"Influence of Traffic Context on ADAS Performance"
Influence of Traffic Context on ADAS Performance
Sprache des Titels:
Advanced Driver Assistance Systems (ADAS) have to be able to operate in various
traffic conditions to fulfill their main function that is improving safety and comfort.
Due to improved sensing technology and higher computational on-board power, ADAS
can be extended to consider additional control targets besides delivering their main
function such as reducing fuel consumption, decreasing travel time or lowering the
amount of abraded tire material. But the achievable performance of these extensions
may strongly depend on the specific traffic situation and it might not be necessary
nor sensible to try to improve all additional targets at the same time. In this thesis,
first the question is answered whether traffic context affects the performance of the
additional control targets, and then whether this relationship can be exploited for
vehicle control. Therefore, traffic context is seen from two different perspectives: the
macroscopic and the microscopic perspective. While the macroscopic view takes into
account information of entire traffic situations, the microscopic perspective considers
the existence, type and average velocity of the preceding vehicle.
With the help of a naturalistic vehicle trajectories dataset and simulations of an
Adaptive Cruise Control (ACC) equipped vehicle optimizing additional control targets,
it is shown that traffic context indeed affects the performance of the additional control
targets. A method which allows choosing the optimization target based on on-line
measurements of the microscopic context is proposed. Additionally, a method that
determines the best trade-off between two different additional control targets based
on on-line measurements of macroscopic context is presented. The developed methods
are tested within simulations to show the relevance of the problem and of the proposed
The findings of this thesis indicate that different traffic situations do require prioritization
of additional control targets, and it is shown that the performance of
the controlled vehicle with respect to the additional control targets increases when
considering this traffic context dependent prioritization.