robotics and automation in the food industry || robotics and automation in meat processing

© Woodhead Publishing Limited, 2013

13

Robotics and automation in meat processing G. Purnell , Grimsby Institute of Further & Higher Education (GIFHE), UK

DOI: 10.1533/9780857095763.2.304

Abstract : Tasks in the meat processing sector are physically challenging, repetitive and prone to worker scarcity. Despite the potential for automation, the inherent biological variation of meat and the commercial characteristics of the supply chain have limited the widespread implementation of automated systems. This chapter describes potential benefits and challenges, and gives an overview of some of the robotic and automation equipment available and in development for beef, pork and lamb processing.

Key words : meat processing automation, robotics, primal cutting, boning, trimming.

13.1 Introduction The benefits of automation are well understood and frequently adopted by many manufacturing industries. The food sector generally has been slow to capitalise on the opportunities, particularly in the primary production operations before pack-ing. The specific issues associated with automation for the meat sector are dis-cussed in the following sections.

13.1.1 Scope of chapter Butchery tasks are unpleasant, physically arduous and carry a high risk of worker injury. This suggests them as prime targets for the benefits of robotisation; how-ever, the skilled nature of the butchery task, combined with the biological varia-tion of the raw material, poses substantial challenges. This chapter considers the applications of robotics and automation in primary meat production processes in the abattoir and cutting plant for beef, sheep/lamb and pork meat.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3

Robotics and automation in meat processing 305


Automation in poultry and fish production are dealt with in other chapters, as are operations such as packing and palletising that occur after retail portion cutting.

13.1.2 Drivers for automation in meat production Robotics and automation have been relatively slow to permeate the meat produc-tion industry, and the majority of tasks are still performed manually. Whilst indi-vidual machines can be developed for skilled and complex tasks, their function always remains specific to the task for which they were developed.

Replacing a skilled slaughterman or butcher is difficult. For the more (seem-ingly) mundane tasks, such as putting lamb chops into packs, laying up sliced beef, handling ‘bundles’ of wafer ham, etc., the human is difficult to replace at a viable cost. Research projects have tackled the technical aspects of these problems for sev-eral decades, but commercially viable systems are only just beginning to emerge.

There are a wide variety of commercial and product quality reasons leading many companies to investigate robotics and automation applications on meat pro-duction lines. Ultimately all drivers to adopt automation have the same aim – increased profitability. If no profit or long-term benefit is foreseeable then no changes will be implemented. The use of automation in meat processing in place of human operatives has many potential benefits, which may be tangible, intangi-ble, social or economic. Many generic drivers are quoted to support the introduc-tion of automation including:

• Production quality : It is widely accepted that meat cuts best in the range 2–5°C, just above the initial freezing point. As the temperatures reduce, the cut quality improves, but cutting forces increase (Brown et al ., 2005 ) to an extent where human strength could be insufficient to maintain production rates. Automation can be used to exert higher forces, maintaining or improving on cutting quality and production rates. • Product consistency : Boredom, stress and tiredness are not an issue with auto-mated systems, typically performing a task more consistently than a human. This consistency can additionally permit efficiencies in other aspects of the business thus further aiding profitability. ‘Getting things right’ reduces waste and increases overall yield. • Added functionality : One key benefit of robotics is in performing tasks that the human cannot. Automation can make subtle adjustments beyond the skill of an operative, or be endowed with ‘superhuman’ sensory, recall, reasoning or other capabilities such as infrared detection, increased strength, X-ray vision, huge memory, etc. Machines can be designed to operate under conditions where humans could not perform effectively. This can allow processing in environments beneficial to quality, for example, sustained low temperatures, aseptic atmospheres, etc. • Worker safety : Injuries cause lost production and absence from work, not to mention costly compensation claims. The meat processing industry has a poor

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3


306 Robotics and automation in the food industry

safety record when compared to all types of manufacturing industry surveyed by the UK Health and Safety Executive (HSE). For 2005–2008, the meat and poultry sectors had a mean annual injury rate of 1313 per 100,000 employees compared to a mean of only 913 for all manufacturing industries (HSE, 2008). Injury occurs to both experienced and trained staff, illustrating that it is the nature of the work rather than inexperience causing the danger. Cuts made with high force towards the body, bad knife design and cold fingers contribute to the poor safety record (North, 1991 ). • Food safety : Foreign bodies and microorganisms can be transferred to foods from operatives. Replacement of potentially contaminating human labour by machine can reduce this risk. The costs of preserving hygiene with the large numbers of staff present in a normal meat plant increase the overall production cost. A number of studies of specific systems (Holder et al ., 1997; Clausen, 2002 ) suggest that automating and removing staff from the production process can improve the microbial condition of processed meat. • Legislation : The minimum legal continuous working temperature for a stand-ing, active labourer in the UK is 10°C (UK Factories Act, 1961). EEC directive 95/23/CE states that during cutting meat temperatures should not exceed 7°C, and the processing rooms should be at a maximum of 12°C. Automation and robotics can work closer to the optimum temperatures for meat processing than can be legally achieved with human operatives. • Difficulties in recruiting staff : There is a shortage of skilled labour for many of the tasks in the meat industry. The work is typically repetitive, physically intensive and takes place in an unpleasant environment. Many employers have substantial difficulties in recruiting and retaining useful staff. The continual recruitment and training introduces unwanted additional costs.

13.1.3 Barriers to introduction of robotics and automation into meat processing

A traditionally conservative, cash-poor meat industry with low margins has some fundamental financial, attitudinal and commercial challenges in imple-menting automation systems. Developments made during the last decade have removed or reduced many of the technological barriers to automation of meat production tasks. The predominant limiting factors are now related to business and commercial factors. Despite the advances, automation technology is still a long way from the generic robot-type system capable of replacing people in most food operative situations as envisaged by Khodabandehloo and Clarke ( 1993 ).

Commercial and organisational challenges Automated processing has been successfully implemented in the automotive industry where regular components and a high value product, coupled with relatively low production rates, make vehicle production an ideal process for

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



robotisation. Despite the product and process differences, some business experi-ences and observations can be transferred into the meat sector. A longer-term, less risk averse, company culture is required, and employees at all levels must be prepared to accept change. Where automation projects have failed is often in the lack of ‘buy-in’ throughout the company and lack of awareness of the skills and organisational changes required to support the implementation.

The same organisational risks apply to the food sector, with additional chal-lenges of high product variability and a constricting market structure. The low margin on most meat products reduces the finances available for investment and a marketplace dominated by major multiple retailers exacerbates the situation. The majority of labour in the food sector is unskilled, and thus sums saved by manpower substitution are low. Supply, demand and processing specifications are flexible, seasonal and regional.

Many meat processing plants currently lack the in-house skills to specify and support automated systems. The skills required stretch beyond the basic engineer-ing function of current mechanised lines into more complex aspects of automa-tion system specification, installation, support, maintenance and reconfiguration of the system to deal with changing production requirements. Management, and production staff working alongside the automated systems, need to understand the strengths and weaknesses of the equipment and adjust practices accordingly. The entire organisation, from cleaners to directors, has to embrace a positive ‘mindset’ to automation of traditionally manual operations. Inappropriate attitudes at any of many levels can cause automation projects to fail.

