Jiangsu: How Does "Ai+Agricultural Machinery" Create a "New Engine" for Future Agriculture?

"Many young people don't know how many acres of land could one person and one cow plow in the countryside in the past?" Luo Xiwen, an academician of the Chinese Academy of Engineering, raised two fingers with a little excitement, "Two acres.

 

 

"Many young people don't know how many acres of land could one person and one cow plow in the countryside in the past?" Luo Xiwen, an academician of the Chinese Academy of Engineering, raised two fingers with a little excitement, "Two acres. Now, if our farmers can use unmanned dry rotary tillers, even if they only let the machines work for 10 hours a day, they can cultivate 200 acres of land by themselves."

 

This was a scene at the 2025 Leisi Conference held at Jiangsu University on April 28. Also at this industry summit forum, Liu Jizhan, vice dean of the School of Agricultural Engineering of Jiangsu University, asked a "soul-searching question": "Today, many farmers are considered 'young people' at the age of 60, and most are still engaged in agricultural production over the age of 70. If they still cultivate in the traditional way, how can they meet the development needs?"

 

Leisi, a general term for agricultural tools in ancient China, became synonymous with agricultural machinery equipment at the conference. "We held the Leisi Conference and had an in-depth discussion on 'AI empowering new agricultural productivity', which is an important measure to implement the deployment requirements of a strong agricultural country and make up for the shortcomings of the agricultural machinery equipment industry." Li Hongbo, Secretary of the Party Committee of Jiangsu University, said to more than 300 guests including academicians, experts and heads of 36 companies from 86 universities and colleges.

 

The Urgency Of Reality Has Given Rise To The Transformation Of Agricultural Machinery

 

The update of agricultural machinery equipment calls for the integration of AI elements, which has become an industry consensus. The urgency of reality behind it is self-evident.

 

"Take a simple example, cotton topping." Liu Jizhan introduced that there is a "topping" process in cotton planting, which is to cut off the buds on the top of the cotton to promote the development of side branches and increase yields. But the pruning process is not easy. Until now, it is still mainly manpower. Even if traditional machinery can be used to assist, the labor intensity is still very high.

 

Zhao Chunjiang, an academician of the Chinese Academy of Engineering, also noted that if agricultural machinery is not developed, it is likely to restrict agricultural progress. "With the transfer of rural population, the reduction of labor force and the aging problem, the importance of agricultural robots has become increasingly prominent."

 

"I wonder how much it costs to hire an experienced agricultural machinery operator in Jiangsu?" Luo Xiwen asked on the spot. The old expert paused for a few seconds and slowed down his speech. He said that in Guangdong, a skilled driver earns 500 yuan a day during the busy farming season. "Think about it, how much does an ordinary farmer earn from planting one acre of land? Asking them to spend 500 yuan to hire someone to do a day's work is really unaffordable."

 

With rising labor costs and reduced manpower, agricultural machinery equipment has been pushed to the forefront of the times, and accelerated iteration and upgrading has become a must for industrial development.

 

In fact, the country is also accelerating deployment. On March 28, the State Council's executive meeting pointed out that the development of agricultural machinery equipment should adhere to agricultural needs and focus on key areas, and comprehensively adopt measures from breaking through basic technologies, improving innovation systems, cultivating high-quality enterprises, and building industrial clusters to accelerate the development of iconic products and key core technologies.

 

In This Case, How To Make Agricultural Machinery "Smart"?

 

"We have been talking about 'smart agriculture', but we have not found a clear concept after searching for a long time, so I summarized a paragraph." Luo Xiwen explained, "Smart agriculture is a new agricultural production and management method that integrates digital perception, intelligent decision-making, precision operation and intelligent management formed by the deep integration of advanced technologies such as biotechnology, information technology, artificial intelligence, big data and intelligent equipment."

 

In layman's terms, it is to make traditional agricultural machinery from a pure machine to "like a person", and even replace humans to identify and judge basic information.

 

Under such expectations, there has been a concrete transformation in the scientific research and production demonstration ends-

 

In the early morning of the farm, a drone equipped with a multispectral camera flew over the wheat field, and the AI ​​system immediately popped up an early warning message: Beware of early infection of stripe rust.

 

In the field at noon, an unmanned grain truck followed closely behind an unmanned harvester. The two agricultural machines traveled at the same speed with a position error of less than 5 cm, so that the rice was loaded into the truck bed of the grain truck as soon as it was harvested. When sensing that the cargo box is full, the "smart" grain truck can also automatically navigate to the designated location to "unload" and then return to the harvester to work.

 

The same is true in the field of traditional plant protection. Chen Jianping, an academician of the Chinese Academy of Engineering, pointed to the screen at the conference venue: "Artificial intelligence has begun to reshape the domestic plant protection system, and has fissioned into a wide range of application scenarios in the fields of pest monitoring and migration warning, innovative pesticide development, green control technology, and intelligent plant protection equipment."

 

"Although there are urgent practical demands, it is exciting that agricultural robots are gradually transitioning from scientific research demonstration to autonomous agricultural operations." Zhao Chunjiang said.

 

Although Technology Has Improved, The Pressure Is Not Small

 

In the greenhouse in May, a fully automatic unmanned operating robot arm is picking elevated strawberries. If you switch to the "perspective" of this robot arm, it is not difficult to find that each strawberry in the "field of view" has an independent tracking imaging area. Once focused, the robot arm will retrieve the image in the chip database for comparison. As long as the fruit is determined to be ripe, the robot claw will follow up and execute the fruit picking instruction.

