In the old days, all data collected on the farm was done manually – a time-consuming process prone to errors. Today, data collection is an easy process to automate, streamlining poultry production.
“Agriculture is one of the least digitized industries,” SriRaj Kantamneni, the managing director for Cargill Animal Nutrition’s Digital Insights business, said at the 2019 Poultry Tech Summit, held November 20-22, 2019 in Atlanta, Georgia.
“It has never been more important to integrate technology into our farms. We’ve got to start talking about data, sensors and machine learning. There is a lot of opportunity before us.”
Too much information?
At the end of the day, the volume of data collected by poultry producers is changing.
Modern technology makes it possible to collect video and audio files, giving users a more interactive way to monitor flock health and behavior beyond the written word. However, not all data is good data, so producers need to invest in methods that parse the data in order to get at what’s important.
Cargill has invested in image recognition software and robot “nannies” to help augment where labor is short to detect birds who are sick or need attention. Fully connected feed programs could help farmers determine the best-cost diets for flocks to improve return on investment.
Technology solutions currently used for other agricultural producers could easily translate to chickens and turkeys. “What they’re doing in dairy and swine could be of great value to the poultry industry,” Ketnamneni said.
Cargill recently partnered with Agriness, a Brazilian company that specializes in farm management software for swine, with the intention of developing a digital solution for the poultry market.
The promise of machine learning
The future of data collection will be machine learning, a branch of artificial intelligence where systems learn from data and make decisions with minimal human intervention, Kentnamneni said.
To work well, machine learning will require the integration of multiple kinds of photographic, audio and other kinds of data. Farms will need to be prepared to pilot new solutions – and be ready to pivot when things don’t work ready.
Teaching a computer how to make the right decision for poultry operations will take human intervention at first, Kentnamneni said – likening the process to the test drive of a self-driving car.
“Three years ago, if you were to get into a Tesla and flip the autopilot switch, it would have driven you right into a cone. Today, those cars stop automatically. It took years of reporting and self-correction by drivers for the cars to learn what needed to be done,” he said. “At the end of the day, new technology requires a lot of training before it can be useful.”
Like what you just read? Sign up now for free to receive the Poultry Future Newsletter.
