A driver abnormality recondition model based on dynamic Bayesian network for ubiquitous computing

Due to the difficulties in context management for ubiquitous computing, we propose a model based on dynamic Bayesian network, integrating multi-physiological characteristics of the original context, such as blood alcohol concentration, eye movement and head movement. From one time slice to another time slice, the model applies the simple graphical model language to identify the physical condition of the driver sufficiently in the smart vehicle space, which gives the accurate recommendations under the abnormal state(drunk, fatigue) timely and ensures safe driving behavior. The case study by simulating the environment confirms the effectiveness of the model in a real-time driving environment. In addition, the model can reason according to several context information accurately, and choose the highest priority of body state.