D.D. Schmorrow et al. (Eds.): Augmented Cognition, HCII 2009, LNAI 5638, pp. 13–19, 2009. © Springer-Verlag Berlin Heidelberg 2009 Adaptive Interfaces in Driving Rino F.T. Brouwer 1 , Marieka Hoedemaeker 2 , and Mark A. Neerincx 1,2 1 TNO Human Factors, Kampweg 5, 3769 ZG Soesterberg, The Netherlands 2 Delft University of Technology, Mekelweg 4, 2628 GA Delft, The Netherlands rino.brouwer@tno.nl, marieka.hoedemaeker@tno.nl, mark.neerincx@tno.nl Abstract. The automotive domain is an excellent domain for investigating augmented cognition methods, and one of the domains that can provide the ap- plications. We developed, applied and tested indirect (or derived) measures to estimate driver state risks, validated by direct state-sensing methods, with major European vehicle manufacturers, suppliers and research institutes in the project AIDE (Adaptive Integrated Driver-vehicle InterfacE). The project developed an interface with the driver that integrates different advanced driver assistant sys- tems and in-vehicle information systems and adapted the interface to different driver or traffic conditions. This paper presents an overview of the AIDE pro- ject and will then focus on the adaptation aspect of AIDE. Information pre- sented to the driver could be adapted on basis of environmental conditions (weather and traffic), and on basis of assessed workload, distraction, and physi- cal condition of the driver. The adaptation of how information is presented to the driver or the timing of when information is presented to the driver is of im- portance. Adapting information, however, also results in systems that are less transparent to the driver. Keywords: In-car services, workload, adaptive user interface, central manage- ment. 1 Introduction A major research effort on augmented cognition takes place in the defense domain, aiming at systems that support or extend the limited human information processes for operations in high-demand situations [1]. To augment cognition in dynamic condi- tions, the momentary human state is often sensed via (psycho)physiological meas- urements, such as EEG and heart rate [2]. New non-obtrusive methods can be used, such as camera sensors and microphones to assess emotion out of, respectively, facial expressions and voice [3]. In general, we propose to use a mixture of methods, includ- ing measures of human, task and context [4] In our view, the automotive domain is an excellent domain for investigating aug- mented cognition methods, and one of the domains that can provide the applications. First, the human is in a constrained (relatively fixed, “indoor”) position, sitting in an environment that can be relatively easily enriched with driver-state sensing technol- ogy. Second, the driver’s tasks is rather well-defined, and can be tracked well, and