MOO-ve Over Alexa, Meet the Udderly Efficient IDA (a.k.a. the Intelligent Dairy Farmer’s Assistant): How One Food Conglomerate Helped Pilot Machine Learning to Make the Dairy Industry More Sustainable

In 2017, Danone, a French-based food conglomerate, introduced a new corporate vision, “One Planet. One Health.” As the leading global retailer of dairy-based products (4% global market share), Danone inspires its consumers to care about where their food comes from, how it arrived in their hands, and how it impacts their health and the health of the planet [1].

To create these purpose-driven brands, Danone relies on a supply chain of 140,000 dairy farms around the world. These farms annually produce ~1.8 billion gallons of fresh milk worldwide from ~1.4 million cows. Danone directly controls 8,000 farms, accounting for ~90% of the company’s milk sourcing [2]. With agriculture at the core of its business, Danone actively supports regenerative agriculture, a broad set of practices that supports soil health, farmer empowerment, and animal welfare [3].

Danone’s commitment to regenerative agriculture will be important in the coming years. According to the United Nations Food and Agricultural Organization, global demand for food will increase ~70% by 2050 from current levels [4]. Farmers around the world – including those within Danone’s supply chain – must supply this demand, while overcoming factors such as water scarcity, labor constraints, and climate change. Within such a context, success will depend on farmers’ ability to maximize production while minimizing inputs and the costs of waste.

Technology is one way to optimize the agricultural industry, which currently lags most industries on the digitization curve [5]. Of relevance for Danone’s farmers, the practice of precision livestock farming aims to increase production yield by effectively using digital technologies.

Indicative of this trend, Danone recently participated in an eight-month-long pilot program with Amsterdam-based startup Connecterra, an artificial intelligence company which uses sensors and machine learning to solve problems in the natural world [6]. For the trial, Connecterra provided Danone cows with an Intelligent Dairy Farmer’s Assistant (IDA for short), which essentially is a FitBit for the dairy cow with predictive capabilities for the farmer. Secured around the cow’s neck, IDA’s biometric sensors record the animal’s physical movements, including eating, drinking, ruminating (chewing her cud), sleeping, walking, and standing. Using Google’s TensorFlow open-source platform, IDA’s machine learning algorithms track, monitor, and identify trends from these physical actions. IDA then provides multi-lingual notifications to the farmer about the cow’s feed and water intake, behavioral tracking, and potential health issues. As the farmer responds to and validates insights and more cows are connected to the system, IDA’s predictive analytics become more sophisticated with accuracy currently at 80%-90%. According to a Connecterra co-founder, Danone is exploring building additional functionality on the Connecterra platform [7].

Pathways to Just Digital Future

Watch this tech inequality series featuring scholars, practitioners, & activists

Although feedback from the Danone-Connecterra pilot is limited, other farmers speak favorably of their use of Connecterra’s platform to optimize their herd’s feed intake, improve animal wellbeing, reduce loss, improve labor effectiveness, and increase farm profitability overall [8].

Despite the buzz, Danone-affiliated farmers’ widespread adoption of IDA or other IoT technologies will be hindered by factors such as cost (Connecterra charges ~$75/cow start-up fee plus $3/cow monthly fee), poor wireless connectivity in rural areas, and even farmer skepticism for the underlying premise of substituting machine learning for human experience and intuition [9].

Despite these challenges, Danone should continue to support its farmers’ application of machine learning to livestock management and explore the long-term benefits to the company’s own corporate objectives. For example, Ida’s monitoring and predictive capabilities could encourage farmers to alter their operations to include more pasture-based grazing – which is potentially less resource intensive than zero-grazing systems — as IDA would continue to provide real-time feedback on the animal’s well-being outside the barn [10]. Additionally, by accessing IDA’s data streams, Danone could more accurately assess the environmental impact of its supply chains and even pass this “pasture-to-shelf” auditing on to consumers. Such actions would reinforce Danone’s sustainability commitments and could influence consumers to buy Danone products based on perceived environmental, aesthetic, and animal welfare advantages of sustainably-conscious production systems.

The Connecterra pilot program impacted a fraction of Danone’s farmer-partners. Yet, as a leading global dairy product provider, Danone can influence the spread of innovation across its supply chain. Encouraging its dairy producers to adopt big data and predictive analytics is one way that Danone can achieve its stated goals to improve farm efficiency while also elevating the quality of its supply chain.

As Danone and its farmers explore precision livestock farming, is IDA – a bovine FitBit — really the answer? Are there similar technologies such as image processing coupled with predictive analytics that will prove more adaptable and scalable for Danone’s widespread network of farmers? As the library of generated data becomes more extensive and valuable, who – the farmer, Danone, or a third-party like Connecterra – will own and control the data and insights? Finally, are there genetic implications of using machine-learning to prematurely weed out underperforming cows from the herd?

(Word Count: 800)

Endnotes:

1. Danone S.A., 2017 Annual Report (Paris: Danone S.A., 2018), p. 9-10.
2. Danone S.A., “Danone Animal Welfare: 2016 Position Paper (Updated March 2018),” https://www.danone.com/content/dam/danone-corp/about-us-impact/policies-and-commitments/en/2018/2018_03_16_AnimalWelfare2016PositionPaper.pdf, accessed November 2018.
3. Danone S.A., “Regenerative Agriculture,” https://www.danone.com/impact/planet/regenerative-agriculture.html, accessed November 2018.
4. IIOT in Agriculture (December 2017), Frost & Sullivan, accessed November 2018.
5. Kevin Laczkowski, Asutosh Padhi, Niranjana Rajagopal, and Paolo Sandrone. “How OEMs Can Seize the High-Tech Future in Agriculture and Construction.” https://www.mckinsey.com/industries/automotive-and-assembly/our-insights/how-oems-can-seize-the-high-tech-future-in-agriculture-and-construction, accessed November 2018.
6. Connecterra, “About Connecterra,” https://www.connecterra.io/about/, accessed November 2018.
7. Louisa Burwood-Taylor. “Connecterra Raises €4.2m Series A for AI-powered Dairy Tech,” AgFunder News, May 24, 2018, https://agfundernews.com/breaking-connecterra-raises-e4-2m-series-ai-powered-dairy-tech.html/, accessed November 2018.
8. Damon Cline. “Digital Dairy App Helps Milk Data at Waynesboro Farm,” The Augusta Chronicle, April 2, 2018, https://www.augustachronicle.com/news/20180402/digital-dairy-app-helps-milk-data-at-waynesboro-farm, accessed November 2018.
9. Drew Harwell. “‘Cow Fitbits’ and artificial intelligence are coming to the dairy farm. But some farmers aren’t so impressed.,” Washington Post, April 8, 2018, https://www.washingtonpost.com/business/economy/cow-fitbits-and-artificial-intelligence-are-coming-to-the-dairy-farm-but-some-farmers-arent-so-impressed/2018/04/04/a23a0010-3850-11e8-8fd2-49fe3c675a89_story.html?utm_term=.c3eeeb4403c6, accessed November 2018.
10. M. A. G. von Keyserlingk; N. P. Martin; E. Kebreab; K. F. Knowlton; R.J. Grant; M. II Stephenson; C.J. Sniffen; J.P. III Harner; A.D. Wright; and S.I. Smith, “Invited review: Sustainability of the US dairy industry” (2013). USDA-ARS / UNL Faculty. http://dx.doi.org/10.3168/jds.2012-6354, accessed November 2018.