Robotics
Agricultural robotics and the labour that is disappearing
Fields are unstructured, produce is delicate, and the economics are being forced by a labour shortage rather than pulled by capability.
Agriculture automated its bulk operations a century ago. What remains manual is the work requiring judgement and dexterity, and that is where robotics is now being pushed — largely by labour scarcity rather than by technology being ready.
Why fields are hard
Everything that makes a factory suitable for robots is absent.
Terrain is uneven. Lighting varies from overcast to direct sun to shadow within a row. Plants vary in size, shape and position. Targets are partially occluded by leaves. The environment contains dust, mud, rain and vibration. And the whole scene changes as the crop grows.
A machine that works in June may not work in August on the same field.
Where automation has actually arrived
Autosteer and precision guidance. Satellite-guided tractors following centimetre-accurate paths are now standard on large arable farms.
This is genuinely widespread and delivers real savings by eliminating overlap on every pass.
Variable rate application, adjusting seed, fertiliser and chemical rates across a field according to a prescription map.
Automated milking. One of the clearest successes.
Cows enter voluntarily, are identified, teat position is located by vision or laser, cups are attached robotically, and yield and health data are recorded per animal.
It works because the environment is structured, the task is repetitive, and the animal cooperates for feed.
Robotic weeding. The fastest-growing category and a genuinely good fit.
Machine vision distinguishes crop from weed and removes the weed mechanically, with a targeted laser, or with a precisely aimed micro-dose of herbicide.
The driver is chemical reduction — regulatory pressure, resistance, and organic markets — as much as labour. Targeted application can reduce herbicide volume by a very large fraction.
Harvesting: the hard case
Where most attention and money is going, and where success is uneven.
Bulk crops — grains, sugar beet, potatoes — were mechanised long ago, because the product is robust and can be handled roughly.
Fresh produce is the problem. Apples, strawberries, tomatoes, lettuce and asparagus are bruise-sensitive, must be assessed for ripeness individually, and are surrounded by foliage.
The requirements compound: detect the fruit through leaves, judge ripeness, plan a path to it, grip without bruising, detach without damaging the plant, and place it gently — repeatedly, fast enough to beat a human picker's rate.
Human pickers are quick. Matching their speed at acceptable damage rates is the bar, and most systems have not cleared it economically.
Strawberries have received the most effort because of high value and high labour intensity. Several systems are in commercial trials with harvest rates approaching viability for some varieties.
The economics are unusual
Seasonality. A harvesting robot works for weeks per year. Amortising capital cost over a short window is brutal, which pushes toward robots-as-a-service and toward machines that handle multiple crops.
Labour supply is the driver. Fresh produce harvesting depends heavily on seasonal migrant labour, and availability has fallen in several countries for policy reasons.
Growers have reported crops left unharvested. That is a stronger commercial pull than any efficiency argument.
Margins are thin, which limits what growers can pay and makes payback periods long.
Breeding for machines
An underappreciated part of the answer.
Rather than building a robot that can handle any plant, breed plants and design growing systems that suit machines: uniform ripening, fruit presented outside the canopy, trellis systems that expose the crop.
This is exactly the pattern that made warehouse robotics work — change the environment rather than generalising the machine.
Vertical farming and controlled-environment agriculture take it further, producing a fully structured environment where automation is straightforward. The constraint there is energy cost, which limits it to high-value crops.
The labour question
Worth addressing directly.
Agricultural harvesting labour is physically demanding, seasonal, low-paid and frequently performed by workers with limited legal protection.
Automation displaces jobs that many workers depend on and that few would choose. Both of those are true simultaneously.
The honest position is that the transition's effects fall on a workforce with little bargaining power, and that nothing about the technology determines whether the gains are shared.
