Autonomy and individual mobility of elders becomes a societal issue for many countries. Thus, driving is a complex and demanding task, liable to generate specific driving errors and difficulties for a safe driving, even more for elders. In order to both maintain older drivers mobility and to avoid such road safety risks, a solution concerns driving assistances development, providing a technological support for this group of drivers. In this frame, it is needed to observe and to analyse actual driving behaviours and specific difficulties of older drivers in ecological driving conditions, using an instrumented vehicle on open roads. The aim is then to ergonomically specify and design future Senior-Adapted Driving Assistance Systems (S-ADAS) based on monitoring functions embedding driver's activity model within the technology for adaptive Human Machine Cooperation. This paper introduces the Human Centered Design approach we implemented and presents preliminary results in terms of in depth analysis of older drivers' activity which constitutes the first step towards S-ADAS.