Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Software Update 2018.42.x

This site may earn commission on affiliate links.
How do you view such download activity?

I have a 2018 Model S running 2018.38.6, first wide release of V9. Real interested in getting the new 42.2 V9.

Thx for your help.
You haver to have a wifi router that supports this feature. I think googles wifi system does. I use ubiquiti access points and they support it
 

Attachments

  • SNAG-0225.png
    SNAG-0225.png
    109.2 KB · Views: 84
  • Like
Reactions: VikH
Caveat: it works best when you have that and/or when you have labels inherent in the data. For any non-trivial problem, you’re going to have FAR more data than you can possibly have labeled by domain experts. So you find ways to flag select data where their expertise is required. That can mean having non-experts do a first set of passes and only send to experts those where you get significant disagreement. It can also mean, for example, picking out anomalous sensor data(defined as some significant deviation from the mean).

In some cases, the inherent labels are all that’s needed. AlphaGo and AlphaZero were both trained without any labeling of the data, because the end result of a game could be used as a label in itself. Similarly, here, any crash/near miss is an excellent label. Taking situations where a person took over and running it through a simulator can also provide an excellent label.

All this aside, we *know* they’re using neural networks for object detection and classification. And we know they’re doing a relatively good job at that task and improving over time. They must be getting the data for that from *somewhere*. And, of course, we also know that for a (changing) set of certain select situations and also at random intervals, each car occasionally gathers a short set of videos and sensor data which is sent to the mothership. Either they’re using that data for ML, or they’re lying to all of us and secretly using magic.
Automagic I think is the correct term.
 
I don't really follow what you're saying here. What is "integrated RTK", and what does it have anything to do with the presence of an IMU? RTK is the process of determining, in real time, the relative baseline between a rover's GPS reciever (e.g. a car) and a reference GPS receiver (e.g. a fixed station on top of a building of known location). It seeks to use the precise carrier phase measurements to obtain very high accuracy positioning.

How can Tesla implement RTK techniques on their own? They would need an extensive network of accurately-surveyed reference stations as well as a low-latency medium to transmit their measurements to users. Even then, accuracy is not guaranteed because of the type of environments cars operate in. For example, a car driving in an urban canyon would be subject to severe signal multipath and blockage. The results would likely not be good enough for activities such as driving lane resolution.
I've wondered if tesla has/will use predetermined checkpoints that are recognized with great precision by vision/radar to correct GPS/SBAS data. The correction factor could be broadcast to other Tesla's within a certain proximity over the internet creating a virtual LBAS network wherever Tesla's are driving.
 
My first morning commute with 42.2 and Navigate on Autopilot was... extremely underwhelming. And not just because you have to confirm the lane changes. The car does not know where the it is and what is happening next. The overall experience is terrifying. I'm glad you have to confirm lane changes. It still recognizes shoulders as lanes sometimes. It does not know that the lane you're in is ending in a couple hundred feet. It drives like a drunk when merging onto the highway and exiting. It only realizes it's time to exit when the exit lane is mostly behind you and it's pretty much too late.

NOA engaged as soon as I got onto the on-ramp to the highway. I thought that was pretty good, it knew when I was on the ramp. It was all downhill from there (figuratively, not literally). It could not merge into slow-moving traffic at the end of the ramp. It didn't really even try -- it never suggested that I get out of the acceleration lane and onto the highway. I initiated a lane change on my own near the end of the lane because it was getting dangerous to wait for NOA to do it. It then attempted to execute the merge and ended up rapidly ping-ponging in a way that was dangerous and confusing to other drivers so I took over.

It never suggested overtaking during my commute despite traffic moving much slower than my setpoint and the left lane moving slightly faster, though this didn't bother me much as I would not have overtaken either probably. I stayed in the right lane until my exit. The navigation clearly knew when the exit was coming up but NOA did not suggest a lane change until way, way too late. I let it take its time because traffic was slow and I was willing to expirement because risk was low, but basically half of the deceleration lane was behind me by the time NOA finally suggested moving over. I confirmed and once again got a very awkwardly-executed auto lane change into the off-ramp and took over immediately to avoid looking like I was drunk. If I had been going anywhere near the speed limit (traffic was moving at about 10mph on the highway) I would have either missed that exit or ended up very suddenly taking the exit at the last minute, which would have been dangerous.

At present, this is the most useless feature, probably more useless than the Atari games. Let's hope it improves once they get those 10M miles...