GPS/GIS Based Electronic Road Pricing System Design
A probabilistic algorithm is designed to match the initial position of the vehicle. Since the actual location of the vehicle is never precisely known, so we determine an error ellipse, i.e. confidence region, that vehicle is likely to be within. From estimation theory, the input and output signals can be modeled as stochastic process. Variable associated with the true and measured values can be modeled as random variables. Variance-covariance information is propagated through appropriate algorithms to derive the variances and co-variances as functions of the original random variables or as functions of parameters estimated from the original observations. These variances and covariances are used to define confidence region. The determination of the confidence region should also consider the map accuracy as well as the road width.
Searching process proceed until there are candidates within the region. A match completed if there is only one road link cross or within the region. If more than one candidates exist, the candidates are eliminated with the following standards until the only correct link is matched: direction difference between road link and vehicle traveling, traffic restrictions such as one-way road, distances between vehicle position and candidate link. One matched link can be used to verify the immediate past link with their topology relationship.
After a valid start point is known, only three situations are to be considered: vehicle on a road link between intersections, vehicle near an intersection while no turn is detected, and vehicle near an intersection while a turn detected.
Suppose the route traveled is O-M-N-P-Q, the map matching processes are as following (fig. 3).
Suppose the recorded links in one month for a vehicle are Li (i=1,2,3, … , m), then the monthly payment P for the vehicle can be calculated according to price scheme in the database,
- Matching the start position to initial location O on link L1, record the distance OM by GPS and DR distance sensor.
- When vehicle near node M, three possible connections are considered and match to link L2 while no turn is detected at M.
- Among 3 possible connections at N, when a turn is detected, link L3 is selected according to azimuth measurements or angle turned.
- Repeat the same processes above.
- The vehicle trace recorded is L1-L2-L3-L4, …
When vehicular position is collected via wireless communication, traffic flow analysis will be made in the management center to provide dynamic route guidance for vehicle on the road and then improve the quality of service of the transportation. If MM is done in the management center, no additional cost is added into the in-vehicle unit. When MM is done in the in-vehicle unit, additional storage space is required for map database, but the storage media is so cheap even for hundreds of megabytes memory that it can be neglected. But How to make the charging scheme and accurately analysis the traffic flow to provide efficient dynamic route guidance need to be further studied.
- X F He. C L Law. & K V Ling. GPS based IPS for ERP vehicles. IEEE AESS systems Magazine, Vol. 16, No.9 p10-14, Sept. 2001
- Y L Zhao. Mobile phone location determination and its impact on Intelligent Transportation System. IEEE Transaction on ITS. Vol. 1, No. 1 p55-64 Mar. 2000.
- D K Yang. A G Xu. &W D Xiao. Car Navigation system using GPS and wireless communication. 3rd ICICS. Oct. 15-18, 2001, Singapore
- A G Xu. D K Yang. F X Cao. etc Prototype design and implementation for urban area in-car navigation system. The IEEE 5th International Conference on ITS, p517-521, 6-9,Sep. 2002, Singapore
- Joachim Tisal. The GSM network: GPRS evolution. John Wiley & Sons Ltd. 2001
- Xu Aigong, Ling K V, Law C L, Yang Dongkai, In-vehicle GPS/DR Navigation System Development and Test Results in Singapore, Wuhan University Journal of Natural Sciences, Vol.8 No.2B, 2003, pp544-549.