Real-time planning for adjacent consecutive intersections

Real-time path planning constitutes one of the hot topics when developing automated driving. Different path planning techniques have been studied in both the robotics and automated vehicles fields trying to improve the trajectory generation and its tracking. In this paper, a novel local path planning algorithm combining both off-line and real-time generation for automated vehicles in urban environments is presented. It takes advantage of an off-line planning strategy to achieve a smooth and continuous trajectory generation with adjacent consecutive intersections. Therefore, this approach allows a smooth and continuous trajectory by means of planning consecutive curves concurrently, i.e., it will plan the upcoming curve and the following one in advance, taking into account the constraints of the infrastructure and the physical limitations of the vehicle. The real-time planning algorithm has been tested in simulation based on the INRIA-RITS vehicles architecture. The results obtained show an improvement in the smoothness, continuity and comfort of the generated paths.