Technical challenges From an automation viewpoint, the complexity of meat production tasks should not be underestimated. Humans possess sophisticated, integrated sensory abilities with inbuilt reasoning and manipulation capabilities. The majority of tasks within meat production have evolved to utilise these inherent abilities. The human is excellent at evaluating situations and acting accordingly, while a typical machine system has a predestined function, and correction of only a limited number of possible perturbations can be incorporated into the design. An automated system to replicate even a small subset of human abilities can require very sophisticated systems integration.

Despite the advances in meat automation progress made in recent years, the greatest technical problem is still that of coping with the natural biological variation in the product. Variable products require variable production strategies and thus flexible processing methods. This has implications for sensing systems and system elements in contact with the meat such as fixtures, grippers and cutting tools. Many meat products are relatively delicate and can be damaged by inappropriate handling. These factors tend to exclude direct technology transfer from other industries.

The secondary technical challenge is in equipment longevity and suitability for food production environments. Hygienic and robust systems that can resist high-pressure wash down, hot or cold, and condensation can be designed and built, but

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



at additional cost and complexity. This further increases costs for implementation of automation for food production. A number of studies of specific automated meat production systems (Holder et al ., 1997; Clausen, 2002 ) acknowledge the automation production benefits but express concern over cleanability of the often complex equipment.

13.1.4 Current status of meat processing automation Pigs, cattle and sheep are all quadruped mammals and the basic operations for conversion of animals to meat are similar. There are some differences however. Cattle and sheep have their entire skin removed whereas pigs are de-haired. The details of the evisceration process are different for ruminant cattle and sheep and for omnivorous pigs. Pork and beef carcasses are split whereas smaller lamb car-casses are typically not split. All species have different cutting patterns to pro-duce different meat products and portions; these cutting patterns can vary between countries and regions, and seasonally.

Pork meat production is the most widely automated. Many of the key develop-ments have been made by the Danish Meat Research Institute (DMRI). Whilst DMRI have a stated goal of producing a virtually fully automated pork process, some operations such as shackling, sticking, gambrelling, veterinary inspection, final trimming and removal/separation of specific organs are not included in the plan (Clausen, 2002 ). This ambitious target can be attempted due to the coop-erative and nationally integrated structure of the Danish pork industry, research establishments and equipment producers.

Whilst the size and weight of the beef carcass suggests automated processes would be of benefit to reduce the physically arduous nature of the tasks, this carcass type has received relatively little automation research and development (R&D) effort compared to lamb and pork. The key challenge for automation is the large variation seen in cattle. Slaughter animals may be from a wide vari-ety of breeds, ages, type (bull, steer, heifer, cow, etc.) and range in weight from 200 to 1000 kg. The variations seen in other carcass types are substantially less. Mechanised processing aids, guided by human staff, have been in existence for many years, but the ‘Fututech’ Australian R&D programme (White, 1994 ) sought to develop the world’s first truly automated beef processing line. The system was developed through to a commercial prototype stage and designed for a minimum processing rate of 60 carcasses per hour. The system included a large number of automated or semi-automated modules that performed the majority of the slaugh-ter tasks. These modules included rectum clearing and bagging, aitch bone cut-ting, head removal, brisket cutting, evisceration and tail cutting.

Lamb and sheep farming and meat production form a major part in the econ-omies of New Zealand and Australia. Not surprisingly, the majority of automa-tion for these carcass types has originated in these regions. The Meat Industry Research Institute of New Zealand (MIRINZ) have developed a series of machines for sheep processing that use minimal sensing or compliance to adapt the motion or work piece. However, by rearranging the various tasks on the slaughter line

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



and by redeploying some labour to act as the ‘sensing’ or adaptation element, rel-atively simple machines for sheep and lamb meat processing have been success-fully developed and commercialised.

In the early 1980s researchers at MIRINZ developed an improved manual dressing system, later called the ‘Inverted Dressing’ system because the carcass is hung from the front feet for a proportion of the dressing process whereas on a traditional sheep production line carcasses were hung from the hind feet. This simple change reduced manning levels by 10–20% and achieved a throughput of 3200 carcasses per shift (Annan, 1982 ). By 1990 a typical sheep dressing line making use of all available technology developed by MIRINZ over the previous ten years required only 26 butchers. This is almost half of the manning that had been required for the traditional manual sheep production line ten years earlier.

The majority of slaughter, butchery and meat processing operations are cur-rently performed manually with simple tooling. Automation and robotics has much to offer the slaughter and meat processing industries, and whilst not wide-spread some automated systems are available and have been implemented. The next section describes meat processing automation for pork, beef and sheep/lamb operations.

13.2 Automation of carcass production processes before primary chilling

All slaughterhouses follow a similar sequence of operations to transform the live ani-mal into meat for consumption. After slaughter, inedible and many non-muscle parts (skin, hair, intestines, etc.) are removed to produce a carcass which is then chilled to reduce spoilage rates. After chilling, the carcass is typically subdivided into smaller sections (‘primals’) for ease of handling. These primals are then processed (boned, trimmed, cut, etc.) to produce retail joints and portions. This section deals with auto-mation of carcass preparation processes taking place before chilling.

13.2.1 Lairage Ideally all animals rest in the lairage after arrival at the abattoir. This allows them to recover from the stresses of transport and acts as a pre-slaughter process buffer. Maintaining low stress levels in the live animal is an important welfare issue, but also has a beneficial effect on meat quality. An animal experiencing stress will have physiological changes including changes in heart rate, blood pressure, body temperature and respiration. Several stress hormones are released into the blood stream that speed the breakdown of glycogen stored in the liver and muscles, cre-ating by-products of lactic acid and water in the meat carcass. In turn these can contribute to undesirable effects on the final quality of meat such as pale, soft, exudative (PSE) meat and dark, firm, dry (DFD) meat.

The enforced herding required to move the animals around the lairage increases stress further. An automated pig lairage where these movements are

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



performed gently and without human presence was developed in Demark in the 1990s (Madsen et al ., 2006 ). This equipment uses automated walls to gently herd pigs towards the slaughter raceway and is installed in several Scandinavian pork plants (Pigsite, 2009 ).

13.2.2 Stunning, sticking and killing The slaughter raceway leads animals from the lairage to the slaughter area. The activities of slaughter process are influenced by cultural, animal husbandry and occupational health and safety considerations. Errors have far reaching effects on animal welfare, meat quality and all downstream processes.

Two methods of killing are commonly in use: stun-and-bleed, or gas kill. Stunning is carried out with an electrical shock across the head, or concussive

pistol head blow, to halt brain function, and then a cut is made to the artery in the neck (sticking) to drain the blood. There are automated systems to convey animals to the stun station, most consisting of V-shaped conveyors to carry the animal to the stun operator. The current and voltage of the stunning shock is controlled, and varies for different animal sizes and species.

