Ready…Aim…SPRAY!!!: How Blue River Technology is turning Free Spraying Techniques into Agricultural Surgery
In the search for an answer to the question of how we will feed a growing population, Blue River Technology has created a robot that kills plants and weeds so that others may flourish while protecting the planet in the process.
Global agriculture is facing alarming headwinds as it moves further into the 21st century. “World population is forecast to grow from 7 to 9 billion by 2050, one in six is already hungry and food production must increase by 70-100% if it is to feed this growing population”[i] . Beyond the macro rise in demand, there are environmental circumstances that are challenging supply. Greenhouse gas emissions are impacting the availability of water used for farming and triggering less predictable weather patterns causing farmers to lose crops due to volatile rain cycles. The variables that are challenging the farmer’s ability to grow sustainable crops are many and all of them are measured by their effect on crop yield.
Yield in this context refers to the amount that can be harvested divided by the amount that is planted. The optimal output is to have a harvestable crop for every seed sown, but because of several factors such as “the availability of water and soil nutrients, and a rage of site and soil specific factors (pests and weeds)”[ii] this outcome is rarely achieved, especially on an industrial scale. There are many ways to approach solving the yield problem. One of the approaches involves to the use of herbicides to combat the weeds that leech away nutrients and water from the crop. In 1974, Monsanto released its first iteration of Round-Up®, an herbicide designed to be sprayed on crops to kill the weeds that were susceptible to the products dominant ingredient, glyphosphate[iii] but not harm the Round-Up® resilient crops[iv]. Beyond the serious health concerns that surround the use of Round-Up® products on edible crops, there has been increased scrutiny around the efficiency of application and effectiveness of the herbicide treatment. “The common practice of regarding the field as homogeneous creates a sub-optimal treatment, often leading to an oversupply of nutrients and pesticides. Sub-optimal treatments create considerable costs for farmers and constitute a major source of environmental pollution”[v]. In other words, broadly spraying chemicals instead of intentional targeting may be doing more harm than good.
A company that is using digital technology to optimally distribute these necessary treatments in-real time to the plants and weeds that most need them is Blue River Technology. Blue River has optimized a process around an emerging AgTech trend called “Precision Agriculture” or PA[vi]. “PA technology allows farmers to recognize variations of time and space in the production resources and to apply treatment with a much finer degree of precision than previously possible”[vii]. Their newest product, LettuceBot is a tractor attachment that is equipped with a container of chemical solution delivered though small-nozzle sprayers that correspond with cameras that “identify every plant, makes a decision based on what it sees, and precisely sprays individual plants. LettuceBot automatically thins lettuce fields with a precision that increases yields and gives farmers a valuable alternative to scarce farm labor”[viii]. With innovations like LettuceBot, Blue River Technology has coined the Agtech term “See & Spray” which “leverage(s) deep (machine) learning to enable our machines to identify a greater variety of plants & weeds with better accuracy, custom nozzle designs to enable 1-inch spray resolution, and improved software for faster and more agile crop protection”[ix]. A nuance that cannot be over looked with this technology is that it is not only identifying weeds, but also secondary heads of lettuce that are planted too close together. These secondary heads have a negative effect on the prosperity of the plants required to thrive. While weeds can be killed by broadly spraying herbicides, the “thinning” of the secondary heads is often done by hand (which was one of my jobs as an organic farmer this summer), allowing the LettuceBot to not only “boost the yield of farms by 10%” but also cut down significantly on labor[x] .
Blue River is already looking at new ways to increase its machine learning capabilities with the data collected through the digital platform connected to the plow. “Next year Blue River hopes to deliver a prototype of a weed-killing robot that can operate in a similar way to its LettuceBot but for other crop types. “For the majority of farmers, weeds are not dying anymore. It’s an industry in crisis,” says Heraud. The idea would be to very selectively deliver fertilizer, chemicals or herbicides in ways that kill weeds only, but not the plants”[xi]. To look even further into the future, I can see Blue River developing technology that could measure the soil’s mineral content at the base of each plant in real-time and deliver a mix of organic nutrients that are deemed to be presently deficient in the soil, much like a doctor administering antibiotics through a syringe.
