South Korea's heavily-congested roads frustrate drivers and waste a lot of time and fuel.
To help ease the increasingly chronic problem, local researchers have developed an AI system that can predict traffic conditions.
Park Se-young has more.
Cars and buses line the busy streets of Seoul.
Sitting in a traffic jam is not only frustrating.
It's also linked with a number of health problems, stress and greater exposure to pollution.
In the hope of easing traffic flow, Korean researchers have developed an artificial intelligence system that predicts traffic conditions in the next five to 15 minutes.
Unlike previous systems that rely on probability and statistics, …this new system uses a deep learning algorithm that considers real-time traffic situations.
It learned three months of traffic information …like historical average speeds in certain areas, the congested parts of roads nearby …and traffic during rush hour.

"It can predict conditions for the next 15 minutes at an error rate of less than four kilometers an hour. It can tell with 78-percent accuracy whether roads will be congested."

Congestion levels and average driving speeds predicted by the system are visualized using colors and shapes for easy understanding.
The researchers plan to share the visualized data on the Urban Traffic Information Center website, …and the system can also be used by broadcasters and navigation units to find the best route when traffic gets bad.
Park Se-young, Arirang News.