The Flemish government and the Agency for Roads and Traffic (AWV) are partnering with ML6 to save energy and reduce light pollution on highways. By using one of our powerful AI solutions, AWV will be able to control highway lighting, turning it off when possible and only illuminating areas that need it for safety. This will be done in smaller regions and for shorter periods of time, which will help to save money and protect the environment. As they progress toward fully automatic lighting control, the operator will make data-driven decisions. Finally, we make sure that the project achieves its primary goals of energy savings and reduced light pollution without jeopardizing road user safety.
AWV manages 7,000 km of roads and highways and more than 7,700 km of cycle paths in Flanders. Their goal is to provide road users with safe and environmentally friendly mobility. To accomplish this, they work closely with different partners, share information, and employ the best ideas and technology available.
The data we need for our solution comes from different departments and in different formats. However, bringing that data together is difficult due to poor quality or excessive quantity. To solve this, we built a data lake in which we could store all the data in one location, making it easier to use for data engineering.
Another issue to keep in mind is that end users must trust our solution and recognize that it's reliable and helpful. Using a system that combines explainable AI and a human-in-the-loop can help to build that trust.
1/ Decentralization of data
2/ Data quality
3/ Big data
4/ Dependency of other implementation partners
5/ Gain trust of the end user
To improve the way lighting is switched on Flemish highways, we created a data lake on the Google Cloud Platform. This is how bring together data from various sources, such as traffic intensity, sunrise and sunset, precipitation predictions, and planned road works.
Smart algorithms are then trained to predict the road lighting switching plan 7 days in advance, as well as make real-time suggestions based on unexpected events, like traffic jams or meteorological phenomena. For this we take data from e.g. Waze or KMI.
In addition, all of our switching suggestions include a specific dimming level of the lights based on what is happening. This is an important feature for saving energy.
Finally, this platform also allows for the optimization of other decisions, such as salt sprinkling on highways and cycle paths. The use of DataOps and MLOps best practices helps reduce costs and minimize risks while improving the quality and speed of data analysis. Overall, our innovative solution helps centralize data sources and enable new data-driven and AI use cases for AWV.
For this project, the Agency for Roads & Traffic teamed up with AI specialist ML6. Due to our strong focus on the most recent applied research in artificial intelligence, we were able to provide AWV with unparalleled access to self-learning technology and assist them in meeting their project goals.