Despite technological advances, robots still struggle with agricultural tasks that are easy for humans, such as weeding and harvesting tomatoes. Farmers and researchers are exploring why these jobs remain difficult for machines and what the future holds for agricultural robotics. Dutch newspaper NRC describes the situation.
Dutch organic farmer Ard van Galen faces significant weed problems in crops like onions, carrots, and peas. While mechanical removal is possible before the crops emerge, manual weeding becomes necessary as the plants grow, especially since herbicides aren’t used in organic farming. Van Galen has tested robots from various startups, but they often fail to effectively distinguish between weeds and crops.
Complex field environment
The challenge lies not in the act of weeding but in the complex and variable field environment. Robots must recognize different weed species at various growth stages and differentiate them from the desired crops, all under changing weather and soil conditions. Although advancements in sensors and neural networks have improved robotic capabilities, insufficient training data hampers their reliability.
Similar issues occur in horticulture with tasks like tomato harvesting, where robots struggle with obstacles like leaves hiding the fruit and the delicate handling required for produce. While some companies showcase robots for these tasks, fully autonomous and commercially successful solutions are still rare.
Researchers like Gert Kootstra and Erik Pekkeriet from Wageningen University suggest that robots might initially focus on automating parts of the tasks to reduce labor needs. They remain optimistic that with ongoing advancements in artificial intelligence, robots will play a significant role in agriculture within a few years.
Farmers like Van Galen hope for robotic solutions to reduce costs and advance organic farming. However, widespread adoption awaits reliable and effective technology that can handle the complexities of agricultural environments.
Comentarios