An alternative to stun-and-bleed is gas killing, whereby animals are immersed in a CO 2 environment that renders the animal unconscious before bleeding. This automated method is gaining popularity, particularly in pork production. The automated CO 2 stunning units operate like enclosed Ferris wheels, with multiple compartments rotating cyclically. Small batches of around six pigs are herded into each compartment. The compartment then descends into a deep well area filled with CO 2 , emerging on the opposite side to pig entry where the compartment tilts, and the animals slide down a chute to the shackling line below. Residence time is typically around three minutes in 82% CO 2 (Butina, 2010 ). Whilst gas stunning can produce higher quality meat (Channon et al ., 2002 ), there are some concerns for animal welfare (Grandin, 2008 ).

Beef stunning and sticking processes are ergonomically difficult to perform manually because of the size of the animal. Food Science Australia, a joint venture organisation of Australia’s Commonwealth Scientific and Industrial Research Organisation (CSIRO) and the State of Victoria, has investigated auto-matic systems for these tasks. This system is based largely on the Fututech mod-ule, in which a machine vision system was used to determine correct stun and sticking locations. The Fututech slaughter module separated one animal from a group of cattle using a moving floor conveyor that transferred the animal to a moving conveyor between the animal’s legs as the floor dropped away (White, 1994 ). Two bails captured the neck and applied an electrical current to stun the animal. The electrical pathway was then altered to effect a spinal inactivation. A pneumatically powered knife with oscillating blades was used to enter the thoracic cavity and sever the aorta. Horns were also removed at this stage using hydraulic cutters.

A prototype automated sheep stunning machine was developed in the mid 1980s in New Zealand (Authier, 1990 ). This machine was quickly commercialised

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



and is commercially available. For a variety of reasons including ritual slaughter, automated sheep sticking systems have not been successfully developed to date.

13.2.3 Shackling Once stunned, animals are manually shackled, usually with a chain loop around one hind leg, and hoisted to hang head down. A human operative then makes the ‘sticking’ cut to the artery in the throat to drain the blood. These shackling and cutting operations are complex and difficult to automate due to the complexity of the operations, the unstructured environment, the implications on downstream processes if performed incorrectly, and the need to maintain animal welfare if stunning fails.

13.2.4 Removal of hair or hide Pork carcass production differs from beef and sheep carcass production in that typically hairs are removed from the skin, whereas beef and sheep plants remove the entire hide/fleece from the carcass. The removed hide has value as a base product for the leather industry. Whilst it is usual to remove only the hairs from pork and leave the skin on the carcass, some pork plants also perform de-hiding for pig leather, although this is not a common practice.

De-hairing pork Once drained of blood, pork carcasses pass through a sequence of mechanised operations, typically consisting of first a hot water or steam scald to loosen hairs, and then through a de-hairing machine where rotating metal-tipped rubber fingers scrape and brush most of the hairs from the carcass surface. This is followed by a singeing operation, whereby the carcass passes through gas flames to burn off remaining fine hairs. Finally, carcasses pass through a second ‘polishing’ station, where burnt hair stubs are removed by rotating rubber flail fingers.

These operations avoid the need to adapt to carcass geometries by using tech-niques that conform to the product shape. Fingers on flexible rubber mounts, gas flames and water jets can all act on the carcass without detailed knowledge of surface position. This approach allows simple mechanisation to be used for these tasks.

De-hiding beef The first task of beef de-hiding is to cut the hide along the belly from the crotch to the neck. This is a demanding task requiring a consistent cut typically 2 m or more in length, along the centre line of the carcass severing only the skin. Industrial Research Ltd (IRL), based in New Zealand has developed automation for this task (Templer et al ., 2002 ). The profile of the belly is detected with an infrared laser distance sensor, and this information is processed to form a smooth trajectory for the cutting tool. The purpose-designed tool consists of a guidance spike mounted tangentially to a rotating circular knife. The spike protects the

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



underlying meat from cutting damage and serves as an anvil to improve the cut-ting efficiency. The tool is moved by robot to place the spike between the skin and meat and then follow the previously determined path to sever the hide along the belly. The system has been proven in a slaughterhouse in Nebraska, successfully cutting many thousands of carcasses. Whilst the initial development work used a purpose-built robot, later commercialisation work used an off-the-shelf food grade KUKA robot.

Before this hide opening cut is made on feedlot cattle, there are often large ‘dags’ or deposits on the skin that must be removed. In 2000, Food Science Australia staff developed a hand-held dedagging tool. A later MLA project (MLA, 2012 ) sought to automate this process using a robot. The project was not success-ful due to problems restraining the carcass while the robot was operating.

After the skin-opening cut is made, the hide is removed or ‘pulled’. Mechanical pulling arms supply the majority of the effort, but a human butcher is required to make specific preparatory cuts, attach the pulling mechanism and make assisting cuts during the pulling operation.

The Fututech system used bed-dressing for hide removal where the carcass was resting on its back (White, 1994 ). After appropriate manual hide preparation the carcass was suspended from four hooks, one in each hock, while remaining in the supine position. The hide was removed automatically using a 3-stage pro-cess that involved (1) pulling the hide downwards, (2) separating the hide from the back fat using a blunt knife and (3) pulling the hide over the head and off the carcass.

De-fleecing sheep The sheep de-fleecing, or pelting, process is complex and traditionally used 30% of the labour force on a sheep dressing chain (Longdill, 1984 ). Early attempts to automate this process were reasonably successful although the machinery was complex (Robertson, 1980 ). Researchers at MIRINZ developed a rotary pelting machine that automated the majority of the pelting process. The machine was physically large and operated on a rotary turret principle to achieve the required throughput. Commercial versions of this machine were installed in a number of sheep processing plants in New Zealand during the 1980s; however, none are operating now, as they were superseded by the technologies described below.

Sheep pelting starts with the ‘Y-cut’, whereby initial incisions are made on the forelegs and chest. A robotic Y-cutting system was developed at IRL in New Zealand (Taylor, 1993 ) in the 1990s and this is now operating in several meat plants in New Zealand.

Brisket clearing is another pelting subprocess most slaughter staff consider physically difficult that has been automated in New Zealand researchers. This is another example of a human operator being assisted by a relatively simple but powerful machine. The operator performs all the sensing and delicate positioning operations, which are not physically difficult, while the machine can reliably and repeatably perform the difficult and physically demanding tasks without tiring or becoming injured. A similar approach is used for the sheep-fleece shoulder puller

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



with a combination of skilled labour for sensing and positioning, and a machine performing the heavy work. The last in the series of pelting machines developed by MIRINZ is the final puller. This machine is deceptively simple in its operation, although its design and set-up is the key to its successful operation. This machine was released commercially in 1985 and has been installed extensively on lamb chains around the world.

13.2.5 Evisceration and dressing Once the hairs or fleece/hide have been removed, all carcasses are eviscerated, whereby the internal organs are removed. The viscera generally separate into edible (heart, liver, kidneys, etc.) and inedible (intestines and bowel). There are substantial hygiene implications of error when removing the inedible organs as the contents contain faeces and pathogenic organisms that can contaminate the carcass meat if spilt.

