Traveller Information Systems. By Kevin S. Hutchby
Previous page button

University of Washington

Next page button

Seattle traffic An additional visit was arranged to the University of Washington to meet Daniel J. Dailey, Associate Professor in the Electrical Engineering Department, who kindly agreed to join a discussion at short notice about the ground breaking developments taking place at the University of Washington.

Traffic Condition Predictions

I spoke to Dan about the development of the traffic prediction software to find out where it was at and just what kind of data he was using as inputs in order to attempt traffic condition prediction.

All the data required is gathered from the Seattle TSMC computers via the ITS data backbone using standard TCP/IP (Transmission Control Protocol / Internet Protocol) computer networking protocols. The existence of this data source vital to the various development projects the University have undertaken. As well as traffic flow data, CCTV image data is also received, as this is also a key component for the research. The CCTV inputs to the project were currently in development during my visit. The key to the algorithms is that they attempt to measure the overall average speed of a body of traffic, rather than the speed of an individual vehicle, to feed in to the data processing.

The model currently in development aims to predict traffic conditions up to 20 minutes in advance. To achieve this Dan is using a combination of current and historical flow data. The historical data dates back about 26 months and is necessary to provide an adequate database of traffic flow, occupancy and volume information. In addition to this he has access to the AVL Metro Transit information which provides some information regarding travelling times on bus routes. Some speed detector loops do exist but not in any significant number. The more data gleaned then the more accurate travel time predictions can be made, which is an identified requirement for commuters based on TTI surveys.

The software algorithms are designed and written to process all of this information in real time to achieve the prediction results and the project currently focusing on Interstate 5 which runs north/south through Seattle running parallel with Puget Sound and adjacent Lake Union. The results generated are referred to as "cellular automata" which is a microscopic propagating technology that also works very fast on the real time data and statistics. This research is currently active with no delivery date available as yet (August 2000).

As well as providing a commuting tool user, the end product could also help with better ramp metering management and hence, further reduce congestion at intersections. Current ramp metering algorithms only use 2 minutes worth of data processing, expanding this would enable more informed control decisions to be made.

End user tools reflecting these technologies may well include speed colour coded maps similar to the congestion maps. It was agreed that wireless technologies and devices are excellent for this kind of information but that the U.S. was certainly not taking the lead with wireless devices, unlike Japan and parts of Europe!.

The history of this kind of prediction information goes back as far as 1955 but now that ITS has real momentum, combined with affordable & powerful computer hardware capable of the high levels of number crunching required (high specification Pentium P.C.s
are currently used), these technologies can advance and hopefully soon enhance the already impressive array of traveller information tools for the Puget Sound area.

The partnership with the Department of Transport is working well for this project. The State DOT fund some of the research and can provide both staff and hardware. In return the State DOT get to tap into innovative projects that have a real chance of churning out valid and marketable products. They can also achieve this without throwing money at expensive consultants. Prototypes are developed to get private partners on board which usually have a vested interest in the potential end product.

Car Park Status Information

Sample of real-time parking information for Seattle Centre Another software project developed at the University generates real time parking status information on the Internet and forwarded to several VMS signs near the Seattle Centre attraction. A partnership with Seattle Centre and consultants, the IBI group, have helped this project achieve success. Seattle Centre has had a system to monitor entries and exits from three of its parking areas since 1999. The system also directs traffic to available parking areas by utilising VMS signs in the vicinity of Seattle Centre. Software developed at the University processes the raw parking data and transforms it into a data stream known as "Self-Describing Data" or a data stream which carries it's format with it to allow it to be read and analysed easily by other systems. This data is then made available on the ITS backbone and can be accessed by other parties interested in developing applications with it. One such application updates the Seattle Centre Internet server to provide the latest parking information to potential visitors:

Both of the projects discussed highlight the usefulness of the ITS data back bone which carries a lot of the "self describing data" mentioned above. It is designed to tie ITS applications together using software to process source data, from any provider, and redistribute it as "self describing data" to any interested parties, such as Internet service providers. These are ultimately able to present the information to travellers.

Looking at a mere fraction of the projects undertaken it seems to be an ambitious yet fruitful approach to developing and distributing ITS applications.

For further reading on technical aspects of the ITS backbone you can refer to Adobe Acrobat documents available on the Internet from:
    A Component Architecture: described in "A Structured Approach to Developing Real Time, Distributed Network Applications for ITS Deployment"     Self Describing Data: described in "A Self Describing Data Transfer Methodology for ITS Applications."

Of all the projects funded for research and development at the University of Washington, there is an expected 4/5 ratio of success.

My sincere thanks to Daniel Dailey

Seattle Footnote

The quality, depth and vast quantity of the information made available to me in Seattle was almost overwhelming. As mentioned earlier I could have easily spent an extra six months researching the various past, current and future projects of the Puget Sound area and still not have enough time to take it all in. As well as all the information gleaned from tours of facilities and interviews I was constantly handed various reports and project evaluations and referred to an even greater number available on the Internet, some in non public areas. These have so far proved incredibly useful documents and one day I hope to complete reading them.

I am indebted to all in Seattle who granted me valuable time and energy to assist me with my research.

Previous page button Home page button Next page button