A comparative study between subjective and objective results of a Predicting Human Behaviour Method based on the Expected Utility and the Iterative learning Control. Application to the Car Driving

Risk analysis in human-machine Systems has to take into account intentional human errors, called violations, in order to improve both the risk analysis and the safety of the design. Predicting this kind of human error is the best way to decrease their number and/or to reduce their consequences. After introducing the concept of barrier removal (BR) and the benefit cost and potential deficit model (BCD model), this article presents an overview of decision-making analysis. It then moves on to describe a BR decision-making model based on the BCD model. This new mathematical model, which takes the perceived utility of Barrier Removal into account and incorporates the weights of the BCD attributes, is the foundation of a new method for predicting barrier removal that relies on the iterative learning control principle to predict human operator behaviour. The results of an experimental study completed on a car driving simulator are presented to illustrate the benefits and the good performance of this new method as well as the influence of the nature of the B, C and D values (i.e. objective or subjective) on the prediction method performance.