Although all processes follow the general sequence of rectum loosening, belly opening, and viscera removal, there are substantial detail differences for exam-ple, trimming, washing, additional cutting, other organ removal, other processing, etc., between species and plants.

Pork evisceration Work at DMRI in the 1990s ( Fig. 13.1 ) and a later collaboration with SFK has developed automation for pork evisceration (Madsen and Nielsen, 2002 ).

The equipment makes a few simple anatomical measurements that guide the process. These gross measurements allow for coarse positioning of the evisceration automation, and conformation of the flexible carcass or adaptive tooling is also used to reduce complexity and hence increase reliability of a rel-atively complex operation. A prepared carcass is automatically clamped open and an arm moves the viscera to expose the sternum, where a second set of arms loosens the leaf lard and severs the attachment of the diaphragm to the chest cavity wall. A back cutter is then moved into the carcass to penetrate the diaphragm adjacent to the spine and sever the connective tissue between the organs and spine in the hind section of the carcass. A tenderloin tool then moves in opposition to the leaf fat arms to separate organs from the thoracic cavity. Activating these automation motions concurrently resolves forces in the system and removes the need for forceful carcass clamping and fixturing. The released organs are then pulled forward out of the carcass with a horizon-tal movement of the tenderloin tool, the clamps are released and the carcass is moved out of the supports. The automated system ensures tools are washed before the next carcass arrives. This automated evisceration system performs all these operations in 10 s giving a line speed of 360 carcasses per hour. DMRI are currently working on equipment for the subsequent separation and sort-ing of the organs. Microbial analysis has shown that carcasses automatically eviscerated possess fewer pathogens ( E. coli ) and aerobes than conventionally eviscerated carcasses (Clausen, 2002 ).

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



Beef evisceration Once the cattle hide has been detached, the abdominal cavity is opened and the organs removed. Part of the opening process involves sawing the sternum bone to gain full access to the chest cavity. IRL in New Zealand have been work-ing towards automating this beef task (Templer et al ., 2000 ). Through projects running over a number of years, the team have demonstrated first static, then line-synchronised, brisket sawing. Using the same robot and guidance system as the hide opening system, a reciprocating bone saw similar to, but more powerful than, a manually manipulated brisket saw, is moved down the centreline of the sternum. However, when implemented on a production line the equipment did not perform satisfactorily, as a large number of the carcasses had been damaged in the previous de-hiding process. This resulted in a twisted carcass at the sternum saw station. The automation was unable to cope with straightening the carcass and completing the cut in the 9 s cycle time available.

The Fututech system used an automated evisceration system comprising a pad-dle that was pushed against the spine and pulled down the carcass to peel the vis-cera from the abdominal cavity and push it into the viscera tray for sorting (White, 1994 ). This mechanised approach relied heavily on compliance of the carcass and organs to a fixed trajectory.

Fig. 13.1 Developmental pork evisceration system.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



Sheep evisceration Researchers at MIRINZ developed a mechanical sheep evisceration system in the late 1980s (Authier, 1990 ). The system comprised a brisket and belly cutter, an eviscerator and offal handling system. The system was trialled in an Australian sheep processing plant and offered to several companies for commercialisation (Authier, 1994 ). Later trials using a variant of the pelting Y-cut robot were aimed at opening the brisket and the belly in a more conventional chain configuration. The technology has yet to be commercialised.

13.2.6 Splitting Pork and beef carcasses are generally split into right and left sides to ease han-dling and increase rates of chilling. Lamb and sheep carcasses are typically left unsplit.

Automatic carcass splitting equipment has been available for many years. Systems are sold by such suppliers as Stork, SFK, Danfotech, Durand, Automeat, etc. These machines have a range of cutting actions and complexities. The basic systems use a simple downwards motion of a circular saw through the space where the carcass should be. A higher level of complexity uses a series of rollers to locally position the spine onto the cutting device.

‘Back finning’ is sometimes carried out as part of splitting for pork carcasses. This process reduces damage to the eye-muscle during the splitting operation by separating it from the dorsal spine ‘fins’ before splitting the carcass. An automated system using a relatively complex arrangement of rotary knives, plain blades and active rollers has been developed for this task in the Danish pork industry.

Automation for beef splitting was among the first examples of mechanisation in the slaughterhouse, and many equipment manufacturers now include beef split-ting machines in their product range. Whilst this equipment removes the arduous manual process, many users of the equipment are still dissatisfied with its perfor-mance in terms of accuracy of splitting down the centre of the spinal column and the hygiene aspects associated with deposition of bone dust and other detritus on edible surfaces of the carcass.

The Fututech system included a module that automatically split a beef car-cass into two sides using a guided bandsaw (White, 1994 ). Later work by Food Science Australia funded by Meat & Livestock Australia (MLA, 2012 ) used a robot to guide a band saw for carcass splitting. This equipment had a vertebrae sensing system based on ultrasound and this experienced difficulties on some carcasses due to voids caused by the hide puller disrupting the consistent passage of the ultrasonic wave necessary for ultrasound sensing.

Most splitting equipment producers claim an increased accuracy of auto-matic carcass opening and splitting over human-based splitting operations. However, the experience of some users is that there is still deviation from the precise centre line of the carcass. This can cause problems for carcass inspec-tion and subsequent automated systems using the spine as a reference or datum position.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



13.2.7 Automated inspection and grading Farmers and animal growers are typically paid based on weight and conformation (muscularity) of carcasses. Impartial, accurate and reliable automated grading automation has been the subject of much development activity for all species.

Automated carcass weighing systems are common on most slaughter lines. Automatic grading and classification systems typically compare image(s) of each carcass against standard reference carcass images for the various grades. This is impartial and removes variation due to individual graders. The captured image can be stored and used for traceability, production management or process quality audit.

For pork, DMRI have developed the Danish Carcass Classification Centre, and SFK produce automated grading system called AUTO-FOM (Madsen and Nielsen, 2002 ). Machine vision systems that are non-invasive are in development, but some studies show them to be less accurate in predicting saleable yield than existing technology (McClure et al ., 2003 ).

Whilst laboratory development systems show the potential for rapid, economic, hygienic, consistent and objective assessment systems, there are still limitations in the industrial environment (Brosnan and Sun, 2002 ).

13.2.8 Automated chill rooms Certain wavelengths of visible light can reduce shelf life and encourage rancidity of stored chilled meat (Field, 2004 ). Automation to move carcasses in darkened chill rooms could improve product quality through reducing a contamination route from the human operative to the meat and reducing the spoilage organism growth rate. This type of automated carcass loading and unloading system has been com-monplace in the New Zealand sheep meat industry for the last 20–30 years.

