Attitude estimation with accelerometers and gyros using fuzzy tuned Kalman filter

This paper introduces the attitude estimation method of humanoid robot using an extended Kalman filter with a fuzzy logic based tuning algorithm. A humanoid robot which uses inertial sensors such as gyros and accelerometers to calculate its attitude is considered. It is known that the attitude update using gyros are prone to diverge and hence the attitude error needs to be compensated using accelerometers. In this paper, a Kalman filter model with a modified state is presented and an adaptive algorithm is used to make the filter more robust regarding acceleration disturbances. If the accelerometer measures any disturbances caused by movement of the vehicle, the characteristics of the filter must be changed to ensure confidence of the outputs of the gyros. The performance of the proposed algorithm is shown by the experiments.