Sub-project 3000 is the horizontal subproject especially addressing the optimal joining of driver and assistance/automation in a highly automated vehicle. The general principles of a joint system, e.g. the assistance/automation modes, transitions, driver-automation interaction and dynamic task repartition, will be worked out and applied to a generic joint system demonstrator (developed in SP3000), and to the demonstrator vehicles developed in SP5000.
Starting point for the generic joint system demonstrator is an intelligent co-system, that starts with sensor data about the environment, sensor fusion and the generation of trajectories. This data is that then used to control the vehicle in relation to the automation mode, that gets determined between the driver and the automation via a driver interface and a Mode Selection Unit. A driver observation system estimates the driver state, especially driver distraction and driver fatigue. To minimize underload und overload situations, a layered approach of dynamic task repartition between driver and co-system will be used, starting with a state-of-the-art layer and expanding to more future-oriented layers that give the co-system more authority and allow for a more fluid transition of control.
On the first layer, the co-system proposes available automation levels, taking into account the co-system's state. The driver can then activate these automation levels with buttons or a stalk. As a second layer, the co-system actively urges the driver to change automation levels, e.g. based on the driver state information, and is going to a minimum risk state in case the driver is not responding. As a third, more future oriented layer, automation levels can be changed more fluidly by the driver or by the co-system , e.g. if a situation of immanent overload or danger occurs. This layered approach to dynamic task repartition opens up a clear migration path from todays ADAS systems with limited capability to highly automated systems of the future.