Gluonpilot

Hello again Tom,

Thanks a lot for your help so far. I still ponder about two problems though:

1) When you're at an angle, taking a corner - how can you know the exact angular velocity of the corner (to calculate g-forces), as the gyro axis along which you measure is now under an (unknown) angle.

Do gyro's always measure from a fixed plane? If so then the above problem would be solved

2) When you know the actual driving forces from your GPS/speed + angular velocity, how do you know to substract them from the accelero data? You don't know what direction the forces are applied, so you don't know how to substract them from the accelero data in order to have only gravity direction remaining.

1) Assuming you always know the current roll and pitch angle (as it is an iterative calculation), so the combination of 2 gyro's + some algebra would give you the current angular velocity for the corner you're taking

2) Same as the above, you assume you know the current pitch and roll data to make the calculation

To satisfy these assumptions, the calculation should be done as much as possible, mostly 50 or 100 times per second.

I don't quite understand. You take angle data from a gyro, which is assumed inaccurate. With this inaccurate data, you compensate for lateral forces to aquire the same angle from an accelerometer, which then in turn compensates the gyro angle again. Doesn't this stack error upon error upon error? Is this not the definition of drift, the continuous stacking of errors?

I have the feeling I'm trying to do age-old work which I'm never going to be able to do as good as solutions out there. I can find plenty of IMU's and alike that have everything 'integrated'. Example of this would be the MPU-6050 from InvenSense.

I don't nessesarely need to figure out the algorithm myself. Aren't there simple small chips available for a reasonable price that simply output pitch, yaw and roll angles in analogue / digital format so it can be read straight away by a datalogger?

Sure there are quite a lot: just make sure you get one with the GPS integrated.

Nice reference: http://paparazzi.enac.fr/wiki/Sensors/IMU

I would use the CHIMU for low-cost and good quality. As I explained, you will need to send the vehicle speed to the module in order to get correct calculations. See manual: http://www.ryanmechatronics.com/public_ ... _REV_E.pdf