Compass: absolute NSEW orientation

"tassioknop" wrote:
Hi guys,

I'd like to use a real compass: to get an absolute NSEW heading of the Oculus instead of an orientation relative to the initial state.

I'm trying to to this using the OVRDevice.GetMagnetometer function, do you suggest another way?

Also, is it safe to use the magnetometer to derive an absolute orientation? Even if I only calibrate it once?

Thank you very much!

One of the DK1 tracking modes I experimented with awhile back uses only the magnetometer to measure compass direction, but the Rift must be held level with the horizon. Another version adds accelerometer for simple two-axis inclinometer-based tilt compensation, so the Rift can be tilted up to about +/- 60 degrees or so before head tilt error becomes annoying (though the algorithm documentation says it is "accurate" to only about 40 degrees). The Rift gyrometer is not used in those experimental applications.

My Rift DK1 did not explode or catch fire during those experiments, so yes, I would say it is safe to use the magnetometer to derive an absolute orientation. However, it is only accurate to about three pixels on my DK1, showing jitter when between those pixel positions (with a resolution/latency trade-off when adding a software filter).

One caveat though -- it will not work well in environments where a hand-held compass is unlikely to be useful, such as in a steel building with a steel roof. In such an environment, using the gyro sensors (as in the OculusVR SDK) is the only realistic option. Nearby magnetic anomalies are also a problem for magnetic sensing, which is why the latest prototype of the Sixense STEM controllers now include IMU based orientation tracking to supplement the absolute magnetic sensing.

My Rift DK1 did not explode or catch fire during those experiments, so yes, I would say it is safe to use the magnetometer to derive an absolute orientation. However, it is only accurate to about three pixels on my DK1, showing jitter when between those pixel positions (with a resolution/latency trade-off when adding a software filter).

Haha so funny, I was asking if the information derived from the sensor would be trustworthy.
But you've answered that too, thank you. Probably my measurements would be disturbed by magnetic interferences and steel walls. I have to find another solution to my problem.

My problem is mostly yaw drift error, maybe I can use yaw drift automatic correction and it will be satisfatory.

I've found a very interesting article here: http://www.oculus.com/blog/magnetometer/

Your environment will have a huge effect on it. If you have a lot of steel anywhere near you, the compass will read that more than the Earth. Your sensor readings will only be locally consistent at best, move just a few feet and you might get very different results.

The physical cause for distortion on the STEM and Hydra are actually quite different from a magnetometer, so you can't really predict the performance of one based on the other. Magnetometers read a static field while STEM reads the change over time of a pulsing field.