In developing a traffic information center in Seoul, Korea, the authors developed a data fusion formula for deriving the link travel time with insufficient traffic data sources. The sources are loop detectors, probe vehicles, CCTVs, and qualitative information from reporters and anecdotal sources containing the dynamic traffic information. For the loop detector, the average link travel time is supposed to be derived from the linear regression formula based on flow and occupancy. For the probe vehicle, it is to be from the actual cruise time report based on the GPS and wireless data communication, with digital map referenced along with GPS data. The CCTV snapshot has been converted into the six level qualitative traffic status, providing the roughly equivalent estimate from the historical link travel time. The reflection of the anecdotal sources on the link travel time is now in progress. The fusion formula is based on the linear combination of the above sources and expected to come out after the calibration process is fully finished. It uses fuzzy logic for allocating the weights of each data source's reliability. It is expected that the formula would play an important role in calculating the link travel time, which in turn will be distributed and stored as the historical link profile.
Abstract