Build cycle: AI for reusing and recycling of material from construction waste

Purpose and goal

The project will develop technology that enables circular material flows and increases the recycling rate of what today is seen as waste from a construction site. The project aims to improve the sorting of waste at source by developing and using AI-based computer vision models to improve the sorting and management of construction and demolition waste. By collecting the right data, these models can create the conditions for more of today´s waste to become resources in the circular economy.

Expected effects and result

The theory the project is based on will be further developed and then demonstrated in an operational environment. The models for AI-based computer vision must identify missorted objects both at the collection point and during storage, which increases recycling rates and enables circular material flows. The AI ​​models will be improved and adapted for new material flows. This ensures that the method and technology that the project develops are future-proofed and can be adjusted for future needs.

Planned approach and implementation

The project aims to develop and implement AI-based computer vision models to improve the sorting of waste already at the collection point. Data collection will take place at selected locations to generate as representative data as possible. All data will be processed and harmonized for training the AI ​​models. Through transfer learning and training of AI models from previous and parallel projects, active learning methods shall be used to reduce the need for manual annotation.