“Please watch right” – Evaluation of a speech-based on-demand assistance system for urban intersections

•A driver assistance system for urban intersections was evaluated within a driving simulator study.•The system provides recommendations when to enter an intersection.•The driver activates the system by speech only if support is desired (on demand).•The system provides a safe driving behaviour and facilitates monitoring of crossing traffic.•It is preferred compared to a visual system. In a driving simulator study we evaluated a speech-based driver assistance system for urban intersections (called Assistance on Demand AoD system) which supports the driver in monitoring and decision making. The system provides recommendations for suitable time gaps to enter the intersection based on the observation of crossing traffic. Following an “on-demand”-concept, the driver activates the assistance only if support is desired. In one drive, drivers used the AoD system in every situation they experienced to guarantee that every driver had the same exposure to the system when evaluating it. During another drive, drivers were free to decide if they want to use the system or not. The experimental study compared the AoD system with driving manually and with driving supported by a more conventional visual-based system which was always active at intersections (system showing colored arrows in a simulated head-up display (HUD) to visualize the crossing traffic). This resulted in four drives the drivers had to perform. Every drive consisted of several intersections with varying traffic conditions. The drivers had to turn left at every intersection. A total of 24 drivers took part in the study; one group with 14 middle-aged drivers and another group with ten high-aged drivers. Several questionnaires and online ratings were used to assess drivers’ acceptance, perceived usefulness, benefits and specific characteristics of both system variants. In addition, driving behaviour with regard to gap choice and drivers’ monitoring behaviour (using head tracking data) were analyzed. The results show that the AoD system reaches high acceptance ratings and is preferred compared to the visual, always active system. Using the speech modality for communication and the on-demand concept were both highly appreciated by the drivers. With regard to driving behaviour, the AoD system is comparably safe as manual driving while at the same time making driving easier by facilitating the monitoring of vehicles while waiting at an intersection.