Key points

• A technical review of autonomous machinery was undertaken to inform the Code of Practice for Agricultural Mobile Field Machinery with Autonomous Functions in Australia
• It was supported by GRDC, the Cotton Research and Development Corporation and Sugar Research Australia
• It has found that while certain hurdles remain, autonomous on-farm machinery is fast approaching

Some of the key components that will enable autonomous on-farm operations are already commonplace in many top-of-the-range tractors.

Agricultural engineering expert Professor Craig Baillie from the University of Southern Queensland says many grain growers are familiar with autosteer, variable-rate technology (VRT) and global positioning systems (GPS). “The last hurdle is to bring all these components together. It’s a big step but one we are fast approaching,” Professor Baillie says.

He recently reviewed the technologies, standards and regulations for autonomous machinery to help develop a code of practice for the industry. “The purpose was to give some clarity to the obligations behind autonomous machinery for the manufacturers, sellers and users.”

The technical review was supported by GRDC, the Cotton Research and Development Corporation and Sugar Research Australia.

Developed with input from grain growers and with manufacturer support and technical input, the code of practice was a joint project between Grain Producers Australia (GPA), the Society of Precision Agriculture Australia and the Tractor and Machinery Association of Australia. Its development was led by farm technology expert Dr Rohan Rainbow.

Review

Professor Baillie says the technical review behind the code considered industry maturity and carried out a “stocktake” on Australia.

“We concentrated on the technology that tractor manufacturers are using because they offer a clear path to market at broadscale adoption.”

Professor Baillie says full automation is limited by machine perception, the ability of a machine to sense its surrounds, its operation within those surrounds and the ability to monitor and adjust performance without operator intervention.

“The machine needs to know its surrounds and what job it is doing. If it is spraying, for instance, it needs to know that if a nozzle is blocked it should stop until that is fixed.

“Slowly we are starting to see machine perception emerge as functional components on machinery, and spray equipment is a good example of this. Not only can equipment now sense a nozzle is blocked, but it can also perceive weeds from crops and soil for targeted spraying.”

The review also identified that to facilitate autonomous equipment, a machine’s operating requirements need to be defined for different uses and conditions. Called the Operational Design Domain (ODD), it defines machinery and technology requirements relative to different applications. This means machinery is developed fit for purpose and there is clarity on what it can and cannot do.

Although farming can seem to operate in an unstructured environment, which is difficult for autonomy, this can be narrowed, Professor Baillie says. “This can make it structured and the task far simpler.

“There are operating parameters that can be defined to provide particular structure such as defined pathways (tramlines), field-based activities and pre-programmed tasks.

“A structured ODD considerably reduces the expectations and requirements of the technology and provides an earlier entry point to autonomy on-farm.

“We have become used to tractors being able to do everything, but autonomous machinery may be reserved for specific tasks. They may be used for tasks that are relatively simple and laborious or where multiple machines can be supervised by a single operator.”

As part of the review, Professor Baillie also looked at how autonomous machinery is being used in the transport, defence and mining industries. Transport is a good sector for the agriculture industry to follow, he says, because of how it is has approached ODD refinement.

“The level of automation required to operate road vehicles has been simplified by refining the ODD. While the vehicle must be attentive to recognising potential threats – using perception – the operating environment as a whole is expected to remain fixed and predictable. This allows the technology requirements to follow a fixed path and focus on obstacle avoidance.”

The transport industry’s experience can also help in terms of data and data sharing. “If data is shared between vehicles, this could increase cyber security issues – something we need to be aware of.”

One thing is certain, he says – current technology limitations will diminish with rapid developments in sensors, machine perception, machine learning, artificial intelligence, big data and connectivity.

“Apart from machine perception, a suite of other enabling technologies that are commercially available provide some clues as to the maturity of autonomous machinery developments and the relatively close proximity to commercial reality.”

This means the timing of the code of practice is important. “It is critical to consider non-technology impediments that provide clarity for manufacturers to proceed with commercial release or to accelerate commercial release.”

The code of practice provides the basis for these considerations. GPA’s Dr Rainbow says the initiative is proactive. It is designed to enable future access to a rapidly emerging technology that will deliver productivity gains for growers and the industry.

The code includes guidance on mobile machinery with semi-autonomous and autonomous functions, and on developing, implementing and evaluating safe work procedures for using this machinery. It is with the WA Government for formal endorsement before being rolled out nationally.

More information: Craig Baillie, 07 4631 2071, craig.baillie@usq.edu.au

Resources

University of Southern Queensland  Review of Technologies, Regulations and Operating Standards for Field Based Autonomous Agricultural Machinery

Grain Producers Australia, Code of Practice