Conceptual model of tea picking robot. (PHOTO: VCG)

By LIN Yuchen 

Tea picking, mainly done by hand with a short and intensive picking period, has seen a technological transformation with AI-powered robots. In China's southeast province Zhejiang, known for Longjing tea or Dragon Well tea, robots developed by Professor Wu Chuanyu of Zhejiang University of Science and Technology are now seen using mechanical arms to precisely pick tea leaves in local plantations.

A large number of photos of tea tree buds and leaves were input into the tea picking robot system in early stage, according to Associate Professor Gui Jiangsheng from Wu's research team. It can then remember the shape and texture of the buds and leaves, and summarize their characteristics by producing its own photos. The more photos input, the richer the robot's knowledge base is, providing more accurate recognition of tea trees.

The technology uses a small pair of scissors at the end of a robotic arm; it finds the petiole of the buds according to the positioning information. With a single cut, the buds will be separated from the branch, while a negative pressure suction tube attached to the end of the robotic arm absorbs the cut buds into a container in the robotic arm.

The research team revealed that the AI-powered tea picking robot currently takes about 1.5 seconds per tea leaf, still lagging behind human speed.

The recognition accuracy of these intelligent robots is about 86 percent, with a success rate of 60 percent in tea picking.

"This proves intelligent tea picking is feasible, but there is still a long way to go in the future to turn feasible into usable and good," said Professor Chen Jianneng, member of the research team, adding that the robot is still in the laboratory stage, and the team will continue to develop it to improve picking efficiency and picking quality.