An AI-powered robot developed here by experts at a university in Abu Dhabi is capable of accurately identifying ripe strawberries, picking them without causing damage, and operating tirelessly across various environments, from sunny fields to controlled greenhouses.
The ‘Strawberry Picker’ bot project, led by professors from the robotics department at Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI), is done in collaboration with other departments, including machine learning and computer vision. This innovative solution, poised to support the farming industry, aims to help farmers reduce labour costs while maintaining high levels of productivity and fruit quality.
How it works?
The robot leverages advanced artificial intelligence, computer vision, machine learning, robotics, and precision agriculture technologies.
Equipped with high-resolution cameras and sensors, the robot analyses plants in real-time, identifying ripe strawberries based on parametres such as colour, size, and shape. Machine learning algorithms ensure precise identification, distinguishing ripe fruits from unripe or damaged ones.
Once a ripe strawberry is detected, a robotic arm with a sensitive gripper gently picks the fruit without causing harm. The robot combines ‘active perception’ with its manipulation capabilities, enabling it to adjust its position or grip based on environmental factors like light, obstructions, or plant movement caused by wind. Autonomous navigation allows the robot to efficiently move across rows of plants, optimising its route and avoiding obstacles through AI-powered pathfinding algorithms.
“MBZUAI’s expertise in robotics, computer vision, machine learning ensures that these robots can operate with remarkable precision and adaptability, mimicking the care and attention of human labourers but at a faster and more efficient rate,” professor Dezhen Song, deputy department chair of robotics, and professor of robotics, told Gulf News.
In addition to professor Song, core contributors include professors Ivan Laptev and Hao Li, experts in computer vision, along with a multidisciplinary team of AI engineers, roboticists, and agricultural scientists. Together, they tackle challenges in precision agriculture.
Five advantages
Professor Song noted that the AI-driven robot offers several advantages over traditional farming methods, addressing key cost-related challenges.
Enhanced precision: The robot’s ability to accurately detect and pick only ripe strawberries reduces waste and minimises plant damage, leading to higher crop yields, improved fruit quality, and better market value.
Continuous operations: Unlike human labourers, the robot can work around the clock, offering consistent performance and increasing productivity without the need for overtime or seasonal wage costs.
Cost efficiency: By automating repetitive tasks, farmers can significantly lower labour costs while maintaining high productivity. Also, farmers don’t need to train new workers every season, but rely on automated system that requires minimal oversight.
Labour shortage mitigation: As the agricultural sector faces a declining availability of manual labour, especially for physically demanding tasks like fruit picking, this robot provides a practical and reliable alternative.
Scalability: The modular design allows the robot to be adapted for different crops and farming environments, extending its utility beyond strawberries. By automating repetitive tasks, farms can scale operations without proportionally increasing labour costs, making it easier to meet growing demand.
Apples, tomatoes too
Professor Song noted that the ‘Strawberry Picker’ is designed to operate in diverse environments, including varying climates and terrains.
“Its advanced sensors allow the robot to adjust to different lighting conditions, such as bright sunlight or low indoor lighting in greenhouses. The hardware is built to withstand environmental challenges, including varying temperatures, humidity, and dust. The AI models can be fine-tuned for specific environments, ensuring optimal performance in outdoor fields, indoor vertical farms, or polyhouse settings.”
He pointed out that the robot can be deployed for other types of items like tomatoes, apples or bell peppers.
“Yes, the technology behind this robot is inherently scalable because of its reliance on machine learning and adaptable hardware. The same AI algorithms can be trained to recognise different fruits or vegetables. The training process would involve gathering a dataset of images and parametres specific to the new crop. This scalability ensures that the technology can serve as a multi-functional solution for precision agriculture, enhancing productivity across various farming activities,” professor Song added.
