Analog Devices

[AustinSchuh]

Originally Posted by billbo911

We were fortunate enough to receive two of the ADIS16448 boards via FIRST Chiose. Based on this thread we chose to only use the gyro output data and not the Kalman filtered data for our headings.

Did you notice real world issues, or were you just being careful?

For autonomous, the drift is low enough without the filter that you won't have seen an issue. Probably for an entire match as well. The gyro in the ADIS16448 has noise properties of the ADXRS450, which would put it at 25 degrees/hr of drift.

Playing with the ADIS16448 properly is on my list of things to do this summer. Juan also recommended looking at the ADIS16460, which looks even more awesome. 6 degrees/hr of drift, and more dynamic range. I've been working through how to get a ground truth measurement (I think I can borrow a pair of GPS units and use those with drtk to get an accurate position) so we can try various algorithms and benchmark performance.

The ADIS16448 let us add some pretty cool features this year. We wrote a "down estimator" which estimated the direction of gravity using a gyro and 2 accelerometer axis. We used this in auto mode to detect when we crossed the defenses, and also to compensate our shot angle for the robot not being flat on the ground.