wow,
i even see
https://wiki.autosportlabs.com/RaceCapture_on_Raspberry_Pi
so besides their dash app for RPI, you could even run ONLINE in the car
your processing tool ( in the RPI using mk3 connected GPS data … )
and can have all storage you want on system uSD or add on USB.
that should be fun to combine all that!
I might not be understanding, but a concern is that, if you turn late on your data collection pass, doesn’t that mean your anticipation-pass will be giving you values based on a bad previous path? That seems self-reinforcing in a way that you may not want.
If I understand your comment that would be a problem and is one reason why I don’t think i can rely on a “data collection” pass. I’ve actually got it running pretty well right now - and it only uses “target corners” and an understanding of the correct sequence of corners. I have to go into Google Earth and download the GPS location of each corner Apex and manually enter them into the program code along with the proper sequence that the corners are in. Beyond this, the program doesn’t need to know anything about the track. It’s simply looking to see if the car has passed a corner - and then targeting the next corner in the sequence. If the GPS signal were to fail, the program simply waits until you pass a corner to begin reporting again.
There are a few minor disadvantages to this approach:
-
The distance to the corner displayed “as the crow flies” as opposed to following the path of the track (since I haven’t stored the track path) but this isn’t really a problem since the only real areas that the track markers add value are on a corner after a long straight away where the car is moving fast and proper braking is a challenge. You don’t really use track marker on corners nested in corners.
-
Depending on the track layout, there are a few weird locations where the distance to the next corner starts very small (as the crow flies) but actually increases as the car progresses down the track and then decreases as the car approaches the corner for real. There is a turn at Road America (the Carousel) where this is actually the case. When you drive out of turn 8, you are actually quite close to turn 9 but moving away from it. So the program will weirdly count UP as you drive out of turn 8 - then go dark - then begin counting down from 9 as you approach turn 9 for real. I can’t think of an elegant solution for this and I don’t think it’s worth solving.
-
As mentioned - if you lose the GPS signal momentarily, you’ll need to wait until you drive past a corner for the program to re-engage since it will have no idea what you’re aiming for. But I’m hoping that we can begin displaying the track markers far enough in advance that you’d know you’d missed your corner markers and could avoid over-driving the corner.
I think my Processing program was confusing because it did indeed have a full map of the track but this was to allow me to simulate the car driving - not because I wanted my actual program to rely on that.
Although it’s less sophisticated than a full map of the track, it’s also a tiny fraction of the data - 8 GPS points versus a couple thousand. And I have real concerns about data so this is really appealing to me.
I’ve switched over to porting this code to LUA and hope to be running with actual hardware later this weekend. It works well in my simulation. I’m excited to see how it works for real. Thanks for the help!
Dave