Driving intention inference can anticipate the driving risk in advance, drivers have enough time to respond and avoid accident. There are several models for identifying driving intention in recent years. However, these methods infer driving intention without considering the impact of past driver behavior on current station, and only take a few basic factors into account, such as speed, accelerate, etc., which reduce the inference accuracy to some extent. To attack this, a fourstep framework for driving intention inference is proposed. The main contribution includes driving behavior factors selecting analysis which can choose the main impacting factors, and improving the existing inferring model based on pattern recognition method. The improved method can consider the impact of past driver behavior on current station with add Auto-regression (AR). Experiments show that our framework can provide a good result for driving intention, including lane changing and braking intention inference. Moreover, compared to the tradition model, the improved model improves the correct recognition rate.
CITATION STYLE
Li, F., Wang, W., Feng, G., & Guo, W. (2014). Driving intention inference based on dynamic bayesian networks. In Advances in Intelligent Systems and Computing (Vol. 279, pp. 1109–1119). Springer Verlag. https://doi.org/10.1007/978-3-642-54927-4_106
Mendeley helps you to discover research relevant for your work.