No translation data yet?

[paleajed]

The Wrap920 also sends information on head translation on X, Y and Z axis.
Are you planning these for this driver any day soon?

Also what about calibration?

[jpzk]

We're currently working on a broken code in the master branch, because of converting the c++ code into c code - and we thought in this early stage nobody cares about this driver. Nevertheless it works (rotation on X,Y,Z axis & with calibration and error correcting). Check out this commit https://github.com/jph10191/wrap920vr-linux/tree/1a4714761f3dcf958da3155221fa0513e7b18af9 .

Head translation on X, Y, Z axis, this is a new information to us - we've re-engineered the protocol and used the sensor information to calculate the rotation angles, the calculation of the head might be possible but isnt planned yet. Calibration is, as I already mentioned, implemented, to test the driver compile the example application and run it. First you calibrate the zero rotation e.g. by lying the wrap920vr on the table. Then as the output of the example application switches you have to rotate the glasses in any possible way. Both calibration phases start sequentially without a pause. After calibrating the wrap920 the rotation angles of the wrap920vr should work correct.

If you are a C coder and own a wrap920vr feel free to support this free driver by sending pull requests (after the conversion to C), maybe the calculation of head translation ;-). Best regards, Jendrik Poloczek

[jpzk]

I forgot to mention: you have to link /dev/hidrawX, X is for a number associated with your wrap920vr, to /dev/vusix and change permission to read accordingly. See AttitudeSensor.h.

[paleajed]

Dear Jendrik,

You can count me as at least one person thats very interested in the development of this driver. Thanks for all the work you already put in it. One of these days I will put it to the test to see if I can move development of my VR application (currently in very early stage) from Windows to Linux, which is to me a much more agreeable development environment. The Vuzix driver being Windows only forced me on this platform.

As far as developing any head translation features, I can remark here that I read in Vuzix SDK documentation that support for head translation in their driver is rudimentary at least, only hinting to give results about very general head direction. I wonder if they are scamming us all saying they have a 6DOF tracker with no proof of this as yet given. However lets be positively-minded and wait for them to get their SDK up to scratch, then you (or who knows maybe I) can translate their efforts into the Linux driver...

Keep up the good work!

[aubade]

Here's another person interested in Vuzix USB support in linux!

[jpzk]

Great! I'ved converted the C++ into C code in a fork (https://github.com/jpzk/wrap920vr-linux). It hasn't been tested so far, i don't own the glasses, so next time we work together (jph10191 and me) we'll test the C code and write a little example. When everything is fine I'll merge the code into this repository. So watch the repository to notice any important update. Contribution on the C fork (https://github.com/jpzk/wrap920vr-linux) is highly appreciated.

Best regards,
Jendrik Poloczek

[caiosba]

I'm very interested on it too. Thanks and congratulations for the great job! Just gave a try on the working C++ version and seems to work fine. Just a question: when a keep the glasses on a fixed position, pitch and roll do not change (actually, change just a little), but yaw keeps increasing indefinitely. What does it mean?

[jpzk]

Hi caiosba!

It took a while to reverse engineer the protocol. It's been a while since I worked on that code, so I can't remember every detail. Calibration is very important. The code checks if the configuration (calibration data) is already written, if it's not the case, then calibration starts in the example application. At first you need to lay down the glasses (watch the values on the screen) if the output changes you have to rotate it in every possible rotation.

It's not planned that we continue to work on the driver (we both don't own the glasses) - We used it in a VR project in our university. But if you want to hack and improve, the method IWRSENSDATA_PARSED AttitudeSensor::parseData() is the most valueable method (took a while to reverse engineer). Other code only calibrates, calculates the angles (by calculating please keep in mind sensor orientation) and filters (we used a weighted average of angles). The filter part can be optimized by applying intelligent filter algorithms (Kalman-Filter and more).

Best regards,
Jendrik Poloczek

[caiosba]

Thanks! I'll try and let you know if I have something to contribute.