KomilO - Dairy Cow Estrus Detection System

What it does

We design and develop an automatic system for detecting the estrus conditions of dairy cows based on internet of things to reduce errors in estrus detection by human and increase the chances of artificial insemination of dairy and producing milk.

Your inspiration

The demand for dairy products has increased significantly over the past 30 years. The main factors affecting the amount of milk production were a success in dairy cow propagation. However, the major obstacle in cow propagation is due to farmers' deficiencies to detect estrus conditions. As a result, the dairy cows were not fertilized on time. Therefore, they have to wait until the next estrus cycle. According to our surveying, the false insemination rate due to delay detection is 39.02%, which is very high. In Thailand, there are approximately 126,815 dairy cows failed to detect estrus state which costs 266,311,500 baht per estrus cycle.

How it works

The system consists of two sensors, attached to the ears and base of the tail of dairy cows. The ear sensor is the size of a business card, detecting the movement of dairy cows to distinguish behaviors such as walking, eating, standing, and sleeping by using machine learning techniques, then predicting estrus condition with distinguished behaviors. The sensor at the base of the tail is a force sensor. It serves to detect being ridden behavior by other dairy cows. which being ridden is a behavior that indicates entering into estrus as well The data from the sensor is transmitted over a wireless communication network to the server for processing and analyzing the results. We also designed and developed web applications for managing users, managing sensor devices, and alerting the estrus dairy cow including the optimal time for artificial insemination to the farmers' smartphones.

Design process

We surveyed 27 dairy farms in Ratchaburi province to inquire about problems in increasing productivity and found that the main problem was the failure of detecting the estrus of dairy cows by human. Therefore, most farmers wish for an automatic tool for detecting estrus cow and alerting to do artificial insemination. So we designed a sensor to detect the entrus condition, attach to the ear and base of the tail. Ear sensors collect real time data and distinguish the behavior of cow. After multiple trials, we have achieved a compact, lightweight, low-power sensor with active up to 14 days per charging. For the tail sensor, the sensor must be able to attach to the base of tail as long as possible. Because the hair of the cows is slippery, also cows move their bodies quite often. After developed in designing, we have achieved an optimal structure by using a flexible silicone material that adheres to the base of the tail and using a thin cloth to glue it onto the dairy cow with a tail strap that can adjust the length. We have implemented a low-power wireless communication network for transmitting data from sensors. We designed and developed software for managing data in frontend and backend. Then we designed web applications and connected sensors, networks, and software together.

How it is different

- Our sensors can distinguish the behavior of cows without transmitting data via a wireless network to process on the server. This will help in reducing errors due to the transmission and improve data security. - We used a low-power wireless network. Thus sensors will last longer on a single charge. - We develop a complete system that consists of sensors, software for data management in both backend and frontend, and a low-power network. - Our system can predict the perfect timing for artificial insemination. - Our system's accuracy in predicting estrus cow is 80 percent, which is higher than any other system.

Future plans

- We will develop a lighter ear sensor and consume less energy. - We will develop a tail sensor more fit to the cow's physique to avoid injury to the cow. - We will develop an automatic system for detecting injuries in the legs and feet of dairy cows by using machine learning techniques. - We will develop ear sensors to detect movement and distinguishing behaviors to predict the general health of dairy cows. - We will develop a system to predict the health of dairy cows by more than 80 percent. - We plan on exploring the possibilities and finding a partner to invest in and bring this system for commercial use.

Awards