13.3 Automation of carcass separation processes after primary chilling

All slaughterhouses follow a similar sequence of operations to transform the live animal into meat for consumption. After slaughter, inedible and many non-muscle parts (skin, hair, intestines, etc.) are removed to produce a carcass which is then chilled to reduce spoilage rates. After chilling, the carcass is typically subdivided into smaller sections (‘primals’) for ease of handling. These primals are then processed (boned, trimmed, cut, etc.) to produce retail joints and portions. This section deals with automation of carcass preparation processes taking place after chilling.

13.3.1 Primal cutting After chilling, carcasses are commonly cut into smaller ‘primal’ sections that are then further subdivided into retail joints, boned out, or processed into a wide vari-ety of end products. Automated and robotic systems to produce primals are of

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



major importance for meat processing because relative price differences between primals requires accurate and consistent control of cut paths to optimise overall carcass value.

Pork primalisation Early work on robotic systems for pork primalisation was performed in Western Australia (Clarke, 1985 ). The system comprised a computer-controlled pork car-cass break-up machine that automatically broke down a full carcass into eight pieces in less than 30 s.

Simple automated cutting systems that separate a pork half carcass into fore, middle and hind sections were developed in Europe by DMRI and others in the early 1990s. The tenderloins, head and forefeet are manually removed as prepa-ratory operations, then carcasses hanging on a standard gambrel are pulled across a conveyer belt and the hind feet cut off. This releases them from the gambrel onto the conveyor. At a second station each carcass side is moved against a datum surface and the length between the pubic bone and the foreleg is measured. This measurement is used to position circular saws further down the line to anatomi-cally derived cut positions for that carcass side. A second machine is available for the longitudinal cut to separate the belly from the loin.

A robotic solution that performs all cuts in a single system was developed in the early 2000s as part of an EU funded project (Purnell, 2004 ). The Advanced Robotic Technology for Efficient Pork Production (ARTEPP) system has been patented and is arguably the most advanced robotic meat production system avail-able to date. Because of the need for accurate cut placement, compliance of the carcass is not used and each cutting path is specifically adapted to the individual carcass being processed. Significant interaction between various expert organi-sations in cutting-blade design, machine vision, robots, systems integration and meat production were required for the project to be successful. This project is described in more detail as a case study below.

Case study: Development of the ARTEPP pork primalisation robot In initial R&D studies a purpose-built Cartesian food grade robot wielding a pneumatic cutting tool was used to make the cuts. Whilst successful cutting was demonstrated, several factors limited the industrial exploitation potential of the system. The gantry-based Cartesian robot was very large, did not withstand the rigours of the food production environment, and spare parts and engineering sup-port were not readily available. The pneumatic cutting tool was prone to stalling at the high cutting rates possible with robotisation, thus negating some of the potential benefits. This illustrates some of the fundamental differences between human and machine performance of the same task. The physical strength of the slaughterman regulates the cutting process; as the cutting forces build up, the human slows, allowing the tool to make the cut. With a higher strength robot making the cut, the separation made by the pneumatic tool could not keep pace with the rate the robot was moving the tool along the path. Using a more powerful 3-phase electric saw and developing a special blade for high-speed cutting of meat and bone reduced these difficulties.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



A second-generation ARTEPP system used off-the-shelf 6-axis KUKA KR125 anthropomorphic robot to avoid development time and costs within the project, and to produce a system with readily available spare parts and engineering support. A modular approach was taken with the carcass transport, orientation and support subsystem operating independently of the sensing, cut-path derivation and robotic cutter subsystems. This allowed standard subsystems to be used in different installa-tions, with customisation only required in a few subsystems, thus reducing costs.

In one implementation of the ARTEPP system, incoming pork sides are ori-entated to align the carcass split plane to the overhead rail. At the orientation station, use is made of a previous processing line feature in that the hook through the Achilles tendon always faces the split plane. An inductive sensor detects the hook and the side is rotated if required to place split plane facing the robot and vision measurement system. The side then indexes on to the cutting station where adaptive fixturing grasps the side and supports the carcass to resist cutting forces. Once clamped, structured light techniques are used to extract 3D locations of ana-tomically pertinent carcass features. Cut paths are defined anatomically relative to these features for every carcass separately. After cutting the clamps are released and the side ejected from the system.

In another configuration of the system, where fewer cuts per carcass are required but at higher line speeds, an overhead rail and inclined conveyor carry each side past the vision sensing and cutting stations at a fixed speed. The image processing is performed as the side travels to the cutting station and the robot is synchronised to the line speed allowing carcasses to be cut whilst moving.

Comparing the robotic system with manual primal cutting showed manual cut placement was within 20 mm of the anatomically desired location and 89% of manually cut carcasses had cut accuracy of better than ±5 mm. The robot system performed to better than ±5 mm for 97% of cuts ( Fig. 13.2 ).

The system can tirelessly produce consistent, anatomically accurate cuts. However, the most important commercial feature of the automated primal cutting system is its ability to finely adjust cuts in response to seasonal and market price

Frequency

Nominalcut position

Lower value primalHigher value primal

Automated cutting

Manual cutting

20 mm 20 mm5 mm 5 mm

Fig. 13.2 ARTEPP automation versus human cutting performance.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



fluctuations. Shifting cuts to favour high value primals (moving the ‘automation’ peak to the left in Fig. 13.2 ) can result in significant value improvements for each carcass. Other cost benefits connected with not having to find, train and retain human staff for the task are a bonus.

The latest system ( Fig. 13.3 ) has been used online for a full year in a Norwegian plant doing the work of three staff with a 3% yield improvement. This produces a payback in less than 18 months for equipment costs.

Beef primalisation Mechanical boning aids that exert pulling forces while a human butcher makes key separation cuts have been used for many years. Whilst not at the forefront of automation technologies, these human augmentation systems have enabled higher throughputs with less physical effort for the same number of staff than using tra-ditional individual cutting tables (Field, 2004 ).

French researchers at the Institut national de la recherche agronomique (INRA) developed a prototype robotic system for subdividing beef forequarters (Damez and Sale, 1994 ). The system is relatively slow because major sensing and trajec-tory planning problems had to be solved. The prototype was successful in produc-ing beef primals, but has yet to be developed into a commercial system.

An ambitious beef sectioning system was proposed by the Texas Beef Group in a 1993 patent (O’Brien and Malloy, 1993 ). A chilled eviscerated carcass would be mounted horizontally on an automatic guided vehicle and appraised using X-rays, 3D machine vision and ultrasonic sensing. The results of the inspection would be used to generate cutting paths to enable the carcass to be cut into optimal primal sections. A robot would be used to effect this separation with high-pressure water, abrasive and air jets. Flesh would be cut with the water jet while the air jets would

Fig. 13.3 Production ARTEPP system.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



keep the severed meat clear of the cutting area. The abrasive jet would be invoked when making cuts through bone. This is a particularly high-tech proposal in a pat-ent and no record can be found of a prototype or commercial system.

Sheep primalisation automation Up until the 1970s, most of the sheep and lamb traded internationally was in the form of frozen whole carcasses. After that time, carcasses began to be progres-sively broken down into a series of cuts, initially frozen and later chilled and vacuum packed. The main items of technology in a lamb boning room in the early days were band saws and packaging machines.

