Inertial Navigation Trajectory and Attitude Prediction Based on Improved Hidden Markov Model

In order to solve the problem that the trajectory prediction time of inertial navigation system (INS) data is too long and the attitude cannot be predicted, a trajectory and attitude prediction method of Improved Hidden Markov model (IHMM) is proposed. This method improves the conventional HMM, and uses Gaussian weighted-average moving filter to preprocess the data and eliminate the background noise. To solve the parameters more accurately, the Baum-Welch algorithm and maximum likelihood estimation (MLE) method is introduced. The Viterbi algorithm is improved by adding nonlinear transition probability function for decoding and prediction. Through experiments, the performance of IHMM is verified, the simulation and measured data are used for prediction, and the prediction accuracy of X, Y and Z axis acceleration data are 0.85, 0.85 and 0.95 respectively. Moreover, the comparison between IHMM and recurrent neural network (RNN) are also compared. The result shows that, within the range of acceptable prediction accuracy, the running time of the IHMM is about one-third that of RNN. Finally, this method can predict the trajectory and attitude of the vehicle, and provide effective information for environment recognition and prediction in scenarios with strong real-time performance.