Editing VBO Files

I have been getting my GoPro video attached to a vbo file export using the information in this thread: http://www.racechrono.com/forum/#/discussion/1153/external-video-for-circuit-tools-vbo-files

Now that i have that working, I thought it would be nice if additional channels could be added for calculated fields. For example, I've run across this information on calculating a faux braking force channel: https://www.dropbox.com/s/xrqmanqldg8up0p/brake-trace.pdf?dl=0 Taking the concepts from the video attaching, I simply added a "braking" column in the headers and stuffed it with data in Excel, then copied back to the vbo file. Wouldn't you know it, but it worked!

Comments

  • Great idea. Please share any good calculated channels you come up with. Since I'm driving a kart, there is no OBD-II data so I'm very interested in expanding what I can gather. (I'm thinking of building an Arduino-based box to acquire data and simulate it as OBD to RC. Since I'd be pre-processing the inputs before handing off to RC, I'd also have the chance to massage the data in transit.)
  • So, stealing the formula in that file, I have this for the "braking" column: =MAX((-1*$H2)-(($E2^2)/100000),0) With H being longacc and E being velocity. That gave me a pretty jumpy dataset, so I applied a weighted moving average using this: =SUMPRODUCT($S1:$S5,{1;3;10;3;1})/(1+3+10+3+1) wit S being the braking column. I tweaked the constant in the main braking to get it to *just* show up on a coast-down area (the document starts it at 50000).

    The weighted average was just to smooth out the graph a bit. The adjustment factors is somewhat arbitrary--I played with it until I got a reasonably smooth line without distorting from the original calculated feed too much--keeping in mind the "original" is fabricated off other data to begin with. Now it doesn't look like a series of spikes.
  • Great use of SUMPRODUCT - that's new to me and very handy.
  • I just learned about it myself. It's pretty cool.
  • Interestingly, Circuit Tools calculates "Combined G" for us. I had summed longacc and latacc myself before spotting this. They also apply some smoothing. I believe the idea is to ensure the driver is using the entire available grip at any time (primarily in braking and turns, harder to reach the limit under acceleration). The old thinking being straight-line braking at the limit of tire grip, followed by lateral/turning to its own maximum. The "newer" trail braking understands the value of a sliding scale - 100% of tire grip can be split between the longitudinal braking component and the lateral turning component. If the total is less than 100%, you're under-utilizing one of those components. 100% of grip is determined by exceeding the limit (locking up the tires) or also viewing the G-G scatter plot for the oval-shaped outer bound of G forces.

    I'd love to get anyone's confirmation or clarification of the above. I'm just piecing this together myself. :-)

    RaceStudio from AIM also gave me some good direction. Their tool goes much farther into use of math channels (that is, calculated columns). Too bad their interface is more cumbersome, and more importantly their file format is binary and proprietary. I might tackle analyzing their data files at some point, but that will have to be another time.

    Anyway, one of RaceStudio's built-in calculations is for Slip, meaning the ratio of RPM to GPS speed. I'm not sure whether this is just clutch slip or has other factors (it has to be the whole drivetrain, including wheel slip too, right?). In fact, I think isolating the clutch for would be *torque* versus wheel speed, if that's even possible. I'd love some thoughts here as well. This version of slip is interesting but I'd like to understand how to confirm full clutch engagement. This is adjustable in karts and I think an important tuning opportunity (as opposed to other vehicles).

    There's quite a bit available on YouTube at https://www.youtube.com/user/aimdata and http://www.aim-sportline.com/download/software/doc/Math_Channels_101_eng.pdf (Antti, I hope these links are OK. Feel free to remove, though I didn't see any forum rules preventing it.)
  • This is actually pretty interesting. I have no problem with these links, as I could use them to create new features :)
  • ^^ :D

    I actually just found the pre-calculated Combined G trace in CT myself. My understanding of why we want to look at it follows what Karter wrote. That is coming from the Roger Caddell/AIM webinar recording; the Practical Data Acquisition one here:

    I'm not sure how I missed all those useful-looking videos, but it appears I have some time in front of the computer on YouTube ahead of me. The one I have watched already was incredibly useful.

    I can say that now that I've gotten a quick primer on what to do with the tools, the RaceChrono data has given me some very concrete things to work on at my next track day. For me RC has gone from a nice addition to the YT videos I post to being an actual useful tool--I am thinking not just for after-the-event analysis in CT, but for between sessions during the day; combining the two really leverages the value of RC.
  • +1 for the usefulness of RC (plus the added fun of later analysis). My own priorities are predictive lap timing for live feedback on the track, post-race analysis without drowning in data, and videos with data overlaid to serve both analysis and "cool factor."

    Thinking about slip some more, RPM versus GPS Speed must be the best available method without a sensor for wheel speed / axle speed (no differential for my case). Having that would isolate slip from tires versus clutch. I'm not yet prepared for that, so this method will have to do for now. I don't see anyone else mentioning it, so maybe considered negligible in the big picture? Engine torque would have to be dyno'd separately and then you match your clutch's lockup to max torque.

    I'm sure a lot of interesting X-Y plots comparing two channels can be made. I plan to fool around with that for a while, but I think the biggest gain comes from comparing Delta-T of laps (against yourself or another driver) to find gains/losses, then looking at Speed (e.g. carrying speed further into a corner, greater slope of the line showing harder braking, etc.). Backing that up with video is critical to see the exact driving line, hence my work there. Everything else is diving into the case-by-case details. So that's the approach I've gleaned as a newbie.

    I think there's a good opportunity to provide acquisition and basic analysis on site during track days. I'm thinking on the model of photographers who sell their services. Is anyone already doing this, outside of driving schools?

    On resources, Racelogic also has a nice ebook that they'll email from a sign up page on their website. It's not heavy into data per se, more on improving driving through it. Lots of good case studies.
Sign In or Register to comment.