Research at New Zealand’s AgResearch Crown Research Institute in the late 1990s was directed at the automation of cutting lamb carcasses into primals. The drivers were to produce clean, square cuts and hygienic handling of primals, to improve product yield and shelf life from subsequent processing. A prototype machine was produced during 2000 and further enhanced the following year by automatically locating bones within carcasses.

13.3.2 Boning Automation and dedicated machinery for boning out of specific pork sections are commercially available or under development in many parts of the world. Much of this work has been led by DMRI and their commercial partners and includes boning equipment for fore-ends and hindlegs, and combined boning and trimming equipment for belly and loins (Madsen and Nielsen, 2002 ).

Robotic technology has also been used to separate pork flank ribs from the pork belly (Anon, 2000 ). The system uses a machine vision system to assess the size and shape of an incoming belly. The 3D data are used to calculate the cutting path. A Fanuc M710 robot equipped with a curved double-edged ‘Denver knife’ executes the path, pulling the shaped knife through the belly in the prespecified trajectory. The robotic cell includes automatic tool changeover and can select from eight different knives. When not in use knives are sterilised as part of the production process. This system can process 1400 bellies an hour, equivalent to a six-man crew. Final trimming and manpower requirements are reduced and the yield is optimised over both the belly and flank rib set.

Automated systems for deboning specific beef meat sections are under devel-opment or in production, as with the other meat types. A beef rib deboning system has been designed and manufactured by Food Science Australia. This machine automatically strips the meat from a beef ribset in 21 s. Longdell ( 1996 ) describes other beef deboning machines for heads, loins and forequarters. All systems improve carcass yield, but the levels of sensing and adaptive automation are low, with most systems separating meat from bone with combinations of mechanical force, compliance in the equipment, different rigidities of meat and bone in the beef section, and bespoke shaped blades (Trow and Ng, 1994 ). A human butcher is required to operate the equipment and assist cutting in a similar manner to the primalisation pulling arms mentioned above.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



A vision guided, force-feedback controlled beef deboning research system has been constructed at the University of Bristol (Purnell et al ., 1993). The laboratory based system ( Fig. 13.4 ) demonstrated the technical feasibility of sensory guided robotic deboning, but further R&D would be required to bring the concept to a commercial reality.

The technique made an initial 2D visual assessment of the beef joint, and sought to match that current meat section to a database of previous experience. If a match was found, the previous cut paths were replayed for the current meat section; if no match was found then force feedback from the boning blade was

Fig. 13.4 (a) Cutting sub-system for beef foreleg deboning research equipment. (b) Beef

foreleg deboning research system showing machine vision and lighting.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



used to guide the robot along the bone and in doing so create another experience example to augment the database. This process showed promise for the 2D debon-ing of beef forelimb taken as the example process. However, these initial concepts would need to be extended substantially to produce a fully automated beef boning line for commercial use.

Research on automated sheep boning systems began at MIRINZ in the early 1980s as part of the mechanical boning project (Roberts, 1984; Wickham, 1988 ). The automated sheep boning process consisted of the following processes:

1. The load station lifted the carcass off the rail, removed the gambrel and loaded the carcass onto the carcass support.

2. The pedestal rotated the carcass support about the horizontal axis to present it to the boning head.

3. The linear drive cleared the pelvis by grasping and pulling the rear legs on the upward stroke.

4. On the downward stroke, a combination of rotating knives, flexible disks, ploughs and a moving wire separated the soft meat sides from the skeletal frame.

5. The skeletal frame was ejected at the pedestal during rotation of the carcass support.

A programmable logic controller and range of sensors controlled the sequenc-ing and the compliance was used in places to adapt cutting paths to carcass profile and deform the carcass to the blade trajectory. The production rate was estimated at 190 carcasses per hour with a payback period of less than 1 year. However, this machine was never commercialised. The frame boner laid the groundwork for a very successful second-generation boning machine (Wickham, 1990 ). A carriage was manually loaded with a loin section of sheep carcass, and then transported through a set of fixed knives followed by a set of semi-rigid plastic ploughs. The frame for the knives and ploughs could move vertically to partially accommodate different loin sizes.

Further commercial machines to come out of the MRINZ mechanical bon-ing programme were the chine and feather bone removal machine, rib frenching machine, shoulder fleecing machine and the shoulder boning machine (Ng, 1992 , 1994 ; Wickham, 1992 ).

A trunk boning machine and a leg boning machine operating on similar prin-ciples are commercially available (Macpro, 2010 ). In the trunk boning machine the trunk is conveyed away from the operator after manual loading where two blades clear the meat from the vertebrae approximately 50 mm either side of the centreline. The fleecing blades sweep around the ribs to separate the meat, while a second set of knives simultaneously clears tissue from the neck. In the leg bon-ing machine the leg is placed vertically between shaped metal boning chucks. The two chucks move towards each other, boning the leg using a scraping and cutting action until the two chucks meet. The bone is finally ejected through the lower chuck.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



Scott Automation, in association with meat processor PPCS, has developed a robotic system for boning lamb legs (Templer, 2004 ). A KUKA robot is been fit-ted with a boning knife incorporating force feedback allowing the robot to guide the knife along the bones of the lamb leg.

13.3.3 Slicing and portioning The development of slicers illustrates the importance of sensing for benefits of automation, particularly in slicing for multi-slice packs where control of individual slice weight is desired. Initially meat slicing was carried out by hand, as this was the only method possible, then mechanised fixed slice thickness machines with much higher throughput rates came to the fore. Further reductions in giveaway were gained with the first generation of automated slicing machines that changed thicknesses by scaling a slicing pattern from input meat section weight. This gave advantages over fixed-increment mechanised slicing. However, because of the inherent variations in weight–length ratio of nearly all meat sections some give-away losses were still apparent. The latest 3D scanning slicing machines make the next step with greater sensor data to improve performance further. A full 3D representation of input meat section is gained using a laser stripe or similar tech-nique, which allows for better utilisation and less giveaway. However, the slice angle is often constant, limited by the mechanical arrangements of the equipment to maintain production rates. There is a trade-off between machine complexity, line speeds and yield attainable.

13.3.4 Trimming People are becoming increasingly health conscious and consequently there is a grow-ing demand for lower fat products. In traditional manual trimming each individual operative makes the decision on how much fat to remove and then makes the cut using a standard butcher’s knife. This often results in straight cuts to trim fat from a curved surface and a high degree of variability between individual trimmers.

Water-jet based trimming machines have been on the market for some time but high capital costs requires relatively high throughputs, which consequently suit only higher volume production. Many meat producers are too small to afford and benefit from this trimming technology. Additionally, there are specific problems with meat section geometry for some sections. Water-jet cutting is most commonly used on ‘planar’ food sections, such as beef steaks, pork chops, chicken breast fil-lets, pork bellies, etc., that are laid flat on the input feed conveyor and remain sta-ble in this position through sensing and cutting. For some sections, such as lamb or pork chops, the narrow ‘tail’ of the chop can twist under its own weight to lie flat on the conveyor, thus preventing effective trimming.

