Kalman or bust topple

One of the problems you have to address with any robot is noisy sensor data. This is particularly important with a balancing robot, because without a relatively continuous output from your error function, the robot will do a lot more slamming itself from side to side than balancing (granted, a neat effect). There is at least one very simple, and successful homebrew way to accomplish this, but (as it requires a shaft encoder to report motor position, something I’d like to avoid) I’ve decided to go the path of most-common-successful-implementation and use a Kalman filter to create a smooth output from the accelerometers and gyro on my IMU chip. I’m sort of hoping to steal this code, but that will require a better grasp of the actual mechanics of the filter than I have now, and probably will wait until I’ve actually got the platform assembled and can test with the motors and wheels in place.


Post a Comment

Required fields are marked *

%d bloggers like this: