In the context of present challenges for society, our aim is mainly concerned with safety and wellness applied to the experience of driving. In this project we will extend ambient-intelligence concepts and technologies to smart cars using a driver-centred perspective where comfort, safety, and well-being are priority values. In particular, we will focus on assisting the drivers with the aim of improving their awareness and driving performance in a non-intrusive way. The availability of advanced driver assistance systems (ADAS), inspired in safety and well-being, is becoming increasingly important to avoid traffic accidents caused by fatigue, stress, distractions or chronic diseases, especially since the driving population is getting older. We will investigate the suitability of driving behaviour signals, mainly CAN bus signals and pedal-sensor recordings (e.g. gas pedal pressure, brake pedal pressure, vehicle velocity, etc.), to develop models of the drivers behaviour in different environments (highways and roads), and performing different driving tasks, with the aim of identifying anomalous driving situations and warning the driver about it. These signals are obtained in a non-intrusive manner, without disturbing the driver -as opposed to speech-signal based methods; the group will use data obtained with the "Uyanik" instrumented car (University of Istanbul).

The automotive sector has taken advantage of field programmable gate arrays (FPGA), mainly due to the high computational demands of this sector where a huge amount of signals have to be processed in real time by means of very fast electronic systems. Currently FPGAs are used as embedded platforms (SoPC) or hardware coprocessors for algorithm acceleration, and as sensor interfaces (camera sensor interface, infrared or thermal camera interface, radar sensor interface, CAN bus interface, etc). The main objective of this project is the development of embedded electronic systems, based on FPGAs, for in-vehicle deployment of ambient intelligence with the aim of improving driver performance and safety in an unobtrusive manner. This objective proposes an innovative and multidisciplinar approach for the development of ADAS which involves the following partial objectives:

i) To develop driver identification and driver status identification models based on behavioural signals obtained in a non-intrusive way and suitable for digital hardware implementation

ii) To implement efficient real-time electronic systems for the above models using FPGA-based embedded systems for in-vehicle integration

iii) To develop practical implementations of ADAS in order to verify and enhance the proposal, to determine their feasibility, and to propose alternative solutions.