Contributors
Subscribe to our newsletter
Dr. Anastasia Volkova, Founder and CEO of FluroSat, shares the future of agricultural technology and why you should care (published on Medium on 13, June 2019)
If we ask today’s toddlers in 5–10 years, “Who do you want to be when you grow up?” we may hear “farmer” as an answer and clearly, why not.
Whilst many still see farming associated with terms or phrases like “hard labor”, “financial risk” or ”lack of technology savviness” today, I believe that in just a few decades farming might be the most digitized and automated profession of all requiring knowledge in robotics, data science, on-edge computing and more.
As the CEO of an agtech startup, I’m often asked how technology will change the future of agriculture, and here are my thoughts.
1. Hyperconnectivity will allow seamless data flow from fields
Lack of connectivity is hurting “autonomous farming”. This means IoT sensors will be unable to continuously collect data, users on the farm will not be able to upload field observations through their smartphones, or even accessing advanced software in the Cloud that’s full of sophisticated crop stress detection algorithms. It also hinders the ability to remotely manage tractors and drones to fertilize or spray crops.
With LPWAN (low-powered wide area network) such as LoRaWAN, Sigfox and other technologies, a “connected” farm will allow more data transmission from the fields, allowing us to use agriculture technology to its fullest potential, closing the loop between data collection (with satellite imagery, IoT sensors and machinery) through crop stress detection to administering a corrective action (applying fertilizer or weeding).
2. Training of Machine Learning models and AI specific to every farm for maximum potential
Just as a modern farmer seeks to follow the best agronomic practices, the farmer of the future will be competing for commodity prices not only by leveraging their “hedging instincts” but also by applying data science skills and improving the accuracy of their farming models.
Machine learning (ML) as a way of iterative model training can learn in order to recognize, for example, a weed or a particular sign of a disease before our naked eyes could detect it. With better farm connectivity, images from cameras on satellites, planes, drones, tractors and farmers’ smartphones will be the primary sources of data for these ML models to perform its function. We’ll be able to use data to train hyper-local ML models to learn the limitations and potential of every acre on a farm.
3. Scientific modelling used to spot crop stress and recommend precision fixes
How do we go from identifying crop stress to taking preventative or correcting action? By conducting experiments.
With more collaboration between research institutions and commercial entities, we’ll bring together the successes of scientific modelling and ML into the field (“lab to paddock”). Through automating the process of experiment design and data collection, the in’s and out’s of a farm system can be learned.
Likewise for nutrient management — By applying different amounts of fertilizer or chemical to different areas in the field, we can tease out the potential of the field and optimize it by putting just the right amount of nutrient on each acre to achieve optimal yield.
4. Precision fixes can be carried out by robots (tractors, drones and IoT sensors)
So, we have models detecting crop stress and proposing a solution, now what? Our tractors, drones and field sensors — the same machines and devices that are used to collect information about crop stress — can be used to act on it!
Tractors and drones have “autopilots”, GPS-guidance module, which allows them to accurately locate themselves and release the amount of fertilizer or chemical required at the right rate and at an exact location identified by the model. Field sensors, sensing soil moisture can already be programmed to trigger irrigation system when soil moisture gets below a certain level. With IoT sensors, as well as modern drones and tractors we see the loop of sensing and acting, “measuring and managing” beginning to close.
5. Human as the master of data and as conductor of the field orchestra
In summary, with machinery and sensors getting better and lasting longer in the fields, better connectivity enabling data flow without “walking thumb-sticks” across the farm and scientific models having enough data to learn the earliest signs of stress and recommend precision fixes, the role of a human operator — the farmer of the future — will become increasingly important.
To improve farming practices in such an automated system, it’s important that we monitor the quality of data it collects and improves the “intuition” or high-level AI “logic” it follows.
Farming system of the future is a combination of a robotics yard, machine learning playground, and IoT “hatchery”. The automated systems of the future might become plug-and-play at some point, but for a while, they will resemble “algorithm laboratories” — similar to FluroSense, an agronomic analytics engine that we’re building at FluroSat.
Conclusion
The farmers of the future are people who want to see a sustainable planet and can develop intelligent systems to learn about what our planet needs. And, wouldn’t you want our next-generation to re-discover the very secrets of this fascinating place we call “home”? I certainly do!
