Philippe Stoop, Innovation Director at itk, was present at SIDO 2020, on September 3rd in Lyon. He spoke at the IoT, the age of maturity round table, led by Bearing Point, which presented a study on the subject. Philippe Stoop presented the role of Artificial Intelligence in the R&D strategy and the evolution of the itk Business Model, for its packaged services and its Farmlife herd monitoring platform.
With its subsidiary Bretone New Medria, itk is present in all the markets of the monitoring of herds by connected sensors: detection of heat and calving, food monitoring (rumination and ingestion time, grass or trough), and animal welfare. Its originality is to have gradually concentrated all these services on a single non-invasive sensor, the Axel accelerometer, permanently worn by the cow on a collar.
Make herd monitoring services even more accessible to all farmers.
Philippe Stoop answers a few questions back from the show:
How important is artificial intelligence in your strategy?
The herd monitoring sensor market was at a bit of a standstill: zootechnical researchers invented a lot of sensors that could provide valuable information to the farmer. But until now they were specialized sensors, each one being placed in the ideal place to capture the movements characteristic of each event: for example, on the neck for the detection of heat, but on the tail for the detection of calving, because that’s where the most characteristic movements are.
This is a dead end on the one hand, because the breeder cannot spend his time laying and removing sensors on his whole flock, whenever he needs new information; but also economical, because it reduces the accessible market, and confines the production of sensors to small runs, thus with high unit manufacturing costs.
The strategy of itk is therefore to concentrate the maximum of services on our Axel collar, to increase its use value for the breeder. But for the detection of calving, this implies recognizing signals much less clear and much more ambiguous than those perceived by a dedicated sensor attached to the tail. While our previous services had been developed thanks to the human expertise of our modeling team, only artificial intelligence (in this case Deep Learning) made it possible to obtain a reliable algorithm. The result is that now our Axel Collar alone provides all the information needed for real-time monitoring of cattle feeding and reproduction.
How did this change your business model?
There is a first immediate consequence: in consultation with our distributor Medria Solutions, we have moved to a marketing based on an annual subscription, and no longer a sale of sensors: the breeder chooses the package of services he wants to subscribe to, and the sensors are provided free of charge. This is a necessary transition to counter the inevitable trivialization of IoT sensors, and refocus the economic model on what has real value for the farmer: the services, and not the hardware itself.
This first step has already allowed us to revive our sales, the majority of our new customers now equipping each of their cows, while before they bought a sensor for only 3 or 4 cows. This now allows us to move to a second phase of redesign to cost of the sensor, to lower the price, and make these services even more accessible for all farmers.
How does machine learning, machine learning and deaplearning help agricultural production?
We have seen that these techniques have already enabled us to improve herd monitoring. Self-learning capabilities will further improve our algorithms, and make them even more accurate. The next step will be to cross-reference the data from our sensors with the production data of the herd, to allow the breeder to detect early eating or health disorders that could affect its production. All these future applications require so much data that they exceed the capabilities of human expertise. But it is nevertheless the expertise of our data analysts that will allow to identify the most relevant variables and indicators to be stirred by the algorithms.
Download the itk presentation [FR]: “AI and IoT, from the sensor connected to the high added value service”
On the same subject: “itk in Finland: connected cows and Christmas trees”
Predict and Decide