 

"With the support of AI, agricultural machinery should be more flexible." Liu Jizhan said frankly that this change is not only to "transplant" the voice interaction and data reading of artificial intelligence to traditional agricultural machinery, but also to give agricultural machinery a "new look" in an all-round way.

 

To this end, Liu Jizhan's team set their sights on kiwifruit picking.

 

"China is the country with the largest kiwifruit planting area in the world, with a total planting area of ​​more than 3 million mu. If kiwifruit is not picked in time after it matures, it will lead to a decline in fruit quality, a reduction in the market window period and damage to the health of the tree, which will in turn affect the entire industrial chain." Team members told reporters that according to preliminary statistics, the annual loss of kiwifruit due to untimely picking exceeds 2 billion yuan.

 

It is based on this background that the team started the project in 2022, built the first picking prototype in June 2023 and tested it in August of that year. In June last year, the second-generation machine research and development began, and finally achieved a leap from single-arm independent picking "1,000 grams per hour" to multi-arm joint picking "3,200 grams per hour", and achieved night picking and autonomous walking with one key operation.

 

"After integrating and developing relevant AI technologies, we have ushered in a milestone improvement. But it should also be noted that according to the domestic planting area, about 100,000 such robots are needed to meet the picking needs." Liu Jizhan said.

 

Although the technology has improved, the challenges are still huge. Not only is mass production and popularization difficult, but some key core technologies and data interconnection also face the need to break through.

 

"In a complex environment, the machine's perception is not accurate, the robot arm's 'hand-eye-foot-brain' coordination is difficult, and the end effector is not flexible. These are all difficult problems that need to be solved now." Zhao Chunjiang said, for example, how to distinguish the pressure resistance characteristics of different fruits, how much force should be used when grabbing; how to track dynamic fruits when the wind blows; for strawberries with tangled stems, how to accurately select ripe ones without damaging the developing vines... "There are many difficulties. Just for the 'tactile feedback' during picking, it is possible to conduct hundreds of experiments."

 

Gong Liang, a professor at the School of Mechanical and Power Engineering of Shanghai Jiao Tong University, feels the same way. He led the team to develop an agricultural humanoid robot from 2018 to 2022. Although it can cope with most situations during inspections and fruit picking, once it encounters three to four large-fruit tomatoes that grow very compactly, the robot's picking order and movements will be out of shape.

 

What is the ideal technical state? Gong Liang painted a picture: if agricultural robots have arms and can push away obstructions such as leaves and then pick the fruits that were originally covered, the success rate of picking can be doubled from 40% when relying on robotic arms.

 

"But it's not easy." Gong Liang smiled. In his opinion, general robotic arms will enter the market "slightly faster". As for humanoid robots, it is expected to take another 4-5 years to enter the demonstration application stage, and it will take longer to enter ordinary farmers' homes.

 

"When paying attention to these, don't ignore the problems of agricultural machinery data." Chen Jianping reminded that the current data compatibility and interoperability of various agricultural machinery equipment is not enough, and the phenomenon of inconsistent data formats and standards is often present, which makes these agricultural machinery unable to access the same Internet of Things, affecting agricultural production.

 

"One Machine For Multiple Uses" Must Be The Future Direction

 

Although the road ahead is long, the transformation of agricultural machinery equipment has made a good start.

 

Li Jun, dean of the School of Engineering at South China Agricultural University, analyzed that thanks to the integration of technologies such as green energy, intelligent collaboration and digital manufacturing, agricultural drones are no longer single-function operating tools, but key factors that trigger the deep reconstruction of the technology innovation chain, agricultural value chain and low-altitude industrial chain.

 

In Li Jun's eyes, agricultural drones will be able to do far more than industrial drones in the future: "Because the operating objects of agricultural machinery equipment are living crops, agricultural drones will be able to "talk" with crops in the field in the future and perceive the growth needs of crops."

 

Letting agricultural machinery talk to crops, this sounds very idealistic, but Li Jun is sure that it is the direction of the future.

 

"Understand crop needs through multi-source data and then prescribe specific prescriptions." He explained quickly, for example, by collecting images of crop leaf growth through drone lenses, and comparing and analyzing with background historical data and various crop trait pictures, it can be known whether the crops are short of water and fertilizer, and what other demands are there for pest control.

 

"With the breakthrough of AI technology, current agricultural machinery equipment can carry out inspection, fertilization, weeding, spraying, picking, pruning, thinning flowers, transplanting and other tasks, but these are still mainly operated by special robots. Looking ahead to the next 20 years, general-purpose robots with 'one machine for multiple uses' will definitely become an important iteration direction for agricultural machinery equipment." Zhao Chunjiang predicted.

 

Gong Liang also agrees. "In agricultural farming scenarios, whether the width of the lanes, the height of the crops, or the picking operation surface, they are all designed according to the human physique. Therefore, once a humanoid robot with the same physique enters the relevant scene, it can quickly replace people and switch seamlessly." Gong Liang said.

 

In 2022, when the humanoid robot developed by Gong Liang's team was "experimentally" deployed to an agricultural company in Shandong, he was delighted to find that only one humanoid robot could complete the picking volume of three to five people. "With the advancement of technology, I believe that efficiency will be improved. Who will farm in the future? Isn't it a solution?"

 

At the end of the forum, an expert showed an imaginary picture of future agriculture generated by ChatGPT on the screen of the venue. In the picture, facing the direction of the rising sun, humanoid robots are inspecting, protecting seedlings, and spraying pesticides on the vast farm, while unmanned agricultural machinery is accompanying them on both sides, responsible for weeding and fertilizing. The green crops in the field contain infinite vitality.