Page 544 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Significance

This system aims to revolutionize infrastructure maintenance by integrating AI with civil
engineering tools, enabling predictive decision-making, extending the lifespan of critical
infrastructure, and aligning with sustainability and smart city goals.

This use case aims to address the inefficiencies in traditional concrete monitoring methods by
introducing an AI-powered system capable of real-time, non-destructive evaluation of concrete
structures. The system uses CNN models to detect cracks and regression-based models to
predict moisture content using sensor input. It helps construction engineers and maintenance
teams identify early-stage damage and moisture anomalies, thereby enhancing infrastructure
durability, safety, and resource optimization.

Use Case Status: Concept validated, data collection and model development ongoing.

Partners
Mbeya University of Science and Technology (research and lab testing), local construction firms
(validation and implementation), sensor vendors (hardware support).

2�2 Benefits of Use Case

The project contributes to industry, innovation, and infrastructure, sustainable cities and
communities, and responsible consumption and production by improving infrastructure
resilience, supporting early detection of construction issues, and reducing material waste.

2�3 Future Work

Planned steps include:
• Expanding dataset using field data
• Collaborating with construction firms for real-world pilots
• Improving model accuracy and generalization
• Mobile application development for on-site use
• Integration with Building Information Modeling (BIM) platforms

3 Use Case Requirements

REQ-01: Integration of sensors (moisture, strain, load) is critical for data collection

REQ-02: Real-time crack detection using camera images is critical
REQ-03: Data transmission to an edge or cloud server is critical

REQ-04: AI model deployment on edge device or cloud platform is critical
REQ-05: User interface/dashboard for visualization and alerts s critical

508