Design of Real-Time Transition from Driving Assistance to Automation Function: Bayesian Artificial Intelligence Approach

Forecasts of automation in driving suggest that wide spread market penetration of fully autonomous vehicles will be decades away and that before such vehicles will gain acceptance by all stake holders, there will be a need for driving assistance in key driving tasks, supplemented by automated active safety capability. This paper advances a Bayesian Artificial Intelligence model for the design of real time transition from assisted driving to automated driving under conditions of high probability of a collision if no action is taken to avoid the collision. Systems can be designed to feature collision warnings as well as automated active safety capabilities. In addition to the high level architecture of the Bayesian transition model, example scenarios illustrate the function of the real-time transition model.