Have the robots arrived?
Tom Allen-Stevens
With a number of autonomous machines now appearing on farm, 4AR seeks the view of Prof Simon Blackmore on whether they live up to the vision of a new mechanisation system for farming.
In January 2014, on the second day of the Oxford Farming Conference, Prof Simon Blackmore took the stage in its final session on Technology and Innovation to deliver his vision of the future.
He suggested a completely new mechanisation system for crop production where you remove all constraints of the current machines. “We’re familiar with precision farming that takes you down to field and sub-field level. How far down can we go? We can come right the way down to individual plant care,” he suggested.
Simon described a future where a small number of big tractors would be replaced by many small autonomous robots, questioning the sense in applying large machines to the soil.
The individual plant care would be guided by scouting robots on the ground or drones in the air. He talked of non-contact solid-state sensors he believed would come into agriculture to support the robots. 99.9% of herbicides could be removed by micro-dot spraying, or even 100% by using lasers, he pointed out.
The robots would feed accurate information through about the shelf life of produce, giving farmers the ability to add value to the product before it leaves the farm gate. Selective harvesting would reduce waste. “Up to 60% of harvested crop is not of saleable quality. So why do we harvest it?” Simon asked.
Finally, a management information system would be needed to analyse the “big data” generated, make a complex system simple to use and through traceable provenance, allow the farmer to take a fairer share of the value of their produce, he said.
Prof Simon Blackmore took the stage at the 2014 Oxford Farming Conference to deliver his vision of the future.
Perhaps the most interesting aspect of the talk is that it was presented as a vision for 2050. “Is this the future for 2050, or is this now? All of these concepts have been developed and researched, but not many of them are commercially available because farmers aren’t asking for these types of machines or systems,” Simon noted.
“How long will it take for farmers to take advantage of these opportunities?”
Not quite as long as 36 years, it seems. The UK’s Small Robot Company for example launched the latest version of its Tom scouting robot recently, with a limited commercial service due to get underway this autumn, ready to ramp up to more than 100 farms in 2023.
Trialled on the Lockerley Estate in Hants, Tom was delivered to commercial specification by Tharus, a British company that also makes robots for Ocado. He’ll cover 20ha/day autonomously, collecting about 6 terabytes of data in an 8-hour shift – at Lockerley, he identified 12.7 million plants in a 6ha field, of which 250,000 were weeds.
The information is fed back to Wilma, the AI brains of the operation, at a resolution claimed to be sub-millimetre, and weeds, including blackgrass, can now be identified from wheat.
Tom (right) is ready for limited commercial roll-out while protype Dick has been trialled on the Lockerley Estate.
Wilma devises a weed-treatment plan that’s put into action by Dick, a second robot that’s still at the prototype stage. Dick zaps weeds with electricity, using technology developed by Rootwave. Probes mounted on three delta arms, created by motion plastics company igus, home in on the weeds and deliver an 8kV electric charge, effectively boiling it so it dies. Running off a Tesla battery, small weeds need only about 10W for 0.5s.
But Tom won’t just be scanning weeds. Also being piloted is SlugBot, with partners Crop Health and Protection (CHAP), in which Tom uses hyper-spectral cameras to detect slugs at night, and then treat them biologically with microdoses of nematodes.
And he’s also acquired a nose – working in a consortium with PES Technologies, NIAB-EMR, Hutchinsons and the Universities of Essex and Greenwich, the proof of concept has been achieved for a device that ‘smells’ the soil.
Tom takes 400g samples from across a field and analyses them on the fly for Volatile Organic Compounds (VOCs) – microbial particles effluxed by various soil types and cropping systems. The consortium is applying machine-learning to the analyses to profile soil health, which could provide accurate, repeatable carbon measurement at farm scale.
The 6t autonomous AgBot has diesel-electric drive, wheels or crawler tracks, lugging up to 156hp through a standard three-point linkage.
Commercial deployment
Another development is the AgBot, from Dutch start-up AgXeed. In May, Claas announced it had acquired a minority shareholding in the company to help commercialise it – expected to be available in 2022.
The 6t autonomous AgBot has diesel-electric drive, wheels or crawler tracks, lugging up to 156hp through a standard three-point linkage. It’s claimed to deliver advanced digital path planning, autonomous task performing and automatic data collection.
With a lift capacity of up to 8t, there’s an electric PTO, independent of the engine speed, and external high-voltage connections are also optionally available. in addition to the RTK steering system, the electric equipment includes all the technology required for hazard and obstacle detection.
The AgBot was on show at the UK’s Cereals Event in June, along with AgroIntelli’s Robotti. The 150D is one of 50 robots from the Danish manufacturer and has been put to work on Home Farm, Nacton in Suffolk, making it the first robotic tractor to be fully operational on a farm in the UK, says AgroIntelli.
The Robotti has two Kubota 75hp diesel or biofuel engines – the left engine propels the machine, while the right engine drives the PTO.
The model at Home Farm has two Kubota 75hp diesel or biofuel engines. The left engine propels the machine in 4WD and powers the conventional three-point hitch. The right engine drives the PTO. The machine can run continuously for around 24 hours between fill-ups.
