Personalization is the user-specific adaptation of a system towards individual user capability, experience and preference. It can be based on assigning the user to one of a number of pre-specified categories or clusters. More flexible models of personalization may require larger sets of individual user data to allow a robust estimation of the user’s properties and preferences. Personalizing the interface can make the interaction more intuitive or comfortable, i.e. better respecting the experiences and preferences of a user.
Human-Machine-Interaction (HMI) with intelligent assistant systems poses great challenges for the mutual understanding of goals, capabilities and mental models of interaction partners. With HRI-EU‘s incremental learning approaches, user-centered system adaptations are researched and developed aiming at a mutual understanding of learning progress and concepts. The result is trust between machine and user – a core value of HRI-EU.
The Assistance on Demand concept offers a highly personalized and context-based approach by delivering assistance only when the driver actively asks for it. Using a speech interface enables the driver to keep his visual attention on the driving task.
The concept was first evaluated in a simulator study, implemented as a left-turn assistant for urban traffic situations. When approaching a complex intersection, the driver may ask the assistant system to “watch right”. The system then detects traffic participants approaching from the right and informs the driver about potential gaps or changes in the situation. Relying on the system’s information, the driver can visually concentrate on the other directions, only checking the right side before entering the intersection.
In a follow-up study, the system’s recommendations were personalized to the individual drivers according to their preference for the gap size. In both cases, the system was very well received by the drivers, in particular when it included the personalization.
To proof the online capabilities, the system has been implemented on a prototype vehicle and tested in real-world traffic situations.
For more information
D. Orth, et. al, “Benefits of Personalization in the Context of a Speech-Based Left-Turn Assistant”, ACM International Conference on Automotive User Interfaces and Interactive Vehicular Applications, 2017