By John P. Desmond, AI Trends Editor
The pandemic is an excuse for many things. If ever there was a time to supplement (or replace, unfortunately) workers in meatpacking plants in the US, this might be it. But whether the robots are smart enough is a question posed by at least one professor.
The coronavirus has set up the acceleration of a move to have robots do more work in meat cutting plants. In April and May, more than 17,300 meat and poultry processing workers in 29 states were infected and 91 died, according to the US Centers for Disease Control and Prevention (CDC), as quoted recently in The Wall Street Journal. US beef and pork production was reduced by more than a third in April as a result.
The work is dangerous for humans, using knives and saws to work on carcasses moving down production lines, for an average pay of $16 per hour, according to the Bureau of Labor Statistics. The meatpackers have invested to make the workers safer, with protective equipment and workplace partitions, and many have boosted the pay to get the workers to stay.
A longer-term solution for the meatpackers is to bring in the robots. At a former truck maintenance shop near the Springdale, Ark. headquarters of Tyson Foods, company engineers and scientists are pushing into robotics.
The Tyson team is working on an automated deboning system to handle some of the 39 million chickens processed by Tyson each week. Tyson has approximately 122,000 employees to process 20% of the chicken, beef and pork produced in the US. The Tyson Manufacturing Automation Center opened in August 2019, as part of $500 million invested in technology and automation in the last three years. The pandemic will likely cause the pace to be accelerated, Tyson CEO Noel White has said.
The entire meatpacking industry is likely to accelerate automation. “Everybody’s thinking about it, and it’s going to increase,” stated Decker Walker, a managing director with Boston Consulting Group (BCG), who works with meatpackers.
The workers are close together. Workers per square 1,000 feet in meatpacking are three times the average of US manufacturing, according to BCG. This is due to the fact that the industry has not been able to better automate due to the complexity of the task. Humans are better able to cut animal carcasses that differ in size and shape. Finer cutting, such as trimming fat, remains in the hands of human workers and is critical to the bottom line.
A skilled loin boner, for example, can carve a cut of filet mignon without leaving too many scraps on the bone, which are turned into lower-valued products used in hamburger or dog food, stated Mark Lauritsen, an international VP for the United Food and Commercial Workers International Union, which represents many meatpacking workers.
“Labor is still cheaper, and humans can do those skilled jobs much better than machines can,” he stated.
A tight labor market has caused the meatpackers to invest in automation; the pandemic is only accelerating the trend. But teaching the machines to recognize and quickly adjust to meat coloration and shape, makes meat processing for robots more difficult than assembling cars from uniform parts. In meat plants, “our parts are infinitely variable,” stated Marty Linn, previously the principal engineer of robotics for General Motors, who joined Tyson last year to help direct its automation efforts.
Texas A&M Professor Skeptical that the Robots Can Learn Needed Skills
A Texas A&M University professor is skeptical about the potential for robot automation in meatpacking plants.
“It would take almost too much AI for the robot to figure out how the next animal is different from the last animal and to make all the millions of iterations that a human makes almost intuitively without really thinking about it,” stated Ty Lawrence, a professor of animal and meat science at West Texas A&M University, in a recent account in Brownfield. “It would take a fair number of photo eyes and/or end motion sensors because every animal is completely unique.”
His preference is to keep the focus on maintaining the health of the workforce, so they can continue to work. “As long as the beef processors are able to pick up as many Saturdays as possible, they’ll be able to chew through the backlog of cattle that we have right now,” Prof. Lawrence stated.
For companies already invested in robots, the pandemic is causing them to work on extending their capabilities.
As the virus spread through Japan in March, for example, workers at a warehouse in Sugito that processes personal care products were overwhelmed by a spike in demand for masks, gloves, sap and hand sanitizer. The company operating the center, Paltac, introduced protection for workers such as masks, temperature checks and frequent decontaminations. Now the company plans to hire more robots as well, according to an account in Wired.
“We have to consider more automation, more use of robotics, in order for people to be spaced apart,” stated Shohei Matsumoto, deputy general manager of the company’s R&D division. “There are going to be fewer opportunities for humans to touch the items.”
Paltac has been using robots from a US company, RightHandRobotics (RHR) of Somerville, Mass., to pick objects from bins and assemble orders. Now the company is looking for software updates that will allow the robots to recognize and grasp new objects, or retrieve items from new types of bins, Matsumoto stated.
Recent advances in AI hold promise that robots will be able to take on tasks that require greater manual dexterity. RHR grew out of a team of researchers at the Harvard Biorobotics Lab, the Yale Grab Lab and MIT research in grasping systems, intelligent hardware sensors, computer vision and apple machine learning. The firm intends to bring the latest technological developments from the lab to the warehousing industry, with the idea of transforming the supply chain.