Bordertown robotics technology window to processing’s future + VIDEO

Jon Condon, November 28, 2014

THERE’S something unsettling about seeing a series of shiny, articulated, stainless steel robotic devices going about the normally labour-intensive business of processing livestock carcases, without human intervention.

But that’s exactly what’s happening at JBS Australia’s 8000-a-day Bordertown lamb plant in South Australia, where a jointly funded R&D project is providing a window into the future of meat processing in Australia.

Robotics and automation are increasingly being seen as a means by which Australian processing can fight back in its lack of competitiveness in processing/labour costs, often quoted as being at least 50 percent higher than those in the US, and perhaps double those in Brazil.

A short video of the Bordertown plant installation has now been seen by producers attending the recent LambEx conference in Adelaide, MLA’s annual general meeting in Sydney, and again in Tasmania last week where JBS held a supplier awards night for its Great Southern farm assurance programs.

Here are two videos displaying the system at work on lamb carcases at Bordertown


For those who can remember back far enough, the video footage looks vaguely similar to images produced during the industry’s original, failed attempts at kill floor automation back in the early 1990s, called Fututech.

In the past 20 years however, computing and robotics technology has travelled light years, and a process that proved impossible to achieve back then through systems like Fututech and VIAscan is now on the industry’s doorstep.

Not only does the Bordertown robotics project lift labour efficiency at the plant at the carcase splitting and breakdown phase, but other technologies built into the system, including Dual- Emission X-Ray Analysis (see further details below) provide two and three dimensional views of each carcase, providing lean meat yield predictions and optimum cutting lines with a high degree of accuracy.

Dare we say it: it’s starting to look like a true value-based marketing model including parameters like meat yield might be on the horizon for the red meat industry.

While the Bordertown project model is built around lamb carcases, high hopes are held that the same technology can be applied to beef, also, producers were told last week.

MLA’s Dr Alex Ball highlighted the progress being made in the JBS Bordertown project during a presentation to beef and lamb producers attending JBS Southern’s supply chain forum in Launceston last Thursday.

Using the system – completely automated and operating without human interaction – operators know exactly where each skeletal position on each individual carcase is, based on the DEXA scans. Regardless of whether it is an 18kg or 32kg lamb carcase, the robotic saws can be programmed to cut between the four and fifth rib, for example, delivering the optimum yield result with tolerances of less than 5mm.

Other robotics components take off the loins, with cutting lines within 2mm of the optimum. Once the saddle is removed, a hyper-spectral camera is used to take two images, which delivers an intra-muscular fat prediction on both ends, providing an eating quality potential, for each item. Cameras also take eye muscle images for yield calculation, and images for objective meat and fat colour measurement.

As the video embedded in this story will show, the Bordertown system will process up to 15 bodies each minute, at the same time delivering real-time estimates of lean meat yield and eating quality.

Far from being a ‘laboratory’ level experiment, the installation is running daily, at commercial line speeds, while delivering individual carcase data on yield and meat quality performance.

“Every JBS supplier to Bordertown will get access to that yield data on their lamb,” Dr Ball said.

“My challenge now, is to convince a couple of automation people to build a beef version,” he said.

“If we do this – put this system in every plant across Australia – when combined with the MSA index we will have a single system for describing the yield and quality potential for the entire Australian beef herd and sheep flock. We’ve never been in that position before, and it is a credit to JBS that they have seen fit to make the investment.”

Being robotic, the technology could theoretically be run virtually non-stop, 24 hours a day if required, without meal breaks, sick days or holidays. Assessments also show the microbial counts on carcases processed with the technology are considerably lower than those produced by manual labour.

Rather than leading to retrenchments or reduction in staff numbers at the JBS Bordertown plant, the installation of the new technology has simply seen staff who previously worked in that portion of the chain (where there was considerably higher OH&S risk, due to the saws being used) relocated to other parts of the plant, with consequent expansion in plant throughput.


How long before yield based payments?

Having seen the yield-predictive capacity of the new technology at Bordertown in the video, it was an obvious target when a Tasmanian producer asked during Thursday’s Launceston forum how long it would be before meat yield value was included in meatworks grids.

“It’s a commercial decision between processors and producers,” said Prof Dave Pethick, one of the forum speakers said. “But the key point up to now is that processors have not had the confidence in the accuracy of the yield prediction measurements to reflect that in payment systems – in beef or lamb,” he said.

“It’s only when we can prove more reliable yield calculation methodology, using some of these new technologies, that we can start to understand where the optimal aspects of weight and fat and muscle are. When we have more reliable measures of fatness, which hopefully are around the corner, then we can start to experiment, providing feedback first, to give both the processor and the livestock supplier confidence in the yield predictive system, and who knows, after a year or two, things might happen.”

JBS’s Southern supply chain manager Mark Inglis said the company’s Bordertown, Cobram and Longford sheep plants were now linked to Livestock Data Link, the industry’s third-party online feedback system for MSA and carcase data.

