An improved attitude information fusion algorithm based on particle filtering
Lin Meng, Dezhi Chen, Sheng Bi, Wentao Chen, Wenbin Yao, Quanyong Huang, Xiao Zeng · 2013
In view of the noise and measurement errors of sensors, the data in attitude information measurement system should be filtered. Based on the previous algorithm Kalman filtering, this paper proposes a more effective algorithm using particle filtering to solve the problem of accuracy appearing in Kalman filtering. Using Bayes theory, the estimate of the state of a system is accomplished by computation of probability distribution. The data of the sensors is filtered by a prior estimate with the characteristic of the system and a posterior estimate based on the data. This process is implemented recursively and achieves a real-time estimate of the state. The algorithm proposed in this paper tries to approximate the posterior probability density by random discrete measure. It generates two sets particles each time to fuse the data of two sensors which makes the fusion more accurately. The algorithm is verified by Matlab using the data gathering from some motional vehicles and the results show the feasibility and good performance of the algorithm.