EMS-vision: recognition of intersections on unmarked road networks

The ability to recognize intersections enables an autonomous vehicle to navigate on road networks for performing complex missions. The paper gives the geometry model for intersections applied and their interaction with active viewing direction control. Quality measures indicate to performance monitoring processes the reliability of the estimation results. The perception module is integrated in the EMS-Vision system. Results from autonomous turn-off maneuvers, conducted on unmarked campus roads are discussed