Come on, it´s not that simple. Autopilot has shown that it can navigate curves very well, so it knows its trajectory a bit into the future. With that knowledge it can identify if an object is in its path or not.
That said I know it is obviously a problem to switch on the braking for stationary objects due to other corner cases leading to false positives. But I do think that is a serious shortcoming of AP, even it is well documented and any user who read the manual should know that.
We all agree that legally Tesla is not at fault as the driver has to monitor what´s going on and is responsible to brake if there is a stationary object. However, given the level of advancement of the rest of the system one would intuitively not expect AP to ignore those objects.
Also, there have been some serious accidents due to this shortcoming. I am sure Tesla is doing everything they can to improve AP in this regard. But they are making a choice by keeping the system running like it is and that´s up for discussion.
It comes down to statistics. Essentially, a car stopping when it doesn't need to would be a Type I error. To greatly increase the chance of stopping when it should, you would likely also increase the chance of having a Type I error, stopping when it doesn't need to, which if this happens too often, then Tesla would be terrible at doing basic statistics.
The serious issue with with have a Type II error, not stopping when it should. This is always a huge problem because it has more severe consequences, and is a natural reason for frustration (and inevitable statistics wise).
Consider this scenario: You have to take some prescribed drug, but in the label it says you must walk for 10minutes immediately after taking it (or change this condition to whatever you want). Suppose there is only a 0.01% chance that you will die IF you do not walk for 10minutes. And over several times of taking the drug, you forgot to walk or just decided not to walk, and you realize oh hey, I don't really really need to walk ALL the time after taking the drug. So, on average, 1 out of every 10,000 who take this drug will die after taking the drug (if they all think this way), or if you think this way, you'll likely die after about 10,000 times of taking the drug.
I don't think it would be the drug company's fault if those people die for not walking when they should have. Also, I would think in a scenario like this EVERYONE would take the 10min walk (to not risk dying). Yet, there are tons of people who still smoke often, and people who don't put their hands on the wheel while driving in autopilot.....