determine this)? In short (IMO), be prepared for either a significant delay and/or significantly limited functionality on initial release.