I am using the MPU-9150 DMP with MotionDriver 5.1.2 to get 6-axis Quaternions. I have noticed that if I disable the Automatic Gyro Calibration feature (DMP_FEATURE_GYRO_CAL), the DMP then ignores any biases already pushed down to the DMP memory - reverting to '0' calibration. This effectively renders the DMP Quaternions useless due to drift from the Gyro Bias.
The reason this is a problem for us is that we want to use the MPU-9150 as part of our marine propulsion control system - and if we enable the Calibration then a boat drifting in a fixed current will get that movement calibrated out after 8 seconds. This is very undesirable. Basically we want to calibrate ONCE and then use those biases from that point on - but the software is forcing us to either calibrate continuously, or not at all.
Is there a way to get the DMP to use the Gyro Biases that have been pushed down to it BUT NOT re-calculate them after 8 seconds of what it thinks is "no movement"?
Alternatively could anyone provide a bit more information about what is classed as "no movement"? Will a boat undergoing a fixed rate of rotation due to a steady current qualify as "no movement" and so get erroneously calibrated? Or not?
I can't believe I'm the first to notice this. My problem is similar to that described here - viewtopic.php?f=3&t=393.
This did not appear to get resolved.
Unfortunately we are on the other side of the world and could not make it to the Developers Conference to ask anyone in person. Any help will be greatly appreciated.
Cheers and Have a Good Day,
Wolfgang
- Log in to post comments
phpbb Topic ID
16278
Re: Re: MPU-9150 Gyro Calibration with DMP
Ah! Thanks for posting that issue. I've been hitting my head against a wall trying to get DMP gyro bias to work when auto calibrate is disabled. I've emailed developer support and posted a question here - no response yet :/
In the end I gave up and pushed bias values to the mpu rather than dmp.
Rob
Re: Re: MPU-9150 Gyro Calibration with DMP
Glad I could help Rob. I cross-posted this from the other board 'cos it was sitting there not getting answered. My next step is to e-mail developer support - although there seems little point as you've already tried that.
I also spotted the function to set the bias values in the MPU and am resorting to that too. I still hope to find a way to apply the bias in the DMP (without automatic re-calibration) as that seems to have a greater accuracy. I admit I haven't actually calculated the scale of the DMP bias so I could be wrong, but the gyro bias in the MPU is in 1000 deg/s units.
Cheers,
Wolfgang