Citation

Zheng H, Zhou J, Wang H. Int. J. Veh. Des. 2019; 79(4): 292-315.

Copyright

(Copyright © 2019, Inderscience Publishers)

DOI

10.1504/IJVD.2019.103615

PMID

unavailable

Abstract

To improve the lane departure warning algorithm in the vehicle lateral assistance system, an effective approach that can accurately identify a vehicle's lateral state is needed. Since steering events are the primary reason for lane departure, in this paper, a data-based departure warning algorithm is proposed that uses a hidden Markov model (HMM) to detect the lane departure state. In the HMM, the current steering event is the hidden state, and the driving state information is the observed sequence. The parameters of the HMM can be trained using the driving dataset of the driver. In addition, a further judgement strategy is used to distinguish between intentional and unintentional departure to avoid false alerts. Finally, based on a reasonable time window for identification, experiments are conducted to compare the proposed algorithm and the time-to-lane-crossing (TLC) algorithm. Quantitative analyses of the experimental results demonstrate the satisfactory performance of the data-based algorithm.

Keywords: lateral assistance system; lane departure warning; HMM; hidden Markov model; data-based algorithm; driving dataset; steering events; identification; further judgement; time window; comparison.

Language: en