Relative device rotation

ar2rsawseen

Hello,
This is what I'm experimenting with for a gameplay of one of my apps, where the character moves based on device rotation (the same as Labyrinth game). Problem is, that I don't want to force a player to hold the device parallel to the floor while playing (that's just sucks).
So my idea was to get current rotation of device and only use the relative movement. So take the position in which user starts the app and start from that (cause that would be just awesome).

Unfortunately as using accelerometer, acceleration that you get is basically due to gravity (its not that you have to move your device all the time, you simply lean it a bit and let the ball roll, let the gravity do its job) and if I want to use relative position, I can't count on gravity anymore (at least I can't find the way).

Next idea was to try gyroscope. I've implemented relative movement with gyroscope and it worked great. But over the time it begin to behave weird. I've tried to simply put my phone on the table and saw that gyroscope accumulates too much error (at least on my 4gen Ipod), that even low pass filter can't help it. (I've also tried high pass, just for testing and even it accumulated huge error over the time).

So if anyone can point me into right direction to implement relative accelerometer movement or proper filter to gyroscope. Please, I'm all ears

ar2rsawseen

Ok with accelerometer it's probably a task of analytical geometry, to transformating axis to starting position and recalculating all of the acceleration vectors.
But the system equation seems too complex for now, probably need to get some sleep.

But still if anyone knows any elegant solution, or any idea, please share

MikeHart

You are basically on the right track. You need to take the value of the axxis you want to use and put that in relation to the other ones. And then define a x/y value correpsonding to that which you will use to determine the changes in your app.

MikeHart

I always wanted to check this AGK code to see if it works better then my implementation but never found time for it:

http://forum.thegamecreators.com/?m=forum_view&t=194503&b=41

atilim

- Gyro has accumulated error but gives true rotation
- Accelerometer has no accumulated error but has the gravity component problem

I've heart about a technique (called sensor fusion) that uses accelerometer to calibrate the gyro's rotation values. But unfortunately I don't know any technical details. Maybe searching for "sensor fusion"+accelerometer+gyroscope gives some results.

ar2rsawseen

When I first saw sensor fusion it was used for augmented reality markers. Which was cool. I didn't think it could help to solve this problem. However, when I found this presentation:
http://www.slideshare.net/paller/better-motion-control-using-accelerometergyroscope-sensor-fusion

It actually could help a lot. Still thinking that it can be solved using accelerometer, as it's simply a mathematical problem. But the sensor fusion gives me even more ideas of interesting AR apps