Artificial Intelligence Could Be the Holy Grail for the Farming Industry

Thursday, 19 March 2015 - 10:01AM
Thursday, 19 March 2015 - 10:01AM
Artificial Intelligence Could Be the Holy Grail for the Farming Industry
Artificial intelligence is coming soon to a farm near you. An A.I system trained to determine the condition of animals will be seen as early as 2017, says Alex Ball, general manager of livestock and production at MLA (Meat and Livestock Australia). Ball is the one who funded this project, and says beef and lamb farmers are in desperate need of more accurate technology to measure and manage their stock. So, he turned to robotics expert Alen Alempijevic to help him solve the problem. 

Alempijevic's solution was to develop an inexpensive camera- based technology to analyze cattle as they pass through a squeeze chute. A squeeze chute, by the way, is not as bad as it sounds. It's simply a strongly built stall used for examining, marking, and giving veterinary treatment to livestock. It's main purpose is just to hold the animal still to minimize risk of injury during these procedures. Alempijevic's technology takes advantage of this stillness to capture 3D images of fat and muscle deposits. Using mathematical description, as well as fat measurement obtained previously by ultrasound, and the muscle score ascribed by an expert, Alempijevic taught the machine to asses the animal's condition based on the 3D shape it sees. The machine then provides a condition score for each cow. This score could be used by farmers to decide the nature and duration of a specific cow's feeding regimen, or even by buyers to select the animal in the best condition. But most importantly, this technology would mean a transformative shift in predicting yield in livestock. 

But just how big of an impact could this have? According to Ball, "Predicting yield is the holy grail in our industry. At the moment we rely on poor information from a range of different measures...and accuracy is as low as 20 to 30 percent. This technology would mean... accuracy as high as 89 to 90 per cent." 

It's not just a giant step for farmers. Alempijevic's method of "teaching" AI to assess something is also pretty impressive. "Essentially we are enabling computers to think and reason about what they see," he says. This kind of critical thinking could take us one step closer to more human AI, and to passing the Lovelace Test.
Science
Artificial Intelligence

Load Comments