On July 28, at the daylily planting base in Tuofang Village, Yungzhou District, Datong City, Shanxi Province, a tracked daylily harvesting robot moved steadily through the fields. Farmers could remotely issue harvesting commands via their smartphones. The robot's dual arms operated efficiently, using specialized grippers to precisely pick the daylilies and place them into a storage basket. Each single picking cycle took less than five seconds. Due to operational precision limits, the robot uses a zone-based harvesting mode, with each fixed-point picking area covering 4 square meters.
"This marks the first school-enterprise collaborative daylily harvesting robot in Shanxi Province to enter field trials, and it signifies that this provincial key research and development project has moved from the laboratory to real-world field testing," said Professor Zhang Yanjun, Vice Dean of the School of Mechanical Engineering at Taiyuan University of Science and Technology. On the same day, a field application conference for the intelligent daylily harvesting robot was held in Tuofang Village. Technical staff explained the robot's working principles and demonstrated its operation process. The research team also exchanged ideas with farmers and cooperative operators on key technologies such as intelligent recognition, precise positioning, and non-destructive picking.
Yungzhou District is known as the "Hometown of Daylilies" in China, and the plant is a pillar industry for local prosperity. However, daylily harvesting is extremely time-sensitive, with high labor intensity and rising labor costs for manual picking. The harvesting process faces a dilemma of "no machines available." "The daylily harvest period is only about 40 days, and picking usually takes place between 2 a.m. and 8 a.m. We can only pick the buds that are about to open. Once the sun shines on them, they open, and they can't be harvested, or the price drops significantly," said a farmer surnamed Wang. To ensure the moisture needed for daylily growth, the fields are often flooded during the peak flowering season, leading to high harvesting intensity and poor working conditions. This results in difficulties in recruiting workers and high picking costs, and when there is a shortage of labor, many daylilies are left unharvested.
To address these industry pain points, the project "Key Technology Research and Application Demonstration of Intelligent Daylily Harvesting Robot" was approved as a key research and development project in Shanxi Province in 2023. Led by Taiyuan University of Science and Technology, in collaboration with Shanxi Nonggu Modern Agricultural Equipment Technology Co., Ltd. and Shanxi Agricultural University, the team spent nearly three years making continuous breakthroughs in key technologies such as intelligent recognition, precise positioning, and non-destructive picking. This culminated in the launch of the first-generation daylily harvesting robot.
The robot features four-dimensional precise visual positioning, equipped with path recognition and autonomous navigation capabilities at its base, as well as 5G communication, allowing farmers to monitor the robot's status in real-time or issue harvesting commands via their smartphones. It is equipped with two parallel-link mechanisms as harvesting arms, enabling dual-arm collaborative work for higher efficiency. The specialized gripper mimics the human side-picking method to ensure non-destructive harvesting. Professor Li Haihong from the School of Mechanical Engineering at Taiyuan University of Science and Technology stated that the combination of four-dimensional vision, dual-arm collaboration, and specialized grippers allows the robot to "precisely operate in low-light conditions."
During the field demonstration, the robot showcased several core technologies. The four-dimensional precise visual positioning system is its biggest highlight. To address the challenges of low-light, high-humidity, and complex field environments, the robot is equipped with two vision modules that output X, Y, and Z spatial coordinates along with the cutting angle—a four-dimensional picking point—to guide the robotic arms for precise operation and damage-free harvesting. The robot can operate with dual-arm collaboration. It is fitted with two parallel-link robotic arms, and an industrial computer uses algorithms to allocate picking tasks between the arms, planning paths and calculating postures based on the four-dimensional positioning information. Currently, the picking speed is about 5 seconds per flower, with a target of no more than 2 seconds per flower in the future.
In terms of power and endurance, the robot uses a tracked, gantry-type electric chassis, meeting the mobility requirements of daylily fields. A solar panel is installed on the roof to charge the lithium battery during the day when not harvesting, eliminating the need to return to a charging station. LED light strips provide illumination for nighttime operations. Compared to traditional manual picking, this robot offers clear advantages: unified picking standards that do not rely on human experience; truly unmanned operation with remote control; and the ability to work continuously during the day, avoiding missed harvest windows.
From "no machines available" to "actual machines in the field," the intelligent daylily harvesting robot has taken its first step. "But as a first-generation model, this robot still has room for improvement. It cannot operate in rain or strong wind, and the visual recognition algorithms, trajectory planning algorithms, and collaborative control algorithms need further upgrades," Li Haihong admitted. "The first-generation model is like a train that has just left the station; subsequent acceleration and lane changes will have to rely on data from actual use." Through field demonstrations and technical exchanges, the team has collected first-hand feedback data for continuous iterative upgrades.
Comments