AI - from playing board games to optimizing dairy farms

With an ever-increasing population that will need to be fed, agriculture is slowly undergoing a radical transformation thanks to advancement in Artificial Intelligence. Although farming is a newcomer for this exciting technology, the applications grow at a rapid pace and will be more diverse in the near future. But what can AI do for cows?

Everybody is talking about Artificial Intelligence (AI) and Machine Learning (ML) these days, and the pace of implementation of this science in different types of industries is fast. In short, AI concentrates on mimicking human decision-making processes and carrying out tasks in ever more human ways. AI can display reasoning, learning, planning and creativity, all activities that we thought only humans could do. ML in turn is used to be the vehicle that is driving AI development forward. The intention of ML is to enable machines to learn by themselves using the provided data and make accurate predictions. 

Learning to play board games 

The board game Go (an abstract strategy board game for two players) was long seen as too complex for AI to beat expert human players. Eventually, AI was able to do so. ML algorithms were trained to play the game without knowing much about how to play the game. As it played more and more games, it learned to solve the problem through new data in the form of moves. This example of AI is called predictive analytics and decision support, aimed to understand how past or existing behaviors can help predict future outcomes or help humans make decisions about future outcomes based on these patterns. Other forms of AI include hyper personalization, autonomous systems, conversational/human interactions, patterns and anomalies, recognition systems, and goal-driven systems (Forbes, 2019). Companies in all sectors can benefit from one of these patterns of AI to optimise and automate their processes to increase profitability. 

Potential of AI in farming

AI is making our daily lives safer, faster, and more convenient than ever before. And it can do the same on our farms. A recently published market report states that the market for AI in agriculture is expected to reach US$ 2.6 billion by 2025. The growth is driven by the growing demand for agricultural production owing to the increasing population, rising adoption of information management systems and new advanced technologies, increasing productivity by implementing deep learning techniques and growing initiatives by worldwide governments supporting the adoption of modern agricultural techniques. The potential of AI in farming is huge, because farmers have data in large quantities, the data is rich in variety and the data is being collected and processed at ever increasing speeds. 

Robots, monitoring and predictive analytics

But only until recently, most farmers associated the abbreviation AI with Artificial Insemination in dairy cows and pigs or Avian Influenza in poultry. But this is changing at a rapid pace. Currently, the most popular applications of AI in agriculture appear to fall into three major categories: 1) robots, 2) crop and soil monitoring and 3) predictive analytics. Robots are used to compensate for a shortage of labour for repetitive tasks such as harvesting crops. Crop and soil monitoring are based on combining data from drones, satellites, weather and other sources to predict when and where the farmer needs to irrigate and use fertilizer for example. Predictive analytics can be used to learn what policies and treatments work best to improve harvest and accelerate the process to make farming more sustainable. In livestock farming we see a flight in the development of robots and predictive analytics. 

New data coming from sensors and cameras  

How a cow behaves tells a lot about the well-being and health status of the cow. Is she standing a lot or spends more of her time lying and ruminating? And does a farmer know what his cows are doing at night? While we all get used to how cows act when we are around them and think that is normal behavior, the fact of the matter is that they act differently when we are not around. 

An increasing number of companies are active in wearable sensors for dairy cows and target to optimize different assets on the farm. By using sensors attached to the cow or with the installation of cameras on dairies, it is now possible to turn digital imagery of animal behaviors and activities like eating, drinking, heat detection, lameness, body condition score, feed delivery, push-up alerts etc. into meaningful insights that are readily available on a phone. By collecting this wealth of cow behavioral data 24/6, 365 days per year, we can use the data to improve the following:

• Fertility

• Health and welfare

• Feeding

• Labor issues

• Milk quality

Fertility
Farmers used to check the cows with their own eyes to see if they show ‘in-heat’ behavior, meaning they are ready to be inseminated. You can imagine that for large scale dairies this is a lot of work and human error. Over the last years, wearable technologies have been used to assist the farmer when the cow is fertile. These simple sensors are based on a pedometer and an increase in walking often means the cow is fertile. But these systems are not savvy enough to improve and learn from the data. Heat detection technology is not brand-new, the high economic impact of having a good or bad heat detection system has driven the development of many technologies - including AI - to automate this activity.


Health and welfare
Cow health and welfare are key assets to run a profitable and sustainable farm. The use of new technologies such as AI can help in early detection of diseases, enabling quick intervention by the farmer to treat the animal. Quick interventions of health related problems can save money and antibiotics and is better for cow welfare. The company HerdDogg utilizes a smart ear tag and mobile receiver to monitor livestock health and activity. This company uses proprietary algorithms to analyze the data and provides the user with insights about their herds health and behavior. The agtech company Cainthus is focusing on animal health and welfare by monitoring cow behavior and to produce a cow comfort index, an indicator of lameness and other health issues. 

Feeding
Also camera vision technology is entering the dairy sector. Cameras are used for monitoring cows 24/7, 365 days a year, analyzing their well-being, productivity and whether there is enough feed for the cows for example. By combining computer vision (cameras in the barn) with AI, farmers can automate tasks that the human visual system cannot do on a large scale. This can help farmers to secure  that cows always have access to feed, which is being done by Cainthus. As mentioned earlier in this article, this company also monitors cow behavior to be used for health monitoring. AI and sensors are also used to move animals autonomously via virtual fence lines, saving costs on building fences for large scale grazing farms. The company Vence is a good example of this. 

Labor issues
AI and camera vision work around the clock, which makes it ideal to automate labor intensive processes, such as heat detection, checking on cows, and feed management. The use of technology can save hours per day on large dairies. Considering that finding good skilled labor is a true challenge and pain point for many dairy farmers, the idea to use technology instead appeals to many. This will help farmers in restoring their work-life balance and spend more time with their family and friends.

Milk Quality
Milk quality and safety is very important for farmers and dairy processors. AI can help to improve milk quality, as done by companies such as SomaDetect and Labby. Sensor technology with cutting-edge computer-vision and deep-learning algorithms can measure milk contents such as fat, protein, somatic cell counts, fat to protein ratio, progesterone, and antibiotics for individual cows in real time during milking. 

AI to be integral part of food production
The aim of AI driven technologies should be to allow dairy producers to be more productive and efficient with their time, decisions and the management of their operation while allowing for a more cow-centric approach. At the same time, applying new technologies such as AI in livestock is challenging, because it means you have to optimize the external, living world. We are dealing with animals and farmers and each farm is different and unique, depending on where it is located and which farm management approach the farm is aiming for. This means a lot of unpredicted circumstances. Nevertheless, the rapid implementation of AI in dairy farming is truly promising and will soon be an integral part of the dairy farm and the whole food production chain. 

by Tyler Bramble PhD, Cainthus Portfolio Growth Manager