IDRES: A rule-based system for driving situation recognition with uncertainty management

International audienceThis paper deals with the recognition of particular temporal sequences of a dynamic system, particularly in the driving situation context. The aim is to propose a method to recognise the manoeuvre performed by the driver from the data of sensors installed in a vehicle. The developed system named Intelligent Driving Recognition with Expert System is built on a two-level rule-based system that takes into account the measurements from the sensors and the sequence of the states that describes the manoeuvres. Because the inputs can be unreliable or/and inaccurate, a confidence notion is defined and modelled by a mass of evidence proposed in Dempster–Shafer’s theory. The final system recognises the current manoeuvre and evaluates the confidence of this recognition