A new theory of driver vision pressure energy field and its application in driver behaviour decision‐making model

Abstract

Abstract The interpretation of driver behaviour decisions is an essential part of driver behaviour research. Unlike previous studies that use driver's vision indicators or behaviour indicators as the basis for behaviour decision‐making models, this paper proposes a new concept of vision pressure energy field to describe driver's vision perception. Driver behaviours are regarded as the results of the energy fluctuation of the “Potential Energy‐Field Energy‐Kinetic Energy” cycle. The energy field model and corresponding classification method of driving risk level are presented. The micro‐effect and macro‐effect of driver behaviour decisions are considered in the decision effect evaluation process. These models are integrated into an RNN (Recurrent Neural Network) framework. After the field test data training, the model results show that the decision‐making framework with a hidden layer can successfully describe the car‐following and lane‐changing behaviours. The phenomenon of continuous behaviour change can be explained by the prediction result of decision effect level. The vision pressure energy field theory integrates the driver behaviour into the physical energy field theory. It presents a new way to interpret driver's vision perception results. The driver behaviour changes can also be successfully predicted through this theory.

Keywords

• Traffic engineering computing
• Social and behavioural sciences computing
• Neural nets