“For the past four months, we’ve had an algorithm in there based on GR management and carcase weight, which is actually delivering a lean meat yield figure, to ourselves and our livestock suppliers. It’s probably not as accurate as we would ultimately like, but it’s a start,” Mr Inglis said.

Dr Alex Ball said a similar algorithm had just been developed for beef, and would be launched shortly on the LDL feedback system.

“I’m very confident that in the next 12 months we will see yield based payment systems coming on board,” he said.

“The real key, though, will be trying to work out for brand owners what the value proposition for meat yield is, versus the value proposition for eating quality. There’s a big negative correlation between eating quality and yield. Select more for one, and the other declines.”

“The balance will differ, depending on the quality of the product in the relationship between the brand owner and supplier. But we still have a lot of work to do in that space,” Dr Ball said.

“Once we get real time measurements on yield and eating quality, we’ll see processing companies and brand managers automatically gravitate to rewards (grid incentives) based on the quality and quantity of the product presented. Issues like sex and breed content will become much less of an influence in grids.”

Prof Pethick said the researchers had ‘surprised themselves’ when they found that the yield prediction capability within the current MSA grading parameters was actually ‘rather good.’

“It’s better, in fact, that the USDA yield assessment system, way better than the Japanese system, so actually we have been sitting on something that we hadn’t realised for quite some time.”


New technologies show great promise

Dr Ball outlined a host of new technologies that were emerging that held great promise to better define measurements of fat, bone and muscle in a carcase, for both sheep and beef. Some of these were employed in the Bordertown R&D project, while others were still in earlier development.

Alex Ball“In my opinion, there is no better time for the beef and sheepmeat industries to be collectively investing in technology,” he said. “The important point is that JBS is investing with us (MLA) in every one of these technologies.”

Dr Ball said it had taken the last ten years to prove that every existing technology in this field had been a “spectacular failure.”

“We’ve taken technologies that worked really well out of the pork industry, but which fail spectacularly when applied in beef and lamb, primarily because the industry uses hide-pullers that bubble the fat. We’ve taken technologies out of Europe that have failed, because they have no relevance to eating quality or yield in the Australian herd or flock.”

However there were a range of new technologies that did show great promise.

Here’s a sample:

The Danes have developed a device called a ‘Fat-o-meter’ which sends electrical currents into the chilled meat carcase using probes, to use bio-electrical impedance to tell total tissue depth, and gradations between fat and muscle, and intramuscular fat. It has been shown to work well in lamb, and will work in beef. An attraction is that it is a transportable, relatively low-cost technology that could be applied in any plant in Australia, and can be used at chain speeds. “We think we will have something available to industry in the next six months in this area,” Dr Ball said.

Another technology showing promise was hyper-spectral imaging, which operates above infra-red and ultra-violet light bandwidths. The Danish-developed camera technology can objectively define between muscle and fat, and also assess meat colour and fat colour objectively. It could also potentially be used to determine intra-muscular fat, or marbling. Future assessment could even include concentrations of omega 3 fatty acids, iron and zinc. The first prototype camera will arrive in Australia in January to trial. There was also a prospect the technology could deliver a true measure of bone-density, to get rid of the subjective ossification measurement.

Beyond the abattoir, new field-based technologies also hold great promise. As reported earlier on Beef Central, a Sydney University of Technology team is putting together an image analysis system using high quality cameras to get an impression, using laser and colour spectrum to develop a three-dimensional view of a walking animal. “We’ve trialled it and we now know that on a live animal we can predict P8 fat within 2mm and rib fat within 2mm on a live animal,” Dr Ball said. “The really great thing is that the technology is powered by an X-Box gaming console camera, worth about $150. It’s shaping up as a fantastic tool to do body composition predictions in live animals. A feedlot could have one of these sitting in their race, walking cattle through, and making incredibly accurate predictions about carcase fatness and lean meat yield.” At carcase stage, the same technology trialled in abattoirs had shown an accuracy in prediction for lean meat yield of 70 percent, at a rate of eight bodies a minute.

Another technology field, Single Emission X-Ray Analysis (SEXA) and Dual- Emission X-Ray Analysis (DEXA) take X-rays of a carcase, creating images of the skeletal components, and splits bone, fat and muscle components to deliver lean meat yield prediction. It can also objectively measure bone ‘maturity’ across a carcase, negating the need for a subjective ossification measurement. The first 50 animals (lamb) have just been assessed as part of the JBS Bordertown technology project, and the opportunity exists equally to do full lean meat yield analysis with beef.

Other work is scrutinising CT scanning technology adopted out of human medicine, allowing calculation of lean meat yield, intra-muscular fat (within cuts, rather than as a single measurement), and primal weights. Using such technology could lift MSA predictive modelling by another 5pc, Dr Ball said.

“In summary, there is a mass of emerging technologies that we are quietly confident about. These innovations are going to build a real platform for delivery,” Dr Ball said.





Your email address will not be published. Required fields are marked *

Your comment will not appear until it has been moderated.
Contributions that contravene our Comments Policy will not be published.


Get Sheep Central's news headlines emailed to you -