Abstract:
With the development of advanced driving assistance systems, in-vehicle
communication and information systems, there are situations where the driver
becomes overloaded by information, creating potentially dangerous conditions.
In this Thesis a novel strategy is proposed, to prioritise and present
information.
Firstly two main criteria are extracted, that allow the ability to rank
messages: the risk associated with the non-presentation of the message, and
its relevance to the environment. Fuzzy cognitive maps enable to represent
expert knowledge and model these relationships.
Secondly, a strategy to present information is proposed. Using an importance
index, calculated from the previous risk and relevance indices, but
also information nature, time constraints and access frequency, a set of best
interfaces is selected. Furthermore design a model of driver workload is designed,
based on the multiple resources theory. By estimating in real time
the workload of the driver, the system enables to choose an optimal interface,
that should prevent overload.
This Thesis presents then the tools developed for the implementation and
testing of the model. A video capture and data transfer program, based on
the IEEE-1394 bus, enable in-vehicle real-time data capture and collection.
Moreover, a software package for replay of the acquired data, analysis and
simulation is developed. Finally, the implementation of the prioritisation and
presentation strategy is outlined.
The last part of this work is dedicated to the experiments and results.
Using an experimental vehicle, data in different driving conditions are collected.
the experiment is completed by creating data to simulate potentially
dangerous situations, where driver is overloaded with information. The results show that the information management and presentation system is able
to prevent overload in most conditions. Its structure and design allow to
incorporate expert knowledge to refine the classification.
