Fifty-five agricultural robot scenarios, now published
China's industry and agriculture ministries published 55 typical agricultural robot scenarios covering tillage, planting, field management, harvesting and livestock, led mostly by universities, institutes and local firms.
China's industry ministry and agriculture ministry jointly published a list: 55 typical application scenarios for agricultural robots. The document was issued on November 9, and local governments have been republishing the list over the past two days.
First, how it was produced. The selection followed a full process: company applications, recommendations, expert review, public comment, then the two ministries finalising the list. For a company, inclusion puts it in an official scenario library, from which demonstration projects, subsidies and procurement will draw first.
What kind of work is on the list
According to the published table, scenarios are split into crop farming and livestock, then into tillage, planting, field management, harvesting, whole-cycle work and feeding. Examples include high-throughput soil composition testing, automated transplanting and seedling-bed transfer, multimodal crop monitoring with water-fertiliser-pesticide application, laser weeding, greenhouse tomato picking, premium tea picking, coordinated ground-air fruit and vegetable operations, teat-dipping for dairy cows, rail-mounted pig-farm health inspection, and coordinated smart grain-silo levelling.
The notable thing is not how advanced the technology is but how finely the scenarios are cut. From soil testing to cow dipping, each has a defined object, environment and acceptance criterion — exactly what industrial buyers need before purchasing a robot.
Why agriculture suits early deployment
Farming has several features friendly to robots. First, tasks are highly repetitive and structured: weeding is weeding, picking is picking, without the endless surprises of a home. Second, labour is scarce and costly: rural workers keep leaving and seasonal wages rise every year. Third, the payer is clear: large farms, cooperatives and agricultural service companies calculate return on investment, not concepts.
Fourth is policy. Agricultural robots qualify for machinery subsidies, demonstration projects and open scenarios — scarce in the humanoid race. A list of 55 scenarios is essentially an official demand survey and endorsement.
How it relates to humanoids
Most robots on the list are not humanoid. Some are tracked chassis with arms, some rail-mounted inspection devices, some drones — form follows the task, not the aspiration to look human.
That echoes a judgment repeated across the industry this past month: form must serve the scenario. Last week one company chose not to build humanoid (wheeled base with two arms), another put a robot into a semiconductor fab for inspection; this list applies the same logic to fields.
For humanoid companies the list matters differently: it shows how government organises demand — select scenarios, publish a list, then attach demonstrations and procurement. For humanoids that process has not finished (local action plans with output targets dominate), while agriculture has reached the list stage.
An underrated point: agricultural robot data is structured. A field, a row of seedlings, a spacing between trees are all standardised, which makes AI training and generalisation far easier than in homes — the same model often works across farms.
That mirrors the manufacturing line logic: fix the environment, then optimise the motion. In farming the environment is fixed not by remaking it but by the regularity of the crops themselves.
Where to discount
First, a "typical application scenario" is not an order. The list is selection and publication; whether procurement follows depends on demonstration projects and subsidies.
Second, unit prices and margins are far below humanoids. A picking robot costs a few hundred thousand yuan, a weeding machine over a hundred thousand — the market grows through volume, not premium.
Third, the lead units are mostly universities, institutes and local firms, with few national farm-machinery giants. That suggests validation, not standardised mass production.
What is confirmed is the path: scaled deployment of robots is likely to start not with the most human-like form but with the most standardised work. Agriculture, logistics and power inspection share the same features — clear tasks, explicit acceptance, labour shortage and willingness to pay.
This piece summarizes public reporting and did not verify all 55 scenarios. Not investment advice.