A Brief Survey of IoT Solutions for Agriculture

Internet of Things enabling increased visibility into what drives agriculture.

The farming industry has seen rapid advancements in productivity over the last century. Over this period, technologically advanced nations have gone from allocating a majority of their workforce to farming to allocating only a small minority. Advanced nations have leveraged machines to remove much of the manual labor, pesticides to combat pests which lower crop yields, and fertilizers as a kind of steroid to boost growth rates, yield, and keep soil fertile over time. While these advancements have been huge boons to the industry, as a society we still face a 70% increase in consumption of food by the year 2050 according to a report from the Food and Agriculture Organization of the United Nations. 1 We can view the past advancements in farming as a kind of brute force method to improve crop yields. The main motivators were based on ideas that were apparent upon inspection and borrowed from other industries. For example, the idea to use machines to replace manual labor in the field does not take much imagination and was in use throughout the economy during the industrial revolution, granted that the construction of these machines is complex.

However, future improvements must come from a more evidence based approach. We will need to harness the invisible (at least to the naked eye) data that drives crop yields such as sunlight, soil conditions, precise water delivery, and localized genetic properties. A great example of such an effort is the suite of products offered by Kaa.2 Kaa provides a software development kit (SDK) to integrate data from connected objects, such as sensors, with back-end infrastructure commonly referred to as servers. The customer base for such applications is huge as it includes virtually all farmers on the planet that aim to use technology to increase yields.

While it is straightforward to purchase and place sensors to monitor soil attributes, moisture, and sunlight, it can be difficult to aggregate this data and then make informed decisions based upon this data. Kaa enables a user to develop a system to perform A/B testing, remote device configuration, real-time device monitoring, and aid in the collection and analyzation of sensor data. This kind of functionality and ease of use is key for developers. As more developers work with such tool kits the prices of software suites will come down. This will increase adoption for farms under family operation which still constitute a substantial number of commercial farms.

Phenonet is a system that “collects, processes, and visualizes sensor data from the field in near real-time” per their website.3 The article header is an image of their system in use. This particular network is advertised to help cut out labor costs associated with testing different crop varieties by automating the yield measurement process. When comparing crop varieties, it is essential to control for variations in the environment surrounding the crops. The Phenonet system aggregates the sensor data which can then be used to generate statistically significant comparisons. These are the same type of comparisons that were essential in the Indigo case when the research teams were trying to compare the efficacy of different seed coatings.

These two examples provide an overview of a fragmented IoT industry that is attempting to revolutionize the farming industry. Sensors must continue to improve in terms of lower cost, lower power, better accuracy, longer lifetime. The software used to interface with the sensors must continue to improve in providing a clear representation of the data that is both actionable and accurate. An emphasis on machine-to-machine actions will further reduce labor demands on the farm.

Overtime, as these integrated systems become more affordable, effective, and available we will see an increase in their use and more players move into the space to try and generate profits from the increase to the farming value chain.

Word Count: 632


Bigbelly Trash Cans: They know a lot more than you think


Spoiler Alert: At the End, Technology Defeats Food Waste

7 thoughts on “A Brief Survey of IoT Solutions for Agriculture

  1. Great post! I’m happy to see that there are companies continually pushing to increase yields in Agriculture, because at the end of the day that’s what I eat! I’d be curious to see if Kaa has any data around current ROI – I think the big question most farmers will ask is if their yields will increase enough to offset the purchase cost?

  2. Interesting post! Would love to see some results of yield improvement / cost saving after adopting this. Another IoT application to agriculture I’m aware of is drone. DJI (largest drone company globally) Agras MG-1 is an octocopter designed for precision variable rate application of liquid pesticides, fertilizers and herbicides, bringing new levels of efficiency and manageability to the agricultural sector. The powerful propulsion system enables the MG-1 to carry up to 10kg of liquid payloads, including pesticide and fertilizer. The combination of speed and power means that an area of 4,000-6,000 m² can be covered in just 10 minutes, or 40 to 60 times faster than manual spraying operations. The intelligent spraying system automatically adjusts its spray according to the flying speed so that an even spray is always applied. This way, the amount of pesticide or fertilizer is precisely regulated to avoid pollution and economize operations.

