Agerpix launches an advanced digitalisation project for trellis-grown table grapes, combining autonomous navigation, computer vision and predictive harvest models

 

A few weeks ago, we met in Soria with the full PREVID project team to hold our kick-off meeting. Two intensive days of work, plenty of ideas on the table and, above all, confirmation that the consortium we have brought together has exactly the mix of expertise this challenge requires.

 

In this article, we want to explain what the project is about, why we believe it could mark a turning point in table grape monitoring, and what we will be working on over the next three years.

 

The challenge: monitoring table grapes under trellis systems is not easy

Table grapes grown under trellis systems are among the most demanding fruit and vegetable crops when it comes to agronomic monitoring. The trellis structure — a dense canopy covering large areas horizontally — makes aerial monitoring with drones extremely difficult and turns manual scouting into a costly, time-consuming and difficult-to-scale task.

 

However, decisions made in the field — when to thin bunches, when to harvest, which areas are showing ripening or plant-health problems — have a direct impact on export quality and commercial margins. The difference between a well-monitored plot and one that is not can be measured in euros per kilogram.

 

The problem is not a lack of willingness or agronomic knowledge. What is missing are tools capable of moving through the trellis, systematically collecting data and turning that data into actionable information for agronomists and technical teams.

 

That is exactly what PREVID aims to solve.

 

The solution: a robot that walks between the vines

At the heart of the project is the Unitree GO2 EDU, a quadruped ground robot capable of navigating autonomously through trellis rows, adapting to uneven terrain and operating continuously for several hours.

 

Unlike drones — which view the crop from above and have difficulty seeing beneath the canopy — the robot moves at vine level, exactly where the grape bunches are located. This makes it possible to capture images from the same perspective as a technician walking through the farm, but in a systematic, repeatable and fatigue-free way.

 

The robot will be equipped with Agerpix OnFRUIT cameras, a high-resolution imaging system designed to capture detailed images even under challenging lighting conditions.

 

The four outcomes we are pursuing

The project is structured around four development outcomes that will be progressively delivered between 2026 and 2029:

 

R1 — Initial prototype adapted to trellis systems

During the first months, we will define the system requirements, adapt the robot to the real conditions of the pilot farm and establish communication protocols between the onboard hardware and the digital platform.

 

R2 — Functional prototype with integrated sensors and computer vision

We will integrate the cameras and sensors into the robot and develop the first computer vision algorithms, including grape-bunch counting per vine and per hectare, size estimation, colour detection and vegetative-vigour analysis.

 

R3 — Integrated system and predictive models

This is the technological core of the project. All the data captured by the robot — bunch counts, vigour, weather conditions and routes — will flow into Codesian’s AiCrop platform, where we will train machine-learning models capable of predicting yield, export quality and changes in vegetative vigour.

The final result of this phase will be an agronomic recommendation engine integrated into the AiCrop dashboard.

 

R4 — Field validation

We will deploy the complete system in real plots operated by MOYCA Grapes, validate the models using data collected during the growing season and fine-tune the algorithms based on results obtained under real-world conditions.

 

The consortium: four organisations, one goal

PREVID is not being developed by Agerpix alone. The project is an Operational Group made up of four complementary organisations:

 

Agerpix acts as project coordinator and contributes its computer vision technology and OnFRUIT cameras. We lead the development of the image-processing algorithms and the integration of the imaging system with the robot.

 

TECNOVA represents the Operational Group before the public administration and leads all robotics-related activities, including autonomous navigation, adapting the GO2 to trellis systems, developing adaptive movement routines and establishing remote communication with the robot.

 

Codesian is responsible for the digital platform. Its solution will receive and integrate all the data collected by the robot, train the predictive models and deliver agronomic recommendations to MOYCA’s technical teams.

 

MOYCA Grapes, one of Spain’s leading table grape producers, is the end-user partner. The company provides the agronomic expertise, pilot farms and real-world validation needed to ensure that the technology we develop works in the field.

 

In addition, Vector Horizonte supports the project with financial management and reporting.

 

Funding: 100% supported by EAFRD and MAPA

PREVID is 100% funded by the European Agricultural Fund for Rural Development (EAFRD) and the Spanish Ministry of Agriculture, Fisheries and Food (MAPA), under the 2025 Operational Groups call (file reference REGAGE25e00048171574).

 

The total approved budget is €587,281.48, with a three-year implementation period.

 

This level of funding reflects the innovative nature of the project and its alignment with the priorities of European agricultural policy: digitalisation, sustainability and improved competitiveness in the fruit and vegetable sector.

 

What’s next?

The kick-off meeting in Soria marked the real beginning of the project.

 

Over the coming weeks, the consortium has several tasks ahead. Agerpix is currently working on integrating the OnFRUIT cameras with the GO2 chassis and developing the first image-preprocessing modules together with Codesian.

 

It will be a long journey, but the team is committed and the roadmap is clear. We will keep you updated.

 

PREVID is co-funded by the European Agricultural Fund for Rural Development (EAFRD) and the Spanish Ministry of Agriculture, Fisheries and Food (MAPA), under the 2025 Operational Groups call.