It’s been clear over the last century that farms are getting bigger. The economics of a big scale farm are better, and the risk of failure is diluted. However, the level of manpower available to do the work is spread more thinly, meaning bigger and more effective machines are needed. The drawback, however, of driving a wider swath across the farm is that it’s only possible to treat the whole field in the same way.

At harvest time, it will become obvious that not all of the area is performing in the same way – the yield varies from place to place. True, a large farm can accept this variability: but it still represents losses and waste. This variation can arise from a wide range of sources – pests, weeds, nutrient levels and soil condition. Over time this variability could be addressed with controlled, bespoke precision agriculture – but no farmer has the resources to do this. Worse, measuring what’s required with enough detail over a modern arable farm is a formidable challenge.

No-one will take on technology for its own sake, but forward looking farms are starting to think about how to improve the consistency and repeatability of their crop cycles: both to reduce risk and to improve profits. Exactly as textiles moved from being made in the home to being machine made and then highly optimised for cost, the forces pushing for automation and optimisation are inescapable. The key to achieving this is assessing and treating the land at a smaller scale without increasing costs.

Agricultural machinery developers will have to be at the forefront of this revolution. It’s clear that the increase in efficiency by making machines bigger (wider) is close to the limit. New developments need to be about treating plants almost individually, i.e. precision application, but on a massive scale. This isn’t about drones or autonomous robots constantly working the fields, but about a large scale machine individually treating thousands of plants in a single pass.

The technology that this requires exists – up to a point. Academic and commercial researchers have developed systems for, for example, plant phenotyping but these rely on photographing the plants individually and can’t predict yield accurately. To scale this up (both faster and wider) requires both dramatically improved hardware and specialised hardware. Soil condition monitoring is even more difficult as it currently requires taking a soil sample and doing a wet test. Although NDVI is common for remote sensing, hyperspectral imaging systems with much wider range are expected to make an impact in this area.

The economics and scale of this means that there will be consolidation in the industry – for instance the proposed merger between Bayer and Monsanto (and Du Pont and Dow) shows how big a player needs to be to win. There is less interchange between chemical and hardware companies though – and one will soon be entirely reliant on the other as farmers begin to buy crop treatment as a service rather than a shelf of products.

The future’s bright for sharp, new commercial products which are ‘field ready’ – there’s a clear commercial and human need, yet the technology is ripe for rapid development. In fact, it looks a little like the automotive industry in the 1920’s: a period of consolidations followed by massive growth, all enabled by new technology and customer demands. Soon though, the automotive industry could move to a different model - where self-driving cars are a service rather than a product. Farming may go the same way: machines are expensive but highly effective so a farmer doesn't need to own one anymore. That way the development and ownership costs can be spread over a wide area making the cost not only lower but predictable.

Simon Jordan
Senior Sensor Physicist

Working in our sensing systems group, Simon specialises in navigation and communication. Before joining Cambridge Consultants, he spent ten years at Teledyne TSS, working on projects including electromagnetic pipe tracking/survey systems, ship steering systems, marine motion sensors, and the development of high grade inertial navigation systems.