CULTIVO 4.0: Real-time forecasting system for improved crop productivity
The CULTIVO 4.0 project proposes a new digital real-time forecasting system to improve mushroom crop productivity. This improvement will be possible through the use of an ICT tool fed with historical crop and production data (humidity, temperature, harvesting times, quantity of bloom, treatments applied, growth results, etc...) and an image analysis system.
To propose this system, AERTIC, the innovative business association of the ICT sector in La Rioja, coordinates the consortium formed by Bosonit, Ayecue, Cultivos Riojal, Champifresh (belonging to the Riberebro Group), Drónica Solutions, the University of La Rioja and the FOOD +i Cluster. Within this scope of promoting collaboration between companies and knowledge of their needs and resources.
The control of mushroom production has a critical point in the time window of harvesting, which fluctuates unpredictably. This time window varies from hourly intervals to days, which makes harvesting planning very complicated.
Moreover, it is not possible to know how much of the product will be harvested and when it will be available. This problem affects the entire production chain, whether for the canned or fresh market. This means that at both Cultivos Rioja and Champifresh it is not possible to plan the next stages of the production process properly and there is not enough information available to plan in advance the capacity to supply the group's many customers, which directly affects the selling price of the mushroom.
To solve this problem, the CULTIVO 4.0 project is directly linked to the application of Information and Communication Technologies in the agri-food industry, which is one of the KETs established by the RI3 strategy, key to the development of smart and sustainable growth in La Rioja.
Data analysis and construction of predictive growth and production models
Ángela Ruíz Navarro, consultant at Machine Learningand coordinating the project José Ramón Peregrinaconsultant for Data Science at Bosonit, has taken as a starting point the analysis of historical mushroom cultivation and production data, and identifies the significant variables for the construction of growth and production prediction models with Machine Learninga functional tool of technology artificial intelligence.
Drónica Solutions is working together with the University of La Rioja to provide new information to feed the prediction tool by taking digital images of the mushroom from the cultivation plants, and its subsequent storage and processing.
The overall success and objective of the project is based on the following milestones:
- Analysis of current information systems in Ayecue: collection and assessment of historical data.
- Classification of variables through variable engineering processes to detect which variables are relevant for the prediction of mushroom production and harvesting time.
- Assessment of the generation of new variables via integration of other data sources, including new measuring equipment if deemed necessary.
- Integration of an automated digital image processing system as a data source in the prediction tool.
- Generation, evaluation and selection of predictive models.
- Creation of the forecasting tool and its user interface.