Research at FRPERC (Purnell and Brown, 2004 ) developed specific equip-ment for trimming lamb chops. Machine vision provided fat thickness profile along the length of each chop. The system then conformed the meat section to

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



place the fat–lean interface at the desired fat thickness from a fixed cutting path. The cutting path was thus simplified and could be made with basic tools and an uncomplicated trajectory ( Fig. 13.5 ).

After the cut is complete, the meat section is released and returns to its natural shape but with a uniform covering of fat over the fat–lean interface ( Fig. 13.6 ).

Industrial plant trials with the demonstrator system have achieved improved accuracy and product appearance over manual trimming.

13.4 Future trends Until relatively recently, technical issues associated with adapting sufficiently accurately to the inherent biological variations in meat limited the potential for

Required cut path at fixeddistance from fat–lean interface

Conforming wall tochange shape of chop

Clamp forceholding chopagainst wall

Simple pathfollowed by cutter

Fig. 13.5 Uniform fat trimming concept.

Fig. 13.6 Trimmed lamb chops.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



automation in the meat processing sector. Many of these technical issues have now been resolved and demonstrated, and commercial business issues will be the major influence on implementation and uptake of meat processing automation in the near future. Yield control, legislation, difficulties in staff availability will increase commercial pressures and encourage more meat processor organisations to automate, simply to maintain throughput.

The economic breakeven point for implementing automated slaughter lines will be affected by an increasing cost of not automating, coupled with a decreasing cost to automate. As automation levels rise, the staff skill levels will rise accord-ingly. As the pressures imposed by the global marketplace, regulatory agencies, distribution channels, media and customers increase, slaughterhouse staff with profiles closer to surgeons, and skilled automation engineers by their education, training and working habits will begin to emerge.

Due to development costs, new automation meat systems are expected to arrive in a piecemeal manner, rather than as major projects addressing the entire slaugh-ter line. However, as a result more automation subsystems will become available reducing one current barrier of technology cost. These pockets of automation will have significant impact in small areas in their specific roles, but widespread auto-mation will not occur immediately. Pork slaughter automation is a possible excep-tion with fully automated lines expected in next decade.

13.5 Conclusion The development of automated meat processing systems has received substan-tial effort and investment, but uptake has been limited by technical and business issues. With improved technology and reducing production costs for automa-tion, it has become increasingly possible to overcome these limitations. The potential advantages and rewards to the meat industry have resulted in a con-siderable number of process-specific applications and continue to drive R&D of more sophisticated systems. Despite the differing operations for beef, pork and sheep meat production, some general trends are common across a number of projects.

Initially many meat automation research projects developed bespoke robots for their particular task (Maddock et al ., 1989 ; Wadie et al ., 1995 ; Taylor and Templer, 1997 ; Templer et al ., 2002 ). In these projects as the developments neared com-mercialisation, the teams changed direction to using standard industrial robots, protected against the rigours of the food production environment. In conjunction with, and in some cases as a result of, these developments, most robot manufactur-ers now supply off-the-shelf food-grade robots. This in turn provides food sector appropriate subsystems for equipment integrators, easing development, operation and maintenance of robotic meat processing systems (Ranger et al ., 2004 ).

Food companies that have been successful in introducing automation tend to have good working relationships between all grades of staff and have longer-term financial viewpoints.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



A modular approach has been proved worthwhile at both process and individual task levels. DMRI seek to automate all pork production tasks through developing a series of modular components, each performing a different task in the slaughter process. This has allowed a number of different projects and partnerships to be established, leading to more flexibility in implementation for both the automation user and supplier. The ARTEPP primal cutting system uses modular subsystems to accommodate variations in plant specific processes. The series of lamb boning machines developed by MIRINZ use similar mechanised approaches.

Some automation systems have been successful because they perform tasks currently not possible for a human operative. A human butcher could not perform the multi-armed cutting and handling operations achieved by evisceration auto-mation. Even the strongest, most skilled butcher cannot match the consistency and high-force cut accuracy achieved with automated primal cutting. Automation of these types of tasks, unperformable by a human, are often the first to exhibit an acceptable cost–benefit ratio. Currently it is mostly uneconomic to replace a slaughterhouse operative with automation unless the automation yields additional benefits. Some successful projects have demonstrated an improvement over man-ual labour in terms of speed, consistency, accuracy and control.

13.6 Sources of further information and advice ABB robotics: http://www.abb.com/robotics Danish Meat Research Institute (DMRI): http://www.dti.dk/services/danish-

meat-research-institute/31729 Food Refrigeration and Process Engineering Research Centre (FRPERC):

http://www.frperc.com, http://www.grimsby.ac.uk/Industry/FRPERC.php KUKA robotics: http://www.kuka-robotics.com Meat and Livestock Australia: http://www.mla.com.au Meat Industry Research Institute of New Zealand (MIRINZ): http://www.

agresearch.co.nz/mirinz SFK systems A/S: http://www.sfk.com

13.7 References Annan , D. 1982 . MIRINZ manual dressing system. Proceedings of the 22nd Meat Industry

Research Conference , Hamilton , New Zealand , 37–39. Anon . 2000 . Slaughterhouse slashes costs and pork bellies . Advanced Manufacturing , 2

(1), 43 –44. Publ. Clifford Elliot Ltd, Burlington, Ontario, Canada. ISSN 1481–8354. Authier , J.F. 1990 . Mechanical dressing developments. Proceedings of the 26th Meat

Industry Research Conference , Hamilton , New Zealand , 225–231. Authier , J.F. 1994 . Lamb evisceration by machine. Proceedings of the 28th Meat Industry

Research Conference , Auckland , New Zealand , 187–191. Brosnan , T. and Sun , D-W. 2002 . Inspection and grading of agricultural and food products

by computer vision systems – a review . Computers and Electronics in Agriculture , 36 , 193 –213.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



Brown , T. , James , S.J. and Purnell , G. 2005 . Cutting forces in foods – experimen-tal measurements . Journal of Food Engineering , 70 (2), 165 –170. doi:10.1016/j.jfoodeng.2004.09.022

Butina . 2010 . http://www.butina.eu (Accessed August 2010). Channon , H.A , Payne , A.M. and Warner , R.D. 2002 . Comparison of CO 2 stunning with

manual electrical stunning (50 Hz) of pigs on carcass and meat quality . Meat Science , 60 (1), 63 –68.

Clarke , P.T. 1985 . Automatic break up of pork carcasses. Agri-Mation 1, Proceedings of the Agri-Mation 1 Conference & Exposition . ASAE Publication . 01-85, 183–189.

Clausen , V. 2002 . Automation in the pork industry. In Proceedings of The Nordic Veterinary Congress , Helsinki , Finland , 27–29 November 2002.

Damez , J.L. and Sale , P. 1994 . Studies on automation of cutting of the forequarter of cattle carcasses (In French) . Viandes et Produits Carnes , 15 (4), 103 –107.

