Natural variations in the size and thickness of bean pods is a challenge for mechanical threshing machines, often leading to bottlenecks in production. Tierrina Vaqueira is attempting to solve the issue by developing a prototype for a fully-automated thresher. If successful, the project could give bean processors up to 8% higher turnover, improved quality and traceability and less waste.

Tierrina Vaqueira will test artificial intelligence technologies in collaboration with digital solution provider Virtual Intelligence S.L., focusing on automatic detection of pod characteristics to optimise the bean harvest.

The partners anticipate that AI-Thresh will be a valuable innovation for all kinds of delicate field crops.

Final report summary
The main outcome of the project is a pod classifier based on AI and machine learning techniques to enable automatic classification and quantification of pods before threshing.

The following objectives have been reached:

• Mechanical designs of threshing methods developed
  The most effective was the roller train which allows fast and continuous processing, improving productivity and efficiency.
• Integration of the roller train with a control system
  Artificial vision and decision trees were used.
• Virtual prototype developed
  This integrated the roller train method and proposed industry 4.0 technologies

The next steps are to develop a physical prototype and test the integration of the proposed methodologies

For the leading SME in the project Tierrina Vaqueira, the development of an intelligent threshing machine has the potential to improve competitiveness by reducing costs and increasing visibility. Hand threshing represents a bottleneck in fresh bean production and accounts for up to 25% of the final price.

For Virtual Intelligence, this project allowed a deep dive into the technologies, engineering principles and machine learning models used in vegetable processing – creating the basis for future projects.