I apologize for my safe driving. /s
A chain's as strong as its weakest link, and there are a boatload of weak links still. I'm interested to see how my 12.3 issues are addressed (or not) going into the next updates: stop signs, speed, rights into traffic, glaring nav confusion at my funky intersection, etc.
Right. What I want is simple: Drive at the MAX speed if safe to do so.
I've been trying to pay more attention to 12.3's stopping behavior given that you seem to have way more issue with it, and I wonder if it's more noticeable when there's lead vehicles slowing down. There was a previous comment that stopping at a line is trickier than accelerating from the line, but is that actually the common case for you?
I was having a discussion with my buddy on charging % and we all remember Elon said 90%…Poor engineers are probably working around the clock trying to get a stable/safe version FAST to cover Elon's "toilet Tweet".
Is there a tweet or something of Elon saying the goal for v12 is unsupervised operation?
I use vision, prior knowledge of an area, knowledge of the traffic laws for my region, knowledge of behaviours of types of vehicles, ability to read and interpret most signs (some just make no sense and I live in a bilingual jurisdiction so temporary highway sign boards flash the message in one language then the second language and sometimes the message is more than one 'screen' long so one doesn't see both parts of the message before driving by) and SOUND.
Yesterday it was worse without lead vehicles. (Though it is a mixed bag with lead vehicles, depending on how they stop. For some reason the car tracks the lead vehicle and not the traffic light.)
On the surface, that's seems true. But I bet there's a lot more going on under the hood than meets the eye.
Did she somewhat like 11.x or not at all? Here's some comments from Dirty Tesla wife about 12.3.2.1 and Autopark:This was the first experience with my wife. Although it got the lanes wrong twice, she likes this version more.
The idea of a procedural (C++) thread somehow monitoring and overriding FSDb when needed is...problematic. Because the better the NN gets at driving, the less confident the C++ code can be in making the decision to override. I can envision the C++ code as being a safety rail for a short time to address glaring flaws in behavior of the NN, but it does not seem like an effective long-term solution. The NN is just going to have to get really, really good all by itself AFAIK.
That was 2019. I'd been charging both of ours to 90% until my app started saying 80%, so now I do 80% for day-to-day driving (and typically 90-100% immediately before setting out on a longer trip.)
