Attitude Estimation Using Parallel Quaternion Particle Filter Based on New Quaternion Distribution

ZHOU Zhaihe,Yulu ZHONG,ZENG Chuanwei,TIAN Xiangrui,Zhaihe ZHOU,Chuanwei ZENG,Xiangrui TIAN
DOI: https://doi.org/10.2322/tjsass.64.249
IF: 1.0709
2021-01-01
TRANSACTIONS OF THE JAPAN SOCIETY FOR AERONAUTICAL AND SPACE SCIENCES
Abstract:An improved algorithm is proposed to manage with the huge computation burden of the quaternion particle filter in aircraft attitude estimation. Based on the particle filtering frame, the new filter provides robust performance for nonlinear and non-Gaussian stochastic systems. And the posterior distribution of the new estimator is approximated as a new quaternion distribution to realize parallel computation. In addition, similar to the extended Kalman filter, this new method implements time update by replacing particles update with linear transformation to reduce computational complexity. Numerical simulations are carried out to compare the new algorithm to the extended Kalman filter and to quaternion particle filter in simulation results. The simulation results indicate that this estimation technique has faster convergence rate than the extended Kalman filter and takes less computation times than quaternion particle filter under the same accuracy as quaternion particle filter.
engineering, aerospace
What problem does this paper attempt to address?