Driving Trajectory Prediction Method Based on Adaboost-Markov Model Optimization
Download as PDF
DOI: 10.23977/meimie.2019.43058
Author(s)
Shengnan Song, Yongjun Zhang
Corresponding Author
Shengnan Song
ABSTRACT
The driving trajectory prediction method based on the traditional prediction algorithm model has the disadvantages of small prediction accuracy and low matching rate. This paper proposes an improved driving trajectory prediction method based on Adaboost-Markov model. The method adaptively determines the model order m, and uses the Adaboost algorithm to determine the weight coefficients to form a multi-order Markov model. The experimental results show that compared with the fixed-order Markov model, the average prediction accuracy of the Adaboost-Markov model is significantly improved, and it has lower algorithm complexity, which is suitable for vehicle driving trajectory prediction under massive data.
KEYWORDS
Position prediction, trajectory matching rate, Adaboost, multi-order Markov, adaptive weight ratio