Prediction tool reduces traffic stalls
In order to generate short-time traffic forecasts, the system processed huge numbers of traffic data. According to IBM, with an accuracy of better than 90 percent, the results were very encouraging.
The City disposes of a traffic management centre with 20 traffic cameras at critical locations and about 150 measurement points. Hitherto however, no computer-based traffic analysis was available. The purpose of the pilot project was to find an answer to the question how traffic optimizing by means of traffic prediction could help to reduce the number of congestions. Traffic experts believe that precise short-time predictions and traffic situation analysis can detect imminent traffic densification in urban areas before they materialize and counteract by means of appropriate measures. In addition, it is possible to warn drivers early enough. Their benefit is that they can use these informations for better estimations of the time required for a specific trip, and they could chose alternative routes. In addition, these data could support decisions on which means of transport would be the best one in the specific situation.
Experts from IBM Germany and the IBM Watson Research Center analysed data gained through real-time measurement points over a timespan of six weeks. In this task the IBM Traffic Prediction Tool achieved amazingly accurate results. In a comparison of the predictions and the real-world data, the accuracy of the average travel speed short-term prediction over a timespan of 30 minutes was 94 percent, the traffic density could be predicted by an accuracy of 87 percent. These results prove that the traffic management software used in the project is an appropriate tool to optimize traffic flow.
The IBM Traffic Prediction Tool is a component of the IBM Intelligent Operation Center (IOC) which provides integrated processing of traffic data along with ambient events (Cologne frequently suffers from inundations through the Rhine river) and public security events.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
