A Study on Fake and Real Lane Change Behavior Prediction Based on Discrete Hidden Markov Model

In advanced assisted driving, an effective understanding of the driver’s behavior can effectively improve the safety of the driving process. In this paper, we use discrete hidden Markov models to predict lane change and real lane change behavior by collecting natural driving data from a driving simulator. First, we build a lane change road on the Prescan software, collect data from 12 groups of drivers and filter the data, select whether the four characteristic parameters of steering wheel angle, steering wheel angular velocity, yaw rate and yaw acceleration are true for lane change perform analysis. Then, on the effective data set training, the viterbi algorithm is used to predict the fake lane change and the real lane change behavior of the discrete hidden Markov model. The prediction results show that the model can use the above four characteristic parameters to predict fake lane change and real lane change behavior. The accuracy of the prediction respectively is 89.83% ,91.20% ,88.52% ,80.99%, and it has good accuracy.