Browsing by Author "Couto, Pedro Félix"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
- ADVANCING EFFICIENT AUTOMATED HARVESTING: AI -BASED FRUIT DETECTION AND DIGITAL TWIN INTEGRATION FOR AGRICULTURE 5 . 0Publication . Couto, Pedro Félix; Pereira, António Manuel de Jesus; Ramos, João Pedro FerreiraThe rapid growth of the global population has intensified the demand for food production, highlighting the importance of more efficient agricultural practices. Since crop harvesting is one of the earliest stages in the food supply chain, improving its efficiency will positively impact the following steps in the supply chain, reducing waste and ensuring higher quality products reach consumers. This stage, like many others in agriculture, faces significant challenges in becoming more efficient due to the need for human intervention and the shortage of labor. Therefore, innovative approaches and technological advancements are essential to develop solutions that can address these issues. This thesis presents a comprehensive solution leveraging artificial intelligence (AI), robotics, and digital twins to address the challenges of automating the process of fruit harvesting. It uses AI to detect fruit ripeness, robotics for automated harvesting, and digital twins to simulate the evolution of the fruits’ ripeness over time. By automating the assessment of the ripeness, the proposed system helps ensure that fruits are harvested at their optimal stage, maintaining quality standards that are critical for meeting consumer preferences. Among the AI models tested, the Gelan-c model achieved the best performance, with a box precision of 96% and mask precision of 95.8% in ripeness detection. The system’s capability to determine precise 3D coordinates of fruits enables the robotic arm to reliably align itself with the target fruit, demonstrating a high success rate in positioning. In controlled conditions, the system successfully picked and stored ripe fruits 90% of the time. These results suggest that the proposed system can significantly enhance the efficiency of fruit harvesting, reducing reliance on manual labor and improving overall productivity. The integration of digital twins allows for more accurate resource planning and ripeness prediction, contributing to a more sustainable and data-driven approach to agriculture.