Toward Adapting Cars to Their Drivers

46 AI MAGAZINE With the advance of automation and infotainment systems in vehicles, driving can be perceived as a comprehensive interactive activity occurring among human drivers and their vehicles. Classically, these users were considered the supervisors of the automated systems running in the cars. In the last decade, we have seen progress toward more autonomous driving automation, including features such as automatic lane centering, lane keeping, cruise control, and adaptive cruise control. In this article, we concentrate on the adaptive cruise control function and look into the future by attempting to learn automatically how such a system can adapt its settings to its user and context. Cruise control is a known technology that aids drivers by reducing the burden of longitudinal control of the car manually. This technology controls the vehicle speed once the user sets a desired speed. Cruise control is not only convenient, but it has the potential to improve the flow of traffic (van Arem, van Driel, and Visser 2006) and can be effective in reducing driver fatigue and fuel consumption (Bishop 2000). In this article, we focus on a second generation of cruise controls — adaptive cruise control (ACC) as it was used in the Automotive Collision Avoidance System Field Operational Test (ACAS FOT) (Ervin et al. 2005). ACC is designed as a comfort-enhancing system, which is an extension of conventional cruise control (CCC). The ACC system relieves the driver from some of the longitudi-