Combining signals from accelerometers and gyroscopes is a widely used way to estimate robot attitude. However, when using a Kalman filter in this case, the measurements are vulnerable to dynamic accelerations which will result in substantial attitude estimation errors. The attitude acquisition method presented in this paper takes an attitude quaternion as system measurements and uses a Kalman filter to fuse signals from MEMS gyroscopes, accelerometers and magnetic sensors. In order to remove the influence of dynamic accelerations, when dynamic accelerations are found to be significant, a Quaternion-based Strapdown Navigation System (Q-SINS) algorithm is only applied without the Kalman filtering. When the dynamic accelerations are not significant, both the Q-SINS and the Bi-vector algorithms are utilized and fused using the Kalman filter for improved system performance. Compared with some other highly nonlinear and complicated attitude algorithms, the Attitude and Heading Reference System (AHRS) proposed in this paper is computationally less expensive and more suitable for real-time applications.

This content is only available via PDF.
You do not currently have access to this content.