The 1940ha farm includes organic vegetables, so the Robotti is used for mechanical weeding, although there are plans to use it for topping, drilling and even transplanting. It takes around 10mins to map a 1.6ha field and a further 10mins to log the weeding plan into the system. Field obstacles are logged at the programming stage, and any other obstacles outside of this will make Robotti stop in its tracks for safety. With front and back cameras, these, along with the controls and progress, can be accessed through an online portal.
There are a number of other robots now deployed on farms across Europe. Some, like the Sabanto are little more than tractors without drivers, performing conventional tasks autonomously. The e-Tract from French manufacturer Elatec weighs in at under 800kg with lithium batteries capable of working implements on the rear hitch of up to 300kg. The centrally slung mechanical weeder can intelligently guide itself round individual plants, while there’s also a front hitch.
Australian-based SwarmFarm claims its robots have weeded, sprayed or mowed over 80,000ha over the past two years. The concept behind the diesel-fuelled autonomous machines is they’re simple to use with a ‘swapnostic’ system so farmers can fix them themselves.
Dino is a gantry-type machine for weeding large-scale vegetable crops with 2cm accuracy, and can cover 4ha/day
Based in France, Naïo Technologies has three robots operating across Europe: Dino is a gantry-type machine for weeding large-scale vegetable crops with 2cm accuracy. Weighing in at 1.25t, it can cover 4ha/day, powered by batteries. Its smaller brother, Oz, is for smaller growers, weighs 150kg and can follow you around, carrying tools or plants, or follow a predefined route to weed or sow. Ted is the sibling robot for vineyards.
PixelFarming, based in the Netherlands, has its Robot One, another autonomous gantry with ten ‘arms’ that hang down and have a range of interchangeable weeding tools. It has vision recognition and can work to millimetre accuracy, says the company. Five solar panels are supported by 10x 1KWh battery packs.
FarmDroid’s FD20 from Denmark is another solar-powered bot, claimed to be the world’s first autonomous seeding and weeding machine, deployed across Europe in 2019. Because it knows where every seed is planted, the FD20 doesn’t depend on vision for crop or weed recognition, says FarmDroid.
The company was founded by brothers Jens and Kristian Warming along with robot expert Esben Østergaard who invented the world’s first collaborative robot. It can be left alone in the field to work throughout the season within a predefined geofence, and when the sun goes down it just stops until sunrise. It has a working width of 3m with 6-8 rows, weighs 900kg with its batteries and can manage a 20ha field of sugar beet, onions, kale, spinach, rapeseed or certain herbs.
The FD20 is claimed to be the world’s first autonomous seeding and weeding machine, and doesn’t depend on vision.
On the harvesting side, the E-Series from Spanish makers Agrobot is claimed to be the first pre-commercial robotic harvester. 24 independent robotic arms mounted on a gantry pick through the crop, gently plucking the strawberries without even touching the fruit. High quality image recognition and real time AI spot the perfect time to pick.
Shaping up
So how do these commercial (or near-commercial) robots shape up to the concept presented at OFC seven years ago? 4AR caught up with Simon to find out.
Simon Blackmore believes precision farming can come right the way down to individual plant care.
“Size is important,” he says, and that counts out many of the autonomous systems that are simply driverless machines. “Big tractors may be efficient, but they’re designed for traction.
“Every KN of draught force in a horizontal direction applies a KN down, and that pressure is killing the soil. So if we take advantage of the opportunities technology offers, we should get rid of the problems of the old system.”
Simon estimates that 90% of the energy used in cultivating the soil goes towards repairing the damage done by cultivations. “The best thing we can do for the soil is leave it alone, and that’s the opportunity small autonomous robots offer,” he says.
“But if the only tool you reach for is a hammer, everything looks like a nail. Many of those designing autonomous machines haven’t been thinking out of the box.”
The Small Robot Company concept was largely based on Simon’s approach, however. “One of the aspects that makes SRC different is the engagement with farmers. They’ve been involved from the outset with turning the concept into a commercially viable system and have guided the development of every stage.”
But while Simon supports the small form factor, he’s not sure everything is right. “The robot must be small enough to treat each plant individually, but large enough to manage the work rate. And it can be as light as you like, but you need stability.
“I think Tom is too small. Dick’s heading in the right direction, but lasers are still my preferred option for weed control – achieving the work rate is the challenge,” he says.
His favoured machine is the mini gantry as this has the stability and, he believes, the potential to achieve the required work rate. “But when any new technology is introduced it takes a while for the ideas to settle out.”
Selective harvesting is beginning to come through in high value crops, and Simon believes there’s potential in cereal crops, too. “Research has shown for many years just how much variability there is in grain quality within the field. If you knew more accurately before you harvest how the crop varied you could harvest it selectively.
“But again it’s a case of rethinking the design – perhaps a stripper header is the right way forward, in which case you don’t need to take the whole threshing machine into the field.”
One aspect where he believes farmers haven’t begun to make headway, though, is intelligently targeted inputs. “At the moment the farmer treats every acre because they have to. A system where sensors deliver information about every plant not only has the potential to cut inputs, but helps you focus far more on aspects of crop health, rekindling basic crop husbandry skills that have perhaps been lost.”
So Simon doesn’t believe the robots have quite arrived just yet. “It’s too early to sell your tractor. But keep a watching brief and keep an open mind. And while we’re looking towards new technology, we shouldn’t dismiss the techniques of previous generations.”
