文摘
Due to low precision and weak stability, MEMS micro inertial navigation system installed in MAV cannot work alone and achieve high-performance navigation for a long time. In order to improve MEMS performance, it is proposed to fuse MEMS-IMU information with optical flow, and correct MEMS attitude when it is diverged. Then, image coordinate of optical flow is proposed to be firstly transformed to body one for better vehicle attitude calculation. The Allan variances of MEMS sensors are used in KF filter to replace the regular white noise. The algorithm is finally testified by simulation results, which is shown that vehicle attitude modified by combining with optical flow is of good performance, with smaller error, slow divergence and better robustness.