Autonomous Driving Behavior Prediction Method Based on Improved Hidden Markov Model

A behavior prediction method for autonomous driving vehicles based on the improved Hidden Markov Model is proposed in this paper. Firstly, several groups of initial model parameters are randomly generated to obtain a finite set of other vehicles’ possible actions. Secondly, the probability of vehicles switching between different actions, the probability of different actions and the probability of vehicles coming from different directions are studied by the Baum-Welch algorithm. Then, the statistical results obtained by Viterbi algorithm are combined with the Bayesian decision theory to determine the final results. Finally, significance hypothesis test is used to verify the correctness and reliability of the model prediction results.