HARVBot: Automated robotic solution for strawberry harvesting in hydroponics

The primary challenge that led to the engagement of the service was the need for automation in strawberry harvesting in a hydroponic farming environment. ENKI TECHNOLOGIES, S.L. aimed to reducing labor shortages in agriculture, optimise efficiency, and enhance the accuracy of fruit picking while minimising damage to strawberries. Manual harvesting in hydroponic settings is time-consuming, labor-intensive, and prone to inefficiencies.

Key digital transformation needs included:

• Automated harvesting: Developing a robotic system capable of identifying, grasping, and collecting strawberries with precision.

• AI-driven perception: Implementing a computer vision system to detect ripe strawberries and estimate their 3D positions.

• Advanced trajectory planning: Ensuring the robotic arm generates collision-free paths to access the strawberries.

• Real-time monitoring: Enabling a dashboard-based interface for tracking harvested strawberries, system performance, and process efficiency.

• Scalability: Ensuring the robotic solution is cost-effective, adaptable, and ready for commercial deployment.

The project aimed to automate strawberry harvesting in hydroponic farming by optimise efficiency, precision and reduce costs through robotics and AI-driven perception.

To address this challenge, a robotic strawberry harvesting system was developed, integrating AI-based perception, automated trajectory planning, and a sensor-driven gripper. The goal was to boost productivity and lower operational costs in hydroponic farming.

System components include:

• Robotic arm & gripper: A collaborative robotic arm with a gripper featuring a cutting mechanism and soft pads to prevent fruit damage. A custom-designed gripper was created, later optimised with slimmer fingers to access strawberries in dense clusters.

• AI-Powered perception: Real-time fruit detection and 3D position estimation were achieved using YOLO-based deep learning models. A segmentation algorithm distinguished fruit from stems and leaves, while adaptive computer vision handled lighting variability.

• Trajectory planning: A motion planning framework enabled efficient, collision-free paths for the robotic arm. A control orchestrator optimised picking sequences and decision-making in real time.

• Ultrasonic feedback system: This feature confirmed successful grasping, reducing detachment errors. Sensor data was processed in real time via a microcontroller, improving accuracy.

• Monitoring dashboard: Provided real-time data on harvesting metrics, fruit ripeness, and system performance. It enabled operators to track efficiency and make informed decisions for optimisation.

The robotic harvesting system provided by ENKI TECHNOLOGIES, in collaboration with DIH4CAT, marked a transformative shift in the agricultural sector. The technology-driven solution addressed the inefficiencies in manual strawberry harvesting, leading to higher productivity, lower operational costs, and enhanced crop quality. By leveraging AI-driven perception, automated trajectory planning, and real-time monitoring, the system significantly improved harvesting performance and scalability. Below are the measurable outcomes and business impacts achieved through this initiative.

Measurable results:

• 30% reduction in labor costs, mitigating reliance on seasonal workers.

• Enhanced accuracy (95%) in strawberry detection and picking, reducing waste.

• 20% decrease in damage rates, ensuring higher-quality produce.

• Real-time monitoring enabled improved operational insights and data-driven decision-making.

Business impact:

• Higher profitability due to lower operational costs and reduced spoilage.

• Scalability: the modular solution can be adapted for other fruits and agricultural automation tasks.

• Technology validation: the project validated the feasibility of integrating robotics in hydroponic farming, paving the way for further advancements.

Return on investment (ROI):

• Estimated ROI of 2 years, based on increased yield and reduced labor expenses.

• Scalability potential: the robotic system can be replicated across multiple farms, expanding market opportunities.

The results confirm the feasibility and financial benefits of robotic harvesting, enabling significant cost savings, improved efficiency, and a clear path toward broader industry adoption.