- Quickly from A to B
- Reduce noise
- Minimize air pollution
Approach
In the project "KI4LSA", a solution for traffic flow optimization based on real-time data of existing LSA with artificial intelligence (AI) will be implemented and evaluated.
Technological objective
- situation in real time: In addition to the existing data from the LSAs, a real-time sensor system will be implemented that can record traffic flows with sufficient accuracy and lane fidelity using as few sensors as possible and forward them to cloud/edge-based processing. In addition, sensors will be installed to collect relevant environmental data.
- Autonomously learn strategies for real-time traffic flow optimization: The AI used for this purpose will use state-of-the-art algorithms for deep learning, reinforcement learning, and deep reinforcement learning. It will be verified to what extent the AI can be learned with a small amount of sensor data, simulation data and data of atypical events.
- Demonstrate effectiveness in real laboratory: With the help of a prototype, the achievable potential of AI for traffic flow optimization and emission reduction will be exemplarily tested in the planned LSA network in Lemgo and the improvement will be evaluated with commuters and residents.
- Transferability: To this end, scalability tests will be carried out with regard to larger traffic sections on the basis of available traffic data from other locations or through simulations. Furthermore, the collected data will be made available to other stakeholders as open data.