Automatic Orthographic Road Image Generation Using in Vehicle Sensors for ADAS

Abstract

This paper proposes a technique to generate orthographic road images and lane-level road maps suitable for advanced driver assistance systems (ADAS) using in-vehicle low-cost sensors, i.e., a single-frequency GPS, speed sensor, yaw rate gyroscope, and rear-view camera. This research also aims to generate accurate digital maps from vehicle driving data using the probe technique. This proposal should help to reduce the production costs of accurate digital maps that include road markings and road networks. The key feature of this method is the integration of probe data based on accurate trajectories and heading angles to generate digital maps. Comparisons between commercial orthographic road images and our proposals show that the proposed method automatically generates an orthographic image with a similar precision to that of the survey level method of commercial ones.