Agriculture is the sturdy backbone of modern civilisation. Over centuries it has evolved to serve one of humanity’s most basic necessities: to eat. The need for progress is now more acute than ever of course, with the global population set to climb to 10 billion by 2050. We have an epic challenge on our hands to address the shortfall between what we produce now, and what we’ll need then.

Closing the loop for true ploughing automation

For me, the search for answers begins in the soil at the planting stage. What to plant, where to plant and how to plant are three of the most crucial conundrums faced by farmers year on year. Data-driven technology has already made a significant impression, lessening the load on growers and guiding improved outcomes. Innovation will press ahead, freeing farmers to focus on more complex issues by creating further efficiencies. And closed control loops will be at the heart of the emerging technology.

We humans use control loops all the time. They’re hardwired into us to help control our bodies both consciously, when walking, and unconsciously, when we breathe, sweat or have a reflex action. Technological examples have been around for years and include something as simple as a warning light. Driven by sensors, it highlights important information about the state of a machine, such as low fuel. Manual attention is required to rectify the problem, so this control loop is reliant on human intervention.

Before the relative sophistication of warning lights, the operator had to be alert enough to discern any number of problems. This was a tricky and time-consuming process, which was resolved by a warning light that could pinpoint the nature of the problem precisely. The operator then had just one job.

Taking human interaction out of the loop is a lynchpin in technological progress. Automating the simple, repeatable small tasks is the first step on the path to even greater autonomy. Reacting to a signal without human intervention is known as closed loop control. There are many modern agri-tech systems that are closed loop in nature, liberating operators to direct their attention to multivariable problems and focusing their efforts on maximising yield. Follow the link to read about our approach to closed loop ploughing

Innovation driving better control systems 

Sensors are a crucial part of any closed loop control system. They serve as inputs, and their data is used to inform the system on what action to take. Sensing technology continues to advance. And this progress will facilitate the automation of what is now thought to be too difficult. Improvements in sensing do not just mean improving sensitivity.

Changes in form factor allow sensors to fit where they used to be too large. Changes in cost can make it viable to add sensors which were once too expensive. Advances in durability and reliability are especially important in agriculture because of the aggressive working environment. Keeping up with new technologies such as quantum sensing is a vital facilitator of innovation as new sensors create new opportunities which we were too constrained to even notice before.  

Sensors not only allow us to optimise our equipment but also our processes. Data, or more specifically information, has been growing in value for some time and continues to do so. The data provided by sensors, and the information offered by the subsequent signal processing, can guide us to large improvements everywhere. In the context of this article, data collected at harvest time can reveal a lot about how effective the planting was.  

Data relating to seed feed rate, planting depth and seed coverage can be useful to promote a better crop, especially if a control loop ensures that these variables are kept around their optimal values. Leveraging sensing data at planting time will provide a more detailed picture of how well the planting went. The insight can be used to optimise the crop growth in the future. While this isn’t a new idea, the improvement in sensing technology will only ensure finer detail and more accurate data – and so better optimisations and predictions. 

Sensors allow us to discover how to act, but alone they do not perform any further actions. This is where actuators come in. Actuators, like sensors, are a vital form of technology that continue to have an enormous impact on the world as we interact with it. These basic building blocks, in the form of lights, motors, solenoids and so on, have been changing the world for many decades. 

Over time, their cost, power efficiency, size and control precision have been improving, much like sensing technology. In many cases sensors are being built into actuators by default, making them easier to work with and providing extra safety by helping ensure the component can’t exceed certain limits. With advances in robotics and related fields, our ability to create machinery which can adapt to mechanical failure will bring further benefits. 

Control systems in action 

The workhorse of the planting operation is the seed drill, the invention that revolutionised the process. Connected air carts can supply massive quantities of seeds to the drill by blowing them through pipes. There are many control systems in place to ensure the seeds are passed to the drill for planting at the correct rate and with the correct spacing.  

They are not yet perfect. High speed planters are still niche and expensive. Any system with so many moving parts leaves room for mechanical-based errors, such as pipe blockages, pipes falling out of position, or even wheels falling out of alignment. There are plenty of sensors to warn of errors like this, but no actuators yet in place to rectify them. It remains a manual task. 

Another important responsibility of an operator is to make sure the seeds are planted at the correct depth. This is generally a very rough process where the operator manually samples random locations to make sure the seeds are being planted correctly. By combining powerful imaging technologies such as radar, ultrasound and x-ray, it could be possible to generate 3D visualisations of where the seeds are deposited – providing certainly that optimal spread and depth is achieved.  

Calibrations are another important step and are a manual part of many processes which are otherwise fully automated. Auto-calibration is still seen as potentially unreliable, and so calibrating things such as seed dropping rate remains a manual job. Another limiting factor is cost, especially with calibration which remains cheaper as a manual exercise. But as the scale of the agri industry increases to meet growing demands, such manual processes will be bottlenecks. 

There are many other control loops around. Automatic steering is an obvious example of a closed loop control system and has been automated in various farm vehicles by providers like Trimble and John Deere. This is an improvement on wholly-manual steering but has not yet reached a fully-autonomous level.  

How do we scale to feed the 10bn? 

In my view, there are two key challenges to overcome if we are to scale our food production radically and successfully. First, we must use all the planting space efficiently. Planting seeds optimally must be the objective – and to do that at scale means doing it well on the first run through the field. A second run to fill gaps is not sustainable on large areas of land. The control loops associated with optimal seeding are of crucial importance and will have a significant global impact. 

If you would like to learn more about agri-tech

The second imperative is to remove as much manual intervention as possible. Much has been achieved, but opportunities remain. Manual onsite repairs and calibrations slow the vital progress we need. Perhaps it’s time start rethinking our cost benefit analysis of maintenance drones and droids – and fast-tracking the consensus of what jobs should and can be automated.  

So, what is around the corner? We are certainly entering challenging but exciting times. What was once considered science fiction is now shaping real-world applications. There’s plenty more to come, and the team at Cambridge Consultants continues to explore closed loop control innovation for a number of clients across industries. If it’s an area that interests you, please don’t hesitate to drop me an email. Meanwhile, you can read more about our work in agri-tech here

Liam Lombard
Software Engineer

Liam is a software engineer with a strong focus on simulations and control systems. He has also worked on genetic algorithms, secure architectures and concept generation to help clients bring new ideas to life. He enjoys combining technologies in novel ways to produce useful products.