Challenge
Our customer is an infrastructure automation company that specializes in the development and implementation of cutting-edge control systems for critical infrastructures (tunnel systems, traffic control, energy management). Hundreds of sensors are installed in a modern motorway tunnel to detect events such as slow vehicles, stationary vehicles or accidents. The goal was to use a machine learning model to significantly reduce the number of false reports of critical events without reducing the number of correct messages. This required the effective combination of sensor data for more accurate event reports to the tunnel center.
Solution
- Conducting data thinking workshops to analyze processes and data
- Implementation of data pipelines to bring raw sensor data into a suitable target format for a machine learning model
- Test a wide variety of ML models to determine the optimal configuration
- Development of the user interface for employees working in tunnels
- Implementation of APIs to provide ML models for productive operation
- Continuous optimization of machine learning models (MLOPs), data pipelines and sensor technology
Business benefits
- False reports were reduced by 80%
- Highway operator operating costs were drastically reduced
- AI model fully integrated into the backend (highly automated processes) and a user-friendly user interface for employees for a high level of acceptance of the solution