Field , M. 2004 . Evolution in practice . New Food , (Issue 4 ), 41 –43. Grandin , T. 2008 . Effect of genetics on handling and CO 2 stunning of pigs. Meat Focus

International , July, 124–126. (with May 2008 updates). http://www.grandin.com/humane/meatfocus7-92.html (Accessed August 2010).

HSE . 2008 . http://www.hse.gov.uk/food/slaughter.htm#a1 (Accessed September 2012). Khodabandehloo , K. and Clarke , P.T. 1993 . Robotics for Meat Fish and Poultry

Processing . Blackie Academic Press . ISBN 0 7514 0087 4. Longdell , G.R. 1996 . Recent developments in sheep and beef processing in Australasia .

Meat Science , 43 , nS 165 – nS174 . Longdill , G.R. 1984 . Advances in Mechanical Dressing Technology. Proceedings of the

23rd Meat Industry Research Conference , Hamilton , New Zealand , 127–130. Macpro . 2010 . http://www.macpro.co.nz/mutton.html (Accessed August 2010). Maddock , N.A. , Purnell , G. and Khodabandehloo , K. 1989 . Research in Application of

Robotics to Meat Cutting. Proceedings of 20th International Symposium on Industrial Robots (ISIR) . Tokyo , Japan , 4–6 October, 957–963.

Madsen , K.B. and Nielsen , J.U. 2002 . Automated meat processing. In Meat Processing: Improving Quality , Kerry , J. , Kerry , J. and Ledward , D. , eds. Woodhead Publishing Ltd , Cambridge, England . ISBN 1-85573-583-0.

Madsen , N. , Nielsen , J.U. and Mønsted , K. 2006 . Automation – the meat factory of the future. Proceedings of 52nd International Congress of Meat Science and Technology (ICoMST) , Dublin , Eire. 13–18 August. ISBN-10: 90–8686-010-9 / ISBN-13: 978-90-8686-010-4.

McClure , E.K. , Scanga , J.A. , Belk , K.E. and Smith , G.C. 2003 . Evaluation of the E+V video image analysis system as a predictor of pork carcass yield . Journal of Animal Science , 81 , 1193 –1201.

MLA . 2012 . http://www.redmeatinnovation.com.au/innovation-areas/new-technologies/beef-slaughter/products. (Accessed September 2012).

Ng , W.Y. 1992 . Advantages on machine boning of lamb. Proceedings of the 27th Meat Industry Research Conference , Hamilton , New Zealand , 322–326.

Ng , W.Y. 1994 . Lamb shoulder fleecer and rack frenching machines. Proceedings of the 28th Meat Industry Research Conference , Auckland , New Zealand , 201–208.

North , S. 1991 . Public health and safety aspects. DTI Seminar: Robotics for the meat pro-cessing industry. University of Bristol, Bristol, UK, 27 September.

O’Brien , W.H. and Malloy , J. 1993 . Method and apparatus for automatically segmenting animal carcasses. US Patent 5205779, 27 April 1993.

Pigsite . 2009 . Automated Production and Logistics. http://www.thepigsite.com/arti-cles/13/biodigestion-biofuels/2917/environmental-issues-control-slaughter-production (Accessed August 2010).

Purnell , G. 2004 . Robotic Automation for Pork Primal Cutting . Food Technology International . ISBN 1-85938-535-4 / ISSN-1460–3179, 54–58.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3



Purnell , G. and Brown , T. 2004 . Equipment for controlled fat trimming of lamb chops. Computers and Electronics in Agriculture , 45(1–3), 109–124, doi:10.1016/j.compag.2004.06.004

Purnell , G. , Maddock , N. and Khodabandehloo , K. 1993. Robotic deboning: A funda-mental approach to engineering a system. Proceedings of Artificial Intelligence in Food and Agriculture (AIFA 93) Conference , Nîmes , France , 27–29 October.

Ranger , P. , Ottley , G. and Smith , N. 2004. The pitfalls of purchasing. Food Processing , October, 20.

Roberts , C.A. 1984. Mechanical boning and trimming of mutton. Proceedings of the 23rd Meat Industry Research Conference , Hamilton , New Zealand , 90–93.

Robertson , A.A. 1980 . Mechanical pelting: Introduction. Proceedings of the 21st Meat Industry Research Conference , Hamilton , New Zealand , 24.

Taylor , M.G. 1993 . Automated Y-cutting of sheep carcasses. Meat’93. The Australian Meat Industry Research Conference , Gold Coast , Australia .

Taylor , M.G. and Templar , R.G. 1997 . A washable robot for meat processing . Computers and Electronics in Agriculture , 16 , 113 –123.

Templer , R. 2004 . Cutting and Boning | Robotics and New Technology, in Encyclopedia of Meat Sciences , Ed. Jensen W.K., Elsevier Ltd., Oxford, UK, 381 –388.

Templer , R. , Nicolle , T. , Nanu , A. , Osborn , A. and Blenkinsopp , K. 2000 . New auto-mation techniques for variable products. In Proceedings of Meat Automation Congress (MAC) . Malaga , Spain , June 2000.

Templer , R. , Osborn , A. , Nanu , A. , Blenkinsopp , K. and Freidrich , W. 2002 . Innovative robotic applications for beef processing. In Proceedings of Australasian Conference on Robotics and Automation (ARAA) , Auckland , New Zealand , 27–29 November, 43–47.

Trow , D. and Ng , W.Y. 1994 . Beef and pork loin boning. Proceedings of the 28th Meat Industry Research Conference , Auckland , New Zealand , 187–191.

Wadie , I.H.C. , Maddock , N. , Purnell , G.L. , Khodabandehloo , K. , Crooks , A. , Shacklock , A. and West , D. 1995 . Robots for the meat industry . Industrial Robot , 22 (5), 22 –26.

White , R.M. 1994 . Fututech. Proceedings of the 28th Meat Industry Research Conference , Auckland , New Zealand , 165–168.

Wickham , G. 1988 . Progress with boning and cutting technology. Proceedings of the 25th Meat Industry Research Conference , Hamilton , New Zealand , 160–164.

Wickham , G. 1990 . Developments in mechanical boning and cutting. Proceedings of the 26th Meat Industry Research Conference , Hamilton , New Zealand , 232–237.

Wickham , G. 1992 . An overview of boning machinery developments. Proceedings of the 27th Meat Industry Research Conference , Hamilton , New Zealand , 317–321.

Cop

yrig

hted

Mat

eria

l dow

nloa

ded

from

Woo

dhea

d Pu

blis

hing

Onl

ine

D

eliv

ered

by

http

://w

ww

.woo

dhea

dpub

lishi

ngon

line.

com

M

onas

h U

nive

rsity

(76

5-47

-440

)

Tue

sday

, Mar

ch 1

2, 2

013

7:12

:01

PM

IP A

ddre

ss: 1

30.1

94.2

0.17

3

robotics and automation in the food industry || robotics and automation in meat processing

Documents