Detection of changes in driving behaviour using unsupervised learning

In many circumstances, human tests consist of monitoring and controlling complex processes by means of terminal boards and operators who often have to remain alert for a long period of time. Detecting an abnormal situation may be useful to prevent any impairment on the part of the operator. This paper presents a method for detecting any change in a driver´s behaviour, with the use of an unsupervised learning process. The human behaviours are divided into classes using “Boolean connectives” to associate data gathered from sensors. Results obtained from experimental data on a car drives doing a specific task are given