(Word Count: 784)
[i] David J. Bennett, Richard C. Jennings. Successful Agricultural Innovation in Economies : New Genetic Technologies for Global Food Production (Cambridge, UK ; New York : Cambridge University Press, 2013) pi, Google Books
[ii] Benoit A. Aubert, Andreas Schroeder, Jonathan Grimaudo, “IT and an enabler of sustainable farming: An empirical analysis of farmers’ adoption decision of precision agriculture technology” Decision Support Systems (1 July 2012) 510
[iii] Monsanto. “Backgrounder – History of Monsanto’s Glyphosate Herbisides”. 2005. http://www.monsanto.com/products/documents/glyphosate-background-materials/back_history.pdf
[iv] ibid.
[v] Benoit A. Aubert, Andreas Schroeder, Jonathan Grimaudo, “IT and an enabler of sustainable farming: An empirical analysis of farmers’ adoption decision of precision agriculture technology” Decision Support Systems (1 July 2012) 510
[vi] ibid.
[vii] Ibid.
[viii] “Smart Machines – LettuceBot wins 2017 AE50 award,” Blue River Technology. http://smartmachines.bluerivert.com/
[ix] “Company-About US – See & Spray Technology,” Blue River Technology. http://about.bluerivert.com/
[x] Katie Fehrenbacher, “The startup behind the lettuce robot has a new 3D crop scanner“ Fortune (June 3, 2015) http://fortune.com/2015/06/03/3d-crop-scanner/
[xi] Ibid.
Picture Credit:
“Company-About US” Blue River Technology. http://smartmachines.bluerivert.com/
Thanks for the interesting post Steve, I worked on developing soil testing equipment back in my engineering days, but this is a totally different way of going about employing technology for agriculture.
How expensive are these systems and what are the infrastructure requirements needed to employ them? Although technology improvements are great I wonder how realistic it is to employ these systems in less developed markets where they are needed most. I know you have to start in the more developed markets first, but just curious how long adoption will take for this to roll out.
Hi Steve, this sounds like a great development. I’m looking forward to hearing more about their future developments as they adapt to handle multiple types of crops. One way that farmers can organically increase yield is intercropping, and this is probably commonly used in developing countries, because it requires less technological investment, and labor is cheap. If automated systems can be programmed to manage multiple different crop types at once, U.S. farmers can enhance yields through use of both organic and inorganic (fertilizer and herbicides) means which may have additional benefits such as soil maintenance and erosion control.
Interesting post Steve. This technology seems pretty impactful. I’m curious to know what the cost-benefit analysis on the LettuceBot reveals. The machine appears capital intensive, are farmers receiving an economic beneift that justifies investing in this technology? Also, I see the negative societal view on pesticides potentially affecting the success of this technology. As society demands that farmers find alternative solutions to getting rid of weeds, separate from using chemicals, the LettuceBot is at risk of losing its usefulness. All things considers, Blue River Technologies seems like a up and comer, I’m definitely following this company going forward.
Thanks for putting together this article Steve! After visiting the farm that you worked at this fall I can see why farmers are so interested in automating this process. We grew up in an agricultural area and it was very common to see crops being sprayed aerially with planes or with 100′ wide booms on the backs of trackers; both of these methods seemed quite inefficient in how they uniformly applied the product.
My question is in relation the the intervention method(s) that the machine is using. I get the fact that the machine is selectively spraying the ‘bad weeds’ but how is it removing the excess lettuce heads. Does it have mechanical ‘fingers’ that pick out the bad excess weeds? If so, do you think that they could completely eliminate weeds through a mechanical process as opposed to chemical?
Steve, thanks for the post–great read. I really liked how your article pointed out the complexity of agricultural ecosystems, and of the varied impacts of our scientific interventions. Some uses of science increase yields in the short term but cause long-term problems such as weed resistance; others are more long-term solutions, even if harder in the short-term. Some new technologies hurt the environment, others help. I loved the specificity with which you discussed particular impacts.
One other corner of this market, which I’ve been looking at a bit, is the companies that do satellite measurement and tracking of agriculture–another way to get at the same idea of more finely tuning agricultural interventions (as opposed to, for example, spraying out of planes!) The Monsanto subsidiary ClimaCorp, for example, is doing very interesting work in this space. I’d be curious about your opinions here; is this satellite-based approach a promising one, or is on-the-ground robotics and machine learning superior? Or will they both work?
Spencer