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

Software updates in Australia

This site may earn commission on affiliate links.
I think there would be some benefit to posting which updates are received by whom and when, including car specification to determine if a specific update is restricted to a specific configuration. Perhaps this could be made a sticky?
My configuration is below. For the record, no map update or software update since the original 7.1
 
My suggested use of the Update Tracker is as follow: Tesla Firmware Upgrade Tracker Web App

1. Click "View Reports"
1.PNG


2. Check out the last version "Updates by Build/Day"
2.PNG


3. Scroll to bottom and click "Full Data Extract Version.... (full)"
3.PNG


4. Click "Cntry" - sort by country
4.PNG


5. Voila - as Australia is alphabetically in front, we can see who and what version in Australia we are upto!
5.PNG


Of course we need more people to feed in data to make it functional.
 
I think there would be some benefit to posting which updates are received by whom and when, including car specification to determine if a specific update is restricted to a specific configuration. Perhaps this could be made a sticky?
My configuration is below. For the record, no map update or software update since the original 7.1
I got a map update last week, dborn. No new software update for a while though- still on .154...
I agree - Aus specific data would be good!
 
I received the map update about a week ago. Then I went on a short run and it told me I was in a 40 zone on a 100 (marked) section of road. Mmmmm, not, perhaps 100% all the time :crying:

I find the speed limit detection pretty hit or miss, it still often mistake 80 for 30, and often misses limit changes. That's not so much a property of the maps, but of the visual recognition software.
 
Not too sure about that Colin. In an ICE car, using Google Maps nav on a Samsung 6 Edge I see the correct speed limit. GPS can do that without local input, since GPS is accurate to around 5 metres (non-military GPS that is :) One of the reasons I'm sitting back and waiting for Autopiklot and other software to improve (just a little) before dipping a toe in that particular pond. The lads at Tesla will have it worked out in no time, I'm sure. This is the absolute beauty of software updating in a car, you can be "in the swim" and get the updates as they come, or you can be a "fraidy cat", hang back and wait a bit. Not like the poor BMW owner who has to get a bit more hardware in the form of a key to summon their car.
 
Not too sure about that Colin. In an ICE car, using Google Maps nav on a Samsung 6 Edge I see the correct speed limit. GPS can do that without local input, since GPS is accurate to around 5 metres (non-military GPS that is :) One of the reasons I'm sitting back and waiting for Autopiklot and other software to improve (just a little) before dipping a toe in that particular pond. The lads at Tesla will have it worked out in no time, I'm sure. This is the absolute beauty of software updating in a car, you can be "in the swim" and get the updates as they come, or you can be a "fraidy cat", hang back and wait a bit. Not like the poor BMW owner who has to get a bit more hardware in the form of a key to summon their car.

I am with Colin. Tesla uses visual data not GPS for speed zones. It reads temporary speed limit signs (eg. 40 zone for roadworks) and it reads and accepts signs that apply just for trucks (eg. 60 Trucks Only) leading to false alarms for speeding. As we know it has an issue with 30/80 differentiation and it doesn't read 25 for some reason so struggles with Australian roadworks.
 
I am with Colin. Tesla uses visual data not GPS for speed zones. It reads temporary speed limit signs (eg. 40 zone for roadworks) and it reads and accepts signs that apply just for trucks (eg. 60 Trucks Only) leading to false alarms for speeding. As we know it has an issue with 30/80 differentiation and it doesn't read 25 for some reason so struggles with Australian roadworks.
25 isn't in the cars library of road signs, and given very few places use it, its unlikely to happen soon.
Also signs that say "end 40 zone" make the car think its a 40 zone.
I agree its using visual data and not gps.
 
Which, I believe, is partly my point. The system should be arbitrating between visual inputs and GPS to a more sophisticated extent to try and get a more accurate idea of conditions. In this case the "error" was 40 .vs. 100 so something amiss somewhere. Also I would like to see various "avoidance" conditions in the nav such as bypassing toll roads and ferries and even motorways. I understand, given "range issues" (not many) that Tesla would have put emphasis on the shortest route, but it would be nice (and not difficult to implement) to have the option. Visual versus GPS arbitration should be setup in AP as a non-deterministic, bottom-avoiding choice where the system has a "knowledge" of what is "on the map" usually and that will suffice for probably 99% of situations and the visual system maybe override it where there are temporary conditions. That's glossed over 10,000 problems! It might require something as complex as satellite imaging or as simple as a knowledge of traffic flows which will be out of kilter around places with diversions, temporary slow down zones et cetera. What you do when there's a road accident? Forget it :crying:
 
Which, I believe, is partly my point. The system should be arbitrating between visual inputs and GPS to a more sophisticated extent to try and get a more accurate idea of conditions. In this case the "error" was 40 .vs. 100 so something amiss somewhere. Also I would like to see various "avoidance" conditions in the nav such as bypassing toll roads and ferries and even motorways. I understand, given "range issues" (not many) that Tesla would have put emphasis on the shortest route, but it would be nice (and not difficult to implement) to have the option. Visual versus GPS arbitration should be setup in AP as a non-deterministic, bottom-avoiding choice where the system has a "knowledge" of what is "on the map" usually and that will suffice for probably 99% of situations and the visual system maybe override it where there are temporary conditions. That's glossed over 10,000 problems! It might require something as complex as satellite imaging or as simple as a knowledge of traffic flows which will be out of kilter around places with diversions, temporary slow down zones et cetera. What you do when there's a road accident? Forget it :crying:
Its still better than any other gps Ive ever had, but adelaide is a very simple place with no toll roads.
 
As we know it has an issue with 30/80 differentiation and it doesn't read 25 for some reason so struggles with Australian roadworks.
The 80/30 issue is driving me crazy, the software just has NFI, in particular it usually reads a 100-> 80 sign as 30 and then the second 80 sign as 80. The signs of course are absolutely identical. Surely this is just a coding fix?
 
That's one aspect of the problem. However, we don't need to "kill an ant with an atom bomb" (so to speak). The GPS is more than adequate for most speed zones (since they don't move around a lot). This basic method is also much more reliable in foggy, snowy or rainy weather, for obvious reasons -- the zones have not moved, but your (let us assume perfect) visual scanner may not be able to see clearly. Also, that muck you haven't cleaned off the windscreen might be causing issues too -- no offense meant to anyone. What I am driving at is more along the lines of "Making Choices Lazily" (Dr Andrew Moran and Prof. John Hughes) as long ago as 1995; the mathematical foundation anyhow. More recently (2015) this paper:

Improving Big Data visual analytics with interactive virtual reality

Andrew Moran; Vijay Gadepally;

at M.I.T. is looking at this sort of thing, but in a vastly different context to driving a car. Nevertheless, it is the kind of analysis that could, I believe, be applied to the driving "problem". And we always wondered why there were so many "crap drivers" out there?! It looks like it really is rocket science after all, but I know that the reversing problem was solved (in principle) more than 15 years ago using neural nets and it required the equivalent of two brain cells in the net to achieve perfect car or truck reversing up to a loading dock, for example. We were cautioned at that conference never to mention the "two brain cells" part to any truck drivers we might know, for the obvious reasons :crying: