yuri_base 1 #1 December 18, 2006 Yo! Today I made a few test flights with the Eagle Tree Systems flight data recorder. The recorder has built-in pressure sensors to measure airspeed and altitude, and an option to connect various other sensors, including accelerometer, custom sensors, GPS, etc. The data is recorded to onboard memory and can be downloaded to computer via USB. Also, the data can be monitored live using wireless dashboard. The sensors are sampled 10 times per second. The first results are promising! Although the Pitot/static tubes were mounted in fixed positions, the altitude/speed data is relatively smooth when the head was not moved. The next step is to mount the tubes and accelerometer on a weathervane. This will provide much more reliable altitude/speed readings and with the help of accelerometer, the true L/D can be measured. The goal is to build a system which will provide live feedback to the flyer with accurate L/D and horizontal/vertical speed data. Share your ideas, knowledge, results here! Anyone know Kalman filters? 2007 is going to be a very exciting year! YuriAndroid+Wear/iOS/Windows apps: L/D Vario, Smart Altimeter, Rockdrop Pro, Wingsuit FAP iOS only: L/D Magic Windows only: WS Studio Quote Share this post Link to post Share on other sites
notsane 0 #2 December 18, 2006 A Kalman filter is a mathematical transformation you can use on your dataset, like a FIR or IIR. A tool like Mathworks Matlab has a filtering tool that will do the job nicely. Quote Share this post Link to post Share on other sites
Costyn 0 #3 December 18, 2006 Very cool setup, Yuri! It's cool to finally get some raw data to compare to all your theory stuff, right? CheersCostyn van Dongen - http://www.flylikebrick.com/ - World Wide Wingsuit News Quote Share this post Link to post Share on other sites
yuri_base 1 #4 December 18, 2006 It's actually vice-versa: the wingsuit equations will be used in Kalman filter to produce the best estimate of the current flight parameters (horizontal/vertical speeds and L/D) from the noisy data. In a nutshell, Kalman filter is a filtering algorithm which takes the underlying nature/physics of the measured variables into account, as opposed to regular statistics methods which treat these variables just as random numbers. An example: suppose you're lost at sea at night and you only have some inaccurate device (with precision +/- 10 miles) to measure your coordinates by stars, and a speedometer, which shows you're drifting at 10ft per second. You take a series of measurements, 1 sample per second, and it gives you these coordinates: 55 miles, 37 miles, 51, 67, 69, 60, 40, ... etc. If you apply some regular smoothing algorithm (running average, etc.), you will get the "signal" with less noise, e.g. 46, 48, 52, 62, 65, etc. But you know you are not moving at 2, 4, or 10 miles per second. Kalman filter takes the equation of motion x(t) = 10ft/s*t and measured positions to produce the most probable estimate of your position which obeys the equation of motion. The data smoothed by Kalman filter will be much smoother, and your average position will be drifting at 10ft/s, as it should. Wingsuit equations relate several independently measured variables (Vx, Vy, ax, ay, L/D) and thus can be used to effectively smooth the data. By having the redundancy of data (we have 5 measured variables connected by 3 equations), we can eliminate the two unknown coefficients Kl and Kd and obtain accurate and stable real-time Vx, Vy, L/D. Totally sweet stuff! Android+Wear/iOS/Windows apps: L/D Vario, Smart Altimeter, Rockdrop Pro, Wingsuit FAP iOS only: L/D Magic Windows only: WS Studio Quote Share this post Link to post Share on other sites
yuri_base 1 #5 December 19, 2006 Scott, I'm coming to Z-Hills in just 432123 seconds, we'll drool over FIR, IIR, Matlab, and Kalman filter then!!! A real-time horizontal/vertical speed Head-Up Display will be useful for setting the pace for big flocks, although you don't really need one since you're rock consistent leading the flocks, always.Android+Wear/iOS/Windows apps: L/D Vario, Smart Altimeter, Rockdrop Pro, Wingsuit FAP iOS only: L/D Magic Windows only: WS Studio Quote Share this post Link to post Share on other sites
KrisFlyZ 0 #6 December 19, 2006 When you get there have people with a different body types fly with your system and gather data so that you can have data from different sample. I am trying to come to Zhills around that time as well see how that goes. Kris. Quote Share this post Link to post Share on other sites
yuri_base 1 #7 December 19, 2006 As of now, the system is useless because the tubes are not oriented accurately into the relative wind (the measured total speed ~80mph is lower than my real speed), and the accelerometer is mounted statically on the helmet. I won't have weathervane ready for Z-Hills. Maybe, I'll make something right there. I'm thinking about something the size of a smoke braket, and mount it on the left foot, with weathervane sticking out into clean air. The Pitot tubes and 2-axis accelerometer will be mounted on the weathervane, the "black box" will be mounted on the leg, and wireless dashboard - on the cheststrap.Android+Wear/iOS/Windows apps: L/D Vario, Smart Altimeter, Rockdrop Pro, Wingsuit FAP iOS only: L/D Magic Windows only: WS Studio Quote Share this post Link to post Share on other sites
mccordia 73 #8 December 19, 2006 Wouldnt the tubes best be mounted on your body? As the head and feet can move around more?JC FlyLikeBrick I'm an Athlete? Quote Share this post Link to post Share on other sites
yuri_base 1 #9 December 19, 2006 Small movements should not affect readings as the weathervane will maintain the orientation.Android+Wear/iOS/Windows apps: L/D Vario, Smart Altimeter, Rockdrop Pro, Wingsuit FAP iOS only: L/D Magic Windows only: WS Studio Quote Share this post Link to post Share on other sites
yuri_base 1 #10 December 20, 2006 Looking for help from computer geniuses! The recorder is a "black box" with USB output. You can connect it to computer and monitor data live using proprietary software. Now, if you want to "massage" the data in realtime (e.g. apply Kalman filter, calculate the glide ratio, etc.), your code needs to talk to this USB device. Are there any free libraries to do this? How can I stream the data from USB? Any help is much appreciated! Android+Wear/iOS/Windows apps: L/D Vario, Smart Altimeter, Rockdrop Pro, Wingsuit FAP iOS only: L/D Magic Windows only: WS Studio Quote Share this post Link to post Share on other sites
normiss 615 #11 December 20, 2006 Here's some I played with trying toy with USB data. http://www.ontrak.net/ADUSDK/AduStreamPipe.html http://sourceforge.net/projects/usbsnoop/ http://protocoltools.com/usb/?gclid=CLn97cnFoYkCFRaaOAod8V-LQg http://www.phoenix.com/NR/rdonlyres/278702EF-CEFB-470F-9EB1-9BB34CD99684/0/wpusb.pdf as usual...more *nix apps available than Winders.. http://zedomax.com/blog/2006/10/17/diy-hack-howto-snoop-doggy-usb/ I had used some of these to reverse engineer motorcycle ignition programming devices and their software....successfully I might add.... Quote Share this post Link to post Share on other sites
yuri_base 1 #12 December 22, 2006 Thanks for the pointers! I tried SnoopyPro, I can see packets when dashboard is connected and the desktop software is in live mode, but these packets don't seem to have real data. Will try further. Re: Kalman filter. Found a couple of great articles: http://epubl.ltu.se/1402-1617/2000/081/LTU-EX-00081-SE.pdf http://www.casde.iitb.ac.in/Publications/pdfdoc-2004/vikas-ddp.pdf Totally sweet stuff!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Android+Wear/iOS/Windows apps: L/D Vario, Smart Altimeter, Rockdrop Pro, Wingsuit FAP iOS only: L/D Magic Windows only: WS Studio Quote Share this post Link to post Share on other sites