Abstract:
This research presents a methodology to estimated road traffic from mobile probes, in which data
from 3,000-4,000 GPS-equipped taxis are used as a testbed. This research is divided into three parts.
The first part presents a road-traffic estimation system from a GPS-equipped car using the manually
tuned fuzzy logic and the adaptive neuro-fuzzy technique. The system is designed to emulate
human's opinions on three levels of traffic congestion from GPS data. The second part presents a
comparison of two methods used to estimate road traffic from multiple GPS equipped cars that
periodically report their status back to the data center. The first method, the velocity-based method,
uses an average of instant speeds reported from GPS devices to represent traffic status. The second
method, the position-based method, uses an average of velocities from point-to-point on the earth
coordinate to estimate traffic status. The final part, Cell-based traffic estimation, is the prediction of
congestion in areas across Bangkok metropolitan area. We divided the area into several grid squares
called cell. Each cell keeps track of travel time used to cross cell's boundaries. The travel time is
updated in real-time according to the incoming traffic data.