The paper addresses the problem of motion autonomy of Cybercars across a urban intersection. Cybercars are small electric city vehicles aimed at navigating autonomously. In the context of a crossing, the motion generation together with its safety are critical issues. The proposed approach to the problem lies in the coupling of perception and planning capabilities. A new car to car communication algorithm provides necessary information to a trajectory planner capable of iteratively generate safe trajectories within a dynamic environment in order to drive Cybercars safely through the intersection. The main contributions of this work are the development and integration of these modules into one single application, considering explicitly the constraints related to the environment and the system and to provide an original answer to the problem of intelligent crossing.