Development and utilisation of autonomous vehicles and artificial intelligence to grow and harvest food is gaining traction across agriculture.

Robots and AI are now advanced enough to be used for tasks such as weeding, crop sensing and fruit picking.

However, farming is complex and when developing any agricultural technology, innovators must think holistically about how it will be used on-farm, who will be involved in its use and who it might impact more broadly.

An Agricultural Technology Hackathon hosted by Agri-EPI Centre sought to identify solutions to enhance the safety and security of autonomous farm machines.

Agri-EPI ran the event with Innovate UK-funded Hands-Free Farm, a testbed for autonomous farm machinery and drones.

The participating teams originated from a range of disciplines, including robotics, AI and machine learning, drones and computer vision.

They came together to address key challenges around safety, connectivity and cyber security.

Kit Franklin, senior engineer, Harper Adams University, and Clive Blacker, head of business development, Map of Agriculture, provided the challenges.

Considerations An industry paper released following the hackathon raises a series of considerations around agricultures readiness for large scale adoption of autonomous vehicles and offers recommendations around maximising safety, improving connectivity and combating future technology threats.

Speaking after the hackathon, Mr Blacker said: The diverse nature of agriculture and robotics operating in off-road and on-road environments poses many challenges.

Our aim with the hackathon was to bring great ideas from any background into agriculture that have the potential to support robotic safety.

We have been delighted and inspired by the solutions put forward.

Agri-EPI engaged with industry stakeholders, including NFU Mutual and Country Land and Business Association, to explore the challenges posed by autonomous farm vehicles to users of the countryside.

The biggest area of potential risk is the threat of collision with people or other vehicles.

Recommendations: Artificial intelligence and machine vision should be used to identify and avoid collision with certain objects, as well as to recognise the difference between static and moving objects.

Static obstacles, such as pylons and trees, could be identified through a pre-route survey of the field.

Machine vision and supervision could additionally be used to identify and distinguish between objects in-field for example a 6ft tall maize plant versus a 6ft tall human.

Building a mechanism to distinguish between a deliberate collision versus an accident into the machine technology is key to ensuring safety.

There is still work to be done in terms of communicating and raising awareness of farming practices to the public and how to safely navigate the countryside.

Discussion of appropriate signage would be necessary to highlight the operation of autonomous agricultural vehicles being used to produce food.

This could include signage indicating ‘do not enter or ‘stay away zones.

There is also the consideration of whether the Countryside Code would need to be reviewed to reference tech and autonomous systems.

There is a need for cross sector learning between the automotive industry and agriculture.

This collaboration would help to maximise preparedness through rigorous testing of autonomous models.

Companies who brought their experience from the automotive sector included Continental Engineering and Epitomical, which have experience in developing and testing autonomous highway vehicles.

The Transport Research Laboratory also attended the workshop and highlighted the need for rigorous testing and training to be included.

Another key area of focus is on improving the understanding of the integration of human operators with robots for increased performance and reliability.

And there was a strong view there should be an agricultural vehicles autonomous code of practice put forward, something already being explored by Harper Adams University and UK national standards body BSI.

The agricultural industry has for a long time been viewed as at low risk from potential cyber attacks.

However, with more farms adopting new technologies cybercrime is becoming an increasingly severe threat to agri-businesses.

The number of attacks is on the rise and important areas to address this include anti-fraud and anti-theft systems.

More machines and devices connecting to the internet widens the threat landscape and increases potential vectors.

Organisations commonly do not know what is connected to their network and cannot spot anomalous or malicious behaviour.

The problem is only becoming exacerbated as Operational Technology (OT), such as manufacturing and production environments, are being attacked, which ultimately compromises the future of food production.

The ability to take over or glean data from these systems may be a national threat to food supply chains.

In addition, there has been a rise in farm thefts relating to high value GPS kit.

Combat

Recommendations: Network connectivity is wide and varied and will continue to evolve.

A way to combat future technology threats is to include more ingress/ egress points and IT/OT networks which need to be secured.

Compromised systems or devices can cause loss of revenue, reputational damage and loss of intellectual property.

Utilising built-in cyber and anti-theft systems as part of the agritech development process will link in as well to the adoption of better technology on-farm as farmers will be able to deploy this technology safely and securely

Challenge Connectivity

Farms are remote and often located in rural areas with poor connectivity.

Traditional ‘over the air technologies, such as satellite, provide variable performance and can often have expensive usage limits.

As such a move toward the Internet of Things means that more devices are using connectivity which leads to performance issues.

Recommendations: Farms need increased connectivity to support the streamlined adoption of autonomous systems in the future.

Information systems infrastructure must be developed apace of new autonomous technologies to support the rapid adoption of technology such as autonomous agricultural vehicles on-farm.

Superfast networking

An example is the Hands-Free Farm which uses LoRaWAN, 4G, Zigbee, RTK and drones, which are all connected devices, to operate efficiently.

The case for using 5G in farming shows promise, although the superfast networking technology is only beginning to be applied in agriculture.

5G is currently being rolled out through the 5G Rural First project.