If your business is making agricultural equipment, you’d be forgiven a wry smile at the ongoing frustrations of automotive OEMs as they strive to develop autonomous vehicles. After all, semi-autonomous tractors have been commonplace in your industry for more than 15 years thanks to guidance technologies like RTK GPS and camera-based navigation systems. Entire fields can be mapped – and routes planned and navigated – by a few swipes of a tablet.
Improving end-to-end efficiency through the integration of technology.
That’s all well and good of course, but I’d like to play devil’s advocate and ask you this: why is there still an operator in your tractor and what’s stopping broad acre farming from being truly autonomous? The answers to these fundamental questions, I would suggest, can be found in closing the loop – and automating the processes that rely on operator judgement and skill.
The team here at Cambridge Consultants has spoken at length with large corporations and individual farmers alike. We’ve attended countless conferences and viewed reels of YouTube videos to identify the recurring problems faced by farmers, even when operating state-of-the-art equipment. Universally, a potential overarching solution to many of the issues arises… closed loop control.
A control loop takes a measurement, compares it to a target, and then attempts to minimise the error by varying an output. Control loops are everywhere – cruise control in our cars, voltage control in our smartphone chargers and temperature regulation in our office air conditioners. There’s even a control system at play when we sip our way through a coffee, as we don’t tend to miss our mouth despite the changing cup weight. Human brains are running control loops all the time.
And their relevance to farming? Simply put, many of today’s agricultural tasks rely on the skill of the operator to get them just right. The ability to ‘know what good looks like’ and what to adjust to ensure it is achieved is something experienced farmers take for granted. Yet it is this expert knowledge that is inadvertently hindering the move to full, level 5 autonomy. So, before the marketing dream of the cab-less tractor can become reality, we have to nail the automation of the processes that rely on operator input.
Autonomous real-time solutions
This is a broad issue that we’ll be exploring in a series of upcoming articles. I’m focusing this first one on the specific challenges faced during ploughing – and how closed loop control could potentially be used to provide autonomous, real-time solutions.
Let’s start with furrow width. The operator must ensure the top link is aligned and horizontal to the tractor. If it isn't, after swapping over, one side will plough narrow and the other will plough wide. They must also check that the front furrow is ploughing at the same width as the other furrows, and that this is the same as the set width.
Now furrow depth. The operator looks closely for even amounts of soil on the mud boards to ensure furrow uniformity. They must ensure that the skimmers are run deep enough so that the trash is cut and buried under the furrow – and must ensure that the back furrow before a turn is the same as the front furrow after a turn. Once again, the furrow depth should be manually checked against the set point.
Finally, taking the plough in and out. When dropping the plough back in, the operator wants the plough to drop and to touch the floor just before the wheel drops in the trough. When taking it out, they must time it so that the troughs are filled in.
All of these steps involve a visual feedback loop. But with appropriate instrumentation – machine vision, force transducers, proximity sensors for example – the desired properties and their spatial variation can be quantified and fed back to the control system of the tractor. This allows for real-time changes in speed, plough depth, angle and so on.
The true potential of precision agriculture
With greater data collection, understanding of cause and effect, and perhaps even elements of deep learning to build on the collective wisdom of thousands of farmers, closed loop systems could correct non-optimal processes before the inconsistencies are even visible. Such systems could enable closed loop control to deliver on the promise of true autonomy on the farm, freeing up precious time for the farmer and delivering greater yield and consistency of outcome across the farm. It holds the keys to the true potential of precision agriculture.
Farming gets smarter with AI at the edge.
We at Cambridge Consultants have put our extensive cross-disciplinary closed loop control expertise to good use for a number of clients.
In beverage dispensing, we have developed a system which measures motor current to infer the ice content in a frozen cocktail machine. (This proved far more accurate than measuring temperature!)
In drug delivery, we have controlled aerosol dosage by using conductivity sensors to monitor and vary liquid feed to the atomiser
In industrial catering, an ingenious application of AI and robotics streamlined tray cleaning by identifying and removing crockery, cutlery and glasses