Realtime sampling-based trajectory generation in highway driving scenarios
The objective of advanced driver assistance systems (ADAS) in automobiles is to support the driver in his driving tasks and increase safety for all traffic participants. In this context, motion planning is responsible for the generation of feasible driving maneuvers.
The primary purpose of the thesis is to conceptualize and imiplement a planning system for highway driving scenarios. Based on environment and traffic, feasible driving maneuvers are planned and executed by the system. First, the vehicle and the environment are transferred into a phase space (positions, velocity, acceleration, time) to get an abstract model of the problem. Continuous paths, defined by their curvature as polynomial over arc length, describe the spatial movement of the car. Various acceleration profiles define how the paths are followed. Complex maneuvers are achieved by combining trajectories. The behavior of the vehicle is defined by cost components that evaluated aspects like lane and distance keeping fo the trajectories. The planning system was integrated into an existing development and simulation framework for driver assistance systems. The simulation results show, that the proposed systems is able to drive in highway scenarios. Relatively slower moving vehicles are being followed. Lane changes are planned and executed, when obastacles appear on the road.
Für Details bitte den Autor kontaktieren: firstname.lastname@example.org