Naturalistic observation of drivers’ interac- tions while overtaking on an undivided road Evangelia Portouli a,* , Dimitris Nathanael a , Nicolas Marmaras a and Vassilis Papakostopoulos b a Sector of Industrial Management and Operations Research, School of Mechanical Engineering, Ergonomics Unit, Zografou, National Technical University of Athens, Iroon Politechneiou 9, 15780 Zografou, Greece b Department of Product and Systems Design Engineering, University of the Aegean, 84100 Hermoupolis, Syros, Greece Abstract. Current models of driving behavior and designs of driver support systems are not taking adequately into account the interactions between drivers. Although a driver’s intended maneuver may not be physically possible at a specific time point, it may still be safely executed if the driver’s intention is successfully communicated to and interpreted by another driver, who alters the own future trajectory so as to enable the maneuver execution. This paper presents some empirical findings relevant to communicative and cooperative interactions between drivers along naturalistic overtaking maneuvers on an undivided road. The cues used by drivers to interpret other drivers’ intention and the drivers’ interactions were extracted through video obser- vations of the maneuvers together with the frequency of such interactions per level of traffic risk. The findings show that driv- ers use formal and informal cues to anticipate other drivers’ intention. A significant percentage of maneuvers were performed after facilitation by other drivers. Future studies should focus on this phenomenon while future models of driving behavior should incorporate communicative and cooperative interactions among drivers, so as to design adequate cooperative support systems to enhance road safety. Keywords: Communication, communication signal, cooperation, driver model, design of support systems * Corresponding author. E-mail: portouli@mail.ntua.gr 1. Introduction A lot of systems are designed today to enhance traffic safety and avert road accidents by supporting the drivers in several driving tasks [Barnard et al 2011]. These systems try to enhance the drivers’ situ- ation awareness, to predict a possible collision and to warn the driver so that he/she may take the appropri- ate counter-measures and they may even undertake the automatic control of some driving functions, if it is envisaged that the driver will not be capable to take the necessary actions to avoid the accident. Such systems usually employ algorithms to calcu- late the future trajectory of the vehicles involved, based on the vehicle dynamics. Normally they em- ploy some assumptions regarding the dynamics of the vehicles, regarding the maximum or preferred accele- rations and decelerations, regarding the longitudinal and lateral speed and steering wheel turn velocity, and even regarding the drivers’ own needs and prefe- rences [Tideman et al, 2010]. However, the acceptance of such systems by driv- ers is usually not high. The drivers’ opinion on the riskiness of the situations does not often coincide with the system output at a significant percentage [Hegeman, 2007]. Among the main reasons is that the design approach is generally not considering the communicative and cooperative interactions among drivers [Wilde 1976], which may significantly affect the situation evolution. Although a driver’s intended maneuver may not be physically possible at a specif- ic time point, it may still be safely executed if the driver’s intention is successfully communicated to the other drivers, who may alter their own future tra- jectory so as to enable the maneuver execution. This is an interaction, which all of us have encountered on several occasions on the road. Work 41 (2012) 4185-4191 DOI: 10.3233/WOR-2012-0120-4185 IOS Press 4185 1051-9815/12/$27.50 © 2012 – IOS Press and the authors. All rights reserved