Autonomous machines have the capability to work without human intervention. Sensors and control systems allow the operation to be automated. Autonomous robots are commonly used in manufacturing industry; laser cutters and robotic welders, for example, work autonomously to produce precise and reliable work, at a rate that human operators could not match. But this is in a very controlled environment where robots repeat the same process again and again. A machine working in irregular, uneven and sometimes soft, fields is an entirely different situation, where each field and operation is different and there is risk to people. Developing control systems to work safely in this situation is particularly challenging.

The ETAROB from Aachen university is a bed-straddling unit that incorporates advanced guidance/safety technology which adds thermal imaging with artificial intelligence to the array of sensors guiding the unit.

The route to autonomy

The pursuit of the driverless tractor is nothing new. Reading university in the UK demonstrated a driverless tractor carrying out field work in the 1950s. Its guidance system required a guide wire to be buried in the ground, limiting the potential application. But the potential to eliminate labour costs was clearly identified, along with the challenges of having usable and safe systems. With the introduction of GPS, guidance systems have become available and many machines can now operate with little or no steering input – a step towards autonomy, but quite distinct from it. Remote-controlled drive units for mowing steep banks have been produced by companies like Bomford, and have removed the operator from the seat, but they are not autonomous. For a few years now, there has been a move to developing fully autonomous field machines, aided by similar developments in cars.

This radical John Deere concept aims at integrating an autonomous tractor unit with an implement eliminating the tractor front axle and making a very compact unit considering it packs 500kW of power. As current battery technology would not provide sufficient capacity, it’s powered by a cable for testing.

Will driverless power units be smaller?

The increase in size and weight of machines is mainly driven by a desire to reduce labour costs. But if we make tractors and power units autonomous, then the labour advantage of scale is eliminated. It may be possible to move to smaller units, which are lighter with consequent benefits to soil structure. Work-rate would be maintained by working long hours and using multiple power units.

But this assumes that the autonomous component would be inexpensive and there would still be road transport challenges to be overcome.

This concept add-on autonomous control module effectively converts an autosteer tractor to a full autonomous tractor.

What are the driverless options?

Autonomy offers a range of possibilities and it is worth considering a number of categories.

  • Standard tractors with an autonomous capability added.
  • Large tractors built without a cab and operator station.
  • Large tractors in a different configuration using the flexibility of electric drives.
  • Small tractors/units usually weighing less than 1.5t capable of straddling 1.5m beds or working in conventional fields.
  • Micro machines weighing less than 100kg designed to work in small areas or as a ‘swarm’ of units in a field.
  • Flying drones to field scout or to provide specific spot treatment of inputs.
  • Examples of the above were shown at Agritechnica, either as working units or technology display units.

    The Case autonomous tractor is a full scale large tractor designed for sole autonomous control.

    Standard tractor with autonomous controls

    John Deere showed an autonomous control module attached to the front of a ‘mock-up’ 5125R tractor with existing RTK auto-steer capability. The module consisted of radar and lidar sensors to allow the tractor to ‘see’ obstacles, including hedges or field features, but also humans and animals to avoid collisions. This technology would allow a tractor work reliably and safely in a fixed field situation such as a vineyard. The unit is not for sale for liability reasons, as if an accident was to happen without an operator on board, who is liable and how will that liability be insured? This question also applies to autonomous cars. This unit can be driven normally where desired, with full control and a standard cab/operator unit and consequently can be used for all tasks and moved from field to field.

    Autonomous cab-less tractor

    How can you save money when building an autonomous tractor? Building it without a cab and operator station can save weight and cost. This is the approach taken with the futuristic design of the Case autonomous cab-less tractor concept, which appeared in 2016 based on a 200 – 400hp chassis and shown again at Agritechnica this year. It uses lidar, radar and standard cameras with GPS to safely guide the machine. This large scale approach may be feasible if the tractor could carry out all its tasks autonomously and if there was little need to be moved on public roads. This creates particular challenges, but it is conceivable that it could be used on large expanses of open crop land, provided the autonomous control is deemed safe. The Irish-designed Itarra concept falls somewhere near this category, as it doesn’t have a conventional hitch. While many manufacturers have concepts like this to gauge public reaction and test autonomous components, a real market remains to be seen.

    Integrated power unit and implements

    Alongside driverless technology, the electrification of drive systems could bring radical alterations in tractor design. ZF and others at Agritechnica showed transmissions with combined mechanical/electrical elements that allow power to be transmitted to traction wheels on implements.

    Taking this design a little further, John Deere showed an extremely compact driverless tracked 500kW tractor unit without a front axle and with the ability to power implement wheels if necessary. While this only had an electrical motor and no battery storage evident, it still showed potential. In its current form, this unit would need a permanent cable connection to supply power via a reel system (a prototype of which has been developed).

    Full battery operation of a 500kW tractor for fieldwork is not economically feasible today and the incorporation of a conventional engine would add significantly to its size. This improved integration of power unit and implement certainly has scope.

    Small autonomous tool carriers for horticultural use or field use

    Autonomous field vehicles are working in high-value horticultural crops, either in glasshouses/tunnels or in bed systems in the field. They can carry out slow repetitive tasks over long periods - typically weeding operations. The French company Naio has two battery powered models available. The smaller Oz is a 110kg unit designed to weed between rows. Guidance is by camera and laser. There are more than 100 such units working in nurseries and protected horticulture systems today.

    The much larger Naio Dino is an 800kg unit that straddles 1.5m to 2.0m beds, also used primarily for weeding. Its guidance system relies on RTK GPS and cameras.

    There were many of these models on show at Agritechnica, mostly used as test-beds for advanced machine guidance, safety control and implement control systems. Weed control by hoeing or spot spraying and disease/insect control by spot or variable-rate spraying are the main focus, but other uses are possible.

    It’s interesting to see that these technologies and their application were foreseen by the now closed Silsoe research engineering institute in the UK, whose development work in the 1980s and 1990s was not unlike what we see today. That work spawned the Garford ‘robocrop’ guided weeding system, which remains one of the more impressive applications of row-guided technology today.

    The swarm concept

    The most radical embodiment of driverless agricultural machines is the ‘swarm’ concept, where large scale field work is carried out by a multitude of very small units. This is the brainchild of the university-developed ‘Mars’ project, purchased by Agco two years ago and rebranded as the Fendt Xaver. But while the small 50kg unit is inherently safer than larger units and minimises the impact on the soil, is this size practical for field operations? Its original function was as a precision seeder, but a lot of work is needed to see whether this size of unit can play a significant role in field scale agriculture.

    The drone

    The flying drone, although prominent at Agritechnica, should not be forgotten. It can scout fields autonomously (though with significant legal restrictions), potentially using many sensors to indicate crop or field condition. In some cases, they may treat specific areas with targeted inputs

    While autonomous agricultural machines were developed originally in the 1950s, their time is finally arriving. But the shape of this development and its future influence on farming systems is still unclear.