DAISY, A KNOWLEDGEABLE MONITORING AND WARNING AID FOR THE DRIVER ON GERMAN MOTORWAYS

Design and implementation aspects of the driver monitoring and warning system DAISY (Driver Assisting SYstem)1) are presented. The system is designed to work similar to the human driver except his “shortfalls” with respect to traffic safety. A mathematical formulation of danger is given based upon a parameter space description of traffic situations in terms of the time reserve with respect to a danger avoiding action. There are three main modules of the system. First, the reference driver module where a situation analysis and evaluation is accomplished by instantiation of several knowledge representation structures. Second, the model of the actual driver where knowledge about the actual driver's normal behaviour is acquired and processed. Third, the discrepancy interpretation where the output of the other modules are combined in order to warn the driver in dangerous situations. In a learning phase DAISY extracts the actual driver's driving characteristics in order to adapt the warnings later on. At the time being the system is implemented in a well equipped fixed-base driving simulator. It will be reported about some experimental results derived from simulator trials. Parts of the system will be ported to an experimental car where machine vision is the main source of environmental measurement information.