Static Sensor Orientation

Hi Everyone,

I am looking to use the MPU-6050 mainly as a sensor to measure rotation velocity. I have the following situation:

My sensor will be placed inside a compartment in such a way that the gyroscope axes will not match my desired coordinate system. Additionally, the exactly alignment of the sensor will not be known and will not be easy to measure using external instruments (like a protractor, etc.). I was reading up on the DMP and thought that I could perhaps use it to determine the orientation of the MPU-6050 inside the compartment.

My plan was the following: Use the DMP to determine static orientation and from that obtain a correctional alignment matrix R. From here on, I would read the gyroscope values and multiply by R to re-align the sensors to my desired coordinate system.

Now the thing is, I would call this more of a static orientation than a dynamic orientation measurement, which is what I believe the DMP is designed for. I care about the orientation of the sensors when the system is at rest. Is there a more efficient (less embedded code) method to obtain my desired matrix R, without losing accuracy? My concern is that since I only need this measurement at the beginning, it would be unnecessary and use up quite a bit of code space to put a whole program to calculate quaternions inside my embedded processor. I would prefer to take raw measurements from the sensors, send these to a PC via UART, and then do any complex calculations on the PC. However, I am not sure about the feasibility of this approach. Alternatively, I could orient the compartment along the X, Y, and Z axes, read the accelerometer readings in each of those cases, and from this information calculate an R matrix. I would prefer not to use the second approach as it would be a bit difficult (though not impossible) to use my setup in this way.

I would appreciate any feedback.

Sincerely,

Mehdi