Abstract: A unique method of estimating real time traffic speeds using data from moving vehicles is described. The system uses low cost GPS receivers connected to wireless modems transmitting information to a central server. These data are then used to create speed forecasts for each roadway segment. Typically, roadway segments are 100-200 meters in length and the forecasts have time resolutions down to one minute for each day of the week, week of the month, and month of the year. An approach similar to finite element modeling (FEM) is used to compute the speed and travel time of any vehicle from any point to any destination using the information in the FEM like model. The speed estimates within each element are continually improved as more data is available in that cell. Standard recursive statistics are used for this purpose.

Abstract: In the US, over 40,000,000 single occupant vehicles (SOV) commute each day, often spending 30 or minutes more in congestion. In the San Francisco Bay area alone, over 2,000,000 SOVs waste over 190,000 hours in congestion each day. This not only wastes highly productive time but also pollutes at much higher rates than while moving.