Abstract:

Roadway lane marking detection is essential for lane departure warning and enabling autonomous vehicles. Instead of using lane markings only as visual aids to indicate the boundary of two adjacent lanes, it is possible to also include digital information in the lane to provide geographically specific information. This enables in-tunnel navigation data, location-specific service advertisement, construction zone and lane merger warning. Technologically, such digital information should be decodeable by inexpensive equipment, even while vehicle is traveling at highway speed. We propose a Differential Orthogonal Encoding scheme that allows information to be decoded even when the captured image of lane marking suffers from a high level of motion blur, and we show the superiority of the encoding scheme through physical experiment and theoretical analysis. A vehicle equipped with IMP/cm**2 camera traveling at 30m/s (108km/h) can decode lane marking with data density of 77kb/m**2.