Non-linear Model Predictive Control of Connected, Automatic Cars in a Road Network Using Optimal Control Methods

摘要

In the aspiring field of autonomous driving the interaction of cars plays an important role. In particular finding optimal paths for each car whilst avoiding collision raises many problems whose resolutions are significant to ensure the safety of the passengers. To this end, we apply a non-linear model predictive control (NMPC) scheme in combination with a driving hierarchy. Herein, in every step of the NMPC scheme and for every car an optimal control problem with state constraints needs to be solved with the aim to avoid collisions and to minimize travel time. During each NMPC-step the hierarchy among the cars is redefined and adapted depending on the current state with respect to set rules which were derived from common traffic guidelines. We present numerical studies for selected road networks and car pool constellations, specifically concerning varying number of cars and the real time applicability.