I have tested with the state estimation thread slowed down to match the current data rate, which as I understand it is 2 Hz, 1 Hz on state estimation seemed to play nice. I've added print statements that work with Teleplot, which you should get. here is a nice set of plots, with me just leaving the VN sitting on the table
I've added proper scaling of the VN output into the units Fusion expects at the parsing step, and the raw values from the IMU are so good already (aside from gyro drift which the filter does internal compensation for), so I don't think the accel/gyro configuration matrices really need to be modified. Or at least, its not worth the time to conduct an analysis on whether that makes a meaningful difference. Magnetometer is a separate thing with its own empirical compensation parameters, see #57
Next steps for this would be to reconfigure the VN to output raw data at a higher rate, and see if we can boost state estimation up to match.
Finally, I played around a bit with the #defines so that we hopefully don't run into any issues where we load a version that uses the ICM instead of the VN on the fight board.
Closes #41
I have tested with the state estimation thread slowed down to match the current data rate, which as I understand it is 2 Hz, 1 Hz on state estimation seemed to play nice. I've added print statements that work with Teleplot, which you should get. here is a nice set of plots, with me just leaving the VN sitting on the table
I've added proper scaling of the VN output into the units Fusion expects at the parsing step, and the raw values from the IMU are so good already (aside from gyro drift which the filter does internal compensation for), so I don't think the accel/gyro configuration matrices really need to be modified. Or at least, its not worth the time to conduct an analysis on whether that makes a meaningful difference. Magnetometer is a separate thing with its own empirical compensation parameters, see #57
Next steps for this would be to reconfigure the VN to output raw data at a higher rate, and see if we can boost state estimation up to match.
Finally, I played around a bit with the
#defines
so that we hopefully don't run into any issues where we load a version that uses the ICM instead of the VN on the fight board.