Data-driven insights in agriculture – international players discuss AI in Ag at sell-out event

Date:02-22-2018

A deluge of rain during the 2017 harvest slashed profits overnight – wheat for milling and barley for malting were downgraded and producers incurred additional costs for drying. Few industries have so much at stake than agriculture, and so much to gain from accurate, timely information.

The interest was clearly seen at the ‘AI’m of Machine Learning in Agriculture’ Pollinator, which had to change to a bigger venue to accommodate all the interested farmers, plant scientists and technologists.

Data-driven insights – PA Consulting latest research

Aaron Croucher, ‎Engineer & Consultant at PA Consulting, opened the meeting with findings from PA’s report “Transforming Agriculture with Data-Driven Insights”. He explained how advances in Artificial Intelligence are being used to simplify complexity and improve decision making.

“Big names in the machine and equipment and agriscience sectors are now fully alert to the opportunities from digital agritech. They’re looking to collaborate with technology companies and start-ups, assembling and analysing data from different sources to unlock new insight and help farmers make smarter decisions,” he comments.

The report identifies eleven companies that dominate the agritech space. They’re a mix of agriscience leaders (BASF, Bayer, Dow, DuPont, Monsanto, Syngenta) and machine and equipment manufacturers plus two key technology players: Iteris and Trimble.

It suggests that the trigger for interest in AI was Monsanto’s acquisition of the data analytics firm The Climate Corporation. “The $930-million deal seems to have acted as a wake-up call to the rest of the industry.”

Insights from real-time-real world data – Iteris key technology player

John Lord from Iteris spoke at the event alongside industry speakers from After the flood, Fujitsu, Kings College London, and Microsoft.

Lord explains that combining crop and environmental data with artificial intelligence (AI) can help farmers make key decisions.

“The UK harvest in 2017 was very disrupted by weather and farmers had a difficult decision over when to make the cut. Is it better to harvest wet and take the expense of drying or to wait for better weather and risk the chance of the grain sprouting in the ear?

“Mechanical drying is a major investment and grain moisture and environmental conditions can change rapidly so timing is everything. By bringing together crop health modelling with field level atmospheric data, our ClearAg app provides harvesting insights that allow users to make more informed decisions on when and where to harvest and dry crops.” He went on to explain that large US farms are reporting a significant return on investment from implementing the technology in this way.

Iteris is also using AI for smart water control. Lord explains that specific land surface models are used to forecast the soil moisture at crop rooting zones. After submitting further information and user feedback, AI is used to validate and augment the complex model process, thereby building confidence in the models and keeping them current.

New era for active data – After the flood

Humans have evolved to quickly extract information from patterns. After the flood is using AI to take data visualisation to a new dimension. It takes insights from multiple interactions (people-machine and machine-machine) and displays the findings as deceptively simple dashboards.

After the flood’s Chairman Nick Cross, who also manages his family farming business, explains: “Traditional data analytics are based on collecting data and then providing retrospective insights. We are moving into a new era of active data that uses real-time data to provide intelligent services.

After the flood creates a dynamic interpretation of live data. This allows fast reactions and the ability to create systems that learn from experience to respond to changes in their environment.

Within agri-food this could be using customer buying behaviours to predict demand for perishable goods, or monitoring fungal spores and weather conditions to allow preventative, precision spraying.

Cross continues: “I think there will be exciting opportunities to create intelligent data flows between customers, stores and the producers themselves, allowing farmers to be more responsive to specific consumers’ tastes and dietary needs. Perhaps there will come a time when food production will be personalised!”

Read more here about After the flood’s dynamic interpretation of live data.

When is machine learning artificial intelligence  – Microsoft explains

Matthew Smith, Director of Business Development at Microsoft Research, agrees: “I’ve always been excited by creating information services for the food supply chain – finding ways to get the right information, to the right people, at the right time, in the right way. That information supply chain still doesn’t exist as it should; it is fragmented and inefficient.

“Wonderfully, technology is approaching a maturity to create the information supply chains the world needs, harnessing things like cloud computing, IoT, AI and block chain.”

Smith explained the difference between machine learning and AI.

“Machine learning is the ability to infer relationships from data, rather than be explicitly programmed to do so.  Examples of this include: benchmarking, predicting shelf life, estimating soil fertility, predicting oestrus in cattle.

“Artificial intelligence is where machines gain cognitive capabilities for example; image recognition, speech. The applications within agri-food can be described as labour, safety, sustainability, productivity and efficiency.”

Needs to be meaningful and provide ROI – Agri-Tech East

Dr Belinda Clarke, Director of Agri-Tech East, comments: “Collaboration was identified as essential by the PA Consulting report and this networking event provided a good opportunity for farmers and plant scientists to meet personally with technologists from the companies that are shaping this emerging technology on an international stage.

“The opportunity within agri-food for learning systems that can track multiple sources of input from the environment and elsewhere and present this in a way that is easy for humans – or machines – to understand and take action is immense. But it is vital this is directed at producing meaningful information and provides a good return on investment.”