Traffic sign detection, state estimation, and identification using onboard sensors

Abstract

Traffic signs along the roadways are used to convey important information about the road and the environment to drivers, pedestrians, and possibly many Intelligent Transportation System (ITS) applications. A map of traffic signs provides a priori information to ITS applications, allowing the information to be used in a wide variety of detection, positioning, and vehicle control applications. In this research, a probe vehicle with onboard sensors is utilized to survey traffic signs along the roadways of a test site. A sensor platform equipped with GPS/IMU, 3D LIDAR, and a vision sensor is employed in a traffic sign detection, state estimation, and identification framework to map the state of traffic signs and to identify them from a list of known traffic sign templates. Data collected from the test site is used to test the proposed framework. The mapped traffic signs are verified using the carrier-phase Differential Global Positioning System (DGPS) surveyed locations of the sign posts to show that the errors are in the centimeter to decimeter levels.