In the thesis we study the area of the interaction between a human and a machine, specifically we are focus on the special case of the interaction between a driver and the vehicle. We are interested in the multimodal interaction between those two, which means using more interface designs and combing human senses for developing effective interfaces.
In the introductory topics we study theoretical basics of human-machine interaction, because it is limited on one side with human capabilities and on the other with the interface, which must be adapted to those human capabilities. We find that the human capacity to exchange information is limited by the senses, which are quite different from each other and require specific adjustments on the side of the user interface. In addition, the limitation of human information processing capabilities and the limitation of working memory.
A special chapter is therefore intended specifically for studying human working memory, as understanding the concept of its operation is essential for the development of multimodal interaction. In this topic, we find that a person's working memory can process information obtained from different senses in parallel, to some extent. This means that one can effectively receive and process information from different channels, which is the biggest advantage of multimodal interaction.
However, working memory has its limitations, because it can only store and process a limited amount of information at some point. The human working memories workload is also called cognitive workload and plays an important role in human-machine interaction. In the doctoral thesis we consider such interaction in the case of the driver or of the driving environment, therefore it must be borne in mind that driving is a primary high priority task, which is also a safety critical. The driver must at all times take care of the position of the vehicle, monitor the surroundings, road rules and any other hazards. This task in itself causes considerable cognitive load, which must be taken into account when planning secondary tasks, which also include interaction with various devices in the vehicle.
Adequate assessment of the cognitive load of drivers is therefore very important in researching human-machine interaction, so we first developed and analyzed a system that, by measuring eye activity, provides information about the drivers cognitive workload. We used a simple eye-tracking device that was compared to a standardized method of measuring cognitive load on drivers. In the thesis we present the results of a user study in a driving simulator, which we used to evaluate this measurement method and confirm its correctness. We find that it is possible to measure the cognitive load of drivers by measuring the pupil size and blinking frequency with simple and affordable devices. The results are particularly promising because it shows that we can even distinguish between different levels of cognitive load.
In the following part of the thesis, we focus on the human-machine interaction in the vehicle. After reviewing the scientific field, we find that ii is developing rapidly and that new information systems and modes of interaction are emerging in vehicles every year. In the doctoral research, we analyzed the human-machine interaction in two parts. First, we separately observed and analyzed the different modes of interaction as independent input devices. We were particularly interested in their impact on drivers cognitive load, driving safety, user experience, and performance alone of those devices. In the second part, based on the experience and findings of the first user study, we improved the system, analyzed it as a multimodal user interface, and evaluated its usefulness for performing differently difficult tasks in different traffic situations.
As part of the doctoral research, we have developed a prototype of our own user interface that enables communication between the driver and the device through several different modes of interaction. These are a traditional tactile interface in form of buttons on the steering wheel, touchpad, free hand gesture recognition system and touch screen. In addition, the system contained various output devices - a display simulating a central cockpit display, a head-up screen in front of the driver, and an audio interface. This user interface was integrated into a driving simulator to provide a tool for ecologically valid research.
In the final part, we compare different technologies and summarize some of the main advantages of new and not yet implemented technologies, such as use of free-hang gesture interaction. We also highlight some of the major drawbacks of existing interaction methods and user interfaces, such as very popular touchscreens.