|
Re: What to do next?
As long as your samples are occuring consistently, you should be able to have a consistent error, which you seem to have, although I'm just eye-balling the numbers. I would try tweaking the gyro calculation constant (I believe thats what Kevin calls it).
The gyro gives whatever rate of rotation it has at the time of the call. Rate wise, it will always be accurate. If you don't call it quickly enough (your question), you may miss subtle changes. This is important because you are integrating the rate. Not having the smoothest slope (or as many samples of gyro rate) means you MAY be losing accuracy due to that. However, the processor works so darn fast, and our robots move so relatively slow, that I doubt that sample rate is the cause of your problem.
I hope I've made sense....
|