  3. It’s incredible to see the applications for IoT similarly to what you’re mentioning. Digital monitoring at the crop level, coupled with initiatives such as John Deere’s “FarmSight,” which provides remote monitoring of equipment and machinery, could significantly reduce inefficiencies that cause a drag in the farming industry (spoiled crops, equipment downtime, poor yields, etc.). With the growing population and demand for food (and fixed amounts of real estate on our blue planet), and increases in efficiency will go a long way!

  4. Phenonet and Kaa seem like technologies with huge potential, but I’m curious about the rate of adoption of technology in the farming world. The use of tractors and other machinery is clearly widespread, but are many farmers — especially the small, family-owned farms you mention — eager to spend the capital required to add sensors and other connected technology to their operation?

    One British academic paper I found suggested that income, farm size, and access to information are among the chief determinants of whether farms adopt new technology — in the case of this specific research, artificial insemination. [1] Do Kaa and Phenonet have a ton of work cut out for them reaching the smaller, lower-income family farms you mention? Or is there enough of a market among larger farms that both companies can gain a good foothold?

    [1] Peter Howley et al, “Factors Affecting Farmers’ Adoption of Agricultural Innovations: A Panel Data Analysis of the Use of Artificial Insemination among Dairy Farmers in Ireland,” Journal of Agricultural Science; Vol. 4, No. 6; 2012, accessed 20 November 2016 at https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&ved=0ahUKEwiOy-qf8rfQAhVWzGMKHZC1Ay8QFggiMAE&url=http%3A%2F%2Fwww.ccsenet.org%2Fjournal%2Findex.php%2Fjas%2Farticle%2Fdownload%2F10687%2F11101&usg=AFQjCNHOEEWZ8xtuWQeVBSrg5jk0WRNHZQ&sig2=bRtYWKEmFMpjUdZPnh_wzQ

  5. Great post. A couple of things I thought about were (i) I liked your point on the how data for the sake of data isnt valuable, whats really needed is a way to interpret the data and gain valuable insights, and (ii) Im curious as to how these farmers are responding with regards to data rights. For the latter, we touched on it briefly in class on friday, but something that is limiting adoption of these technologies in farming and other fields is the undefined nature of who owns the rights to the data.

  6. Looking at Amy Nordrum’s projections, it would seem that the number of IOT connected devices that will be seeded around the world has been written down in recent years. Given the sources you shared above, I would disagree with Amy. Agriculture has been our world’s oldest and most labor intensive industry, still employing over 40% of our 7+ billion people. While the estimate of total devices may need to be written down for 2020, I think that the extrapolated future potential for devices should still continue to be increasing. If we are able to develop IOT at the nanoscale, then the number may increase by many orders of magnitude. These nanodevices, should they be applied to farming, could interact with the field on a per-plant bases, or even multiple devices interacting with each individual plant. Over time, the question of cost vs value comes into play in terms of adding additional sensors and actuators to the field.

    1. http://spectrum.ieee.org/tech-talk/telecom/internet/popular-internet-of-things-forecast-of-50-billion-devices-by-2020-is-outdated
    2. http://www.momagri.org/UK/agriculture-s-key-figures/With-close-to-40-%25-of-the-global-workforce-agriculture-is-the-world-s-largest-provider-of-jobs-_1066.html

  7. Fascinating! What I find particularly appealing about Kaa is the ability to perform A/B testing. One of the limitations of machine learning, both in agriculture and more broadly, is the ability to generate exogenous variation to help generate a causal estimate. Even with lots and lots of data, if you don’t have sufficient variation, your estimates will lack precision, and if this variation is driven by confounding variables/if you have concerns about omitted variable bias, then your estimates may not even be accurate. For this reason, the ability to do A/B testing is a substantial innovation, as it allows experimental causal estimate to be generated that should be both precise (assuming a large enough sample size) and accurate (assuming the experiment is correctly run).

Leave a comment