After they’ve proved their worth in secondary packaging and grunt applications, robots are slowly making inroads for jobs closer to the process. 

For pick-and-place applications, delta-style robots can make up to 140 or more picks per minute, depending on product and application. Source: Bosch Packaging.

“Hired” in 2009, Lucy and Ethel-two ABB FlexPicker delta-style robots-work untiringly picking candies and arranging them for packaging. The right-hand photo captures a wider view of the line. Source: Wolfgang Candy Inc.

Ask any robot supplier or system integrator whether it’s a good idea to choose robots because you’d like to replace a couple of dozen humans-or take on a complicated task that hasn’t been automated before-and the experts will be the first to say, “Let’s take a very careful look before we continue with this project.”

While robots have had an excellent track record in general manufacturing, putting them to work in a food or beverage plant puts additional demands on them and the processors who maintain them. But there are benefits. Besides performing in heavy lifting and pick-and-place applications, robots and controllers combined with smart vision systems can spot damaged product and keep track of lot and serial numbers when required. And they provide new solutions in traditional, mature applications.

According to Brian Huse, director of marketing for the Robotics Industry Association, robotic systems currently can be used in handling bags, flow packing and packaging/kitting and palletizing and depalletizing applications. With the recent availability of delta-style robots, mid-stream process applications are beginning to employ robots for stacking and assembling food parts, for example, cookies. While most processors find the consistency of robotic/vision systems helps eliminate product waste and improves quality, too “perfect” may not always be desirable. Huse points to a Pepperidge Farm application where delta robots assemble cookies. To keep a handmade appearance, the robots are actually programmed to be a bit imperfect in the cookies they build.

As a processor looks further upstream toward processing, applications tend to get more complex and are sometimes tricky to implement. “In terms of the level of automation, secondary packaging is more mature [than primary packaging], mainly because the complexity to handle primary packed products is [less] than that of ‘naked’ products, which are mainly picked and placed,” says Roland Czuday, Bosch Packaging Systems product manager. Challenges often occur in the design of the gripper to ensure safe and gentle product handling, he adds.

Ostensibly, according to Czuday, it comes down to building into an automation system the positive attributes of human beings at a competitive price. These attributes are what we take for granted: keen eyesight, manual dexterity, quick thinking and flexibility to adapt-which translate to real-time vision systems, multi-purpose gripping tools, easy-to-program and reliable controllers and the ability of software to adapt to product changes.

Flexibility and speed are why Wolfgang Candy (a 100,000-sq.-ft., three-building operation in York, PA) hired “Lucy and Ethel,” two ABB Model IRB 340 FlexPicker robots, to automate a new primary packaging line for its chocolate candies. Installed by JLS Automation in 2009, the robots meet the processor’s increased production needs for contract and private-label production. Rob Wolfgang III, food safety and quality manager, is enthusiastic about the customizable potential offered by the new line. “Another expectation for the robot,” says Wolfgang, “was speed.” We needed a system that could meet or exceed 200 pieces per minute. JLS has given us both flexibility and speed, because it is a pick-and-place system and not a dedicated, hard-engineered system.

“We can pick-and-place nearly anything, which gives us tremendous flexibility to deliver customized product solutions for existing and future customers,” adds Wolfgang. The processor produces more than 120 SKUs on 14 lines, and added 40 employees to its workforce the same season Lucy and Ethel began work.

Solutions that offer the flexibility to handle product types ranging from bags to natural products have been scarce in the food industry. Using vision algorithms, food processors can retrofit their solutions to handle products with highly variable dimensions. Source: Adept Technology.

New solutions in mature applications

While it may seem that secondary packaging applications are mature, there are always challenges to overcome-often ones that don’t show up until changes are made for other reasons. In some of these cases, robotics can help conventional machinery when it gets into trouble. For example, Dick Motley, Fanuc Robotics senior account manager, describes an application for a beverage company that initiated a sustainability program in its secondary packaging, “light-weighting” all its bottles. The processor evolved its cartons from an overwrapped carton to a tray with a shrink configuration-then to a corrugated pad-shrink configuration and, finally, to just shrink wrap. While this new packaging worked well for carbonated beverages (carbonation’s internal pressure adds to the structural support of the bottle), water bottles just didn’t survive conventional bump-and-turn automation. They may as well have been water balloons.

Motley says one of Fanuc’s integrators, Hartness, devised a robotic solution to prevent the destruction of water bottles in the bump-and-turn location. The idea was to have a robot gently grab and guide the product and shuffle it around the bend while it tracked the moving delivery conveyor. The robot was programmed to move at the same velocity as the conveyor and grab and guide bottles around the bump-and-turn location while a larger robot grabbed a layer at a time. Result: No more bursting water bottles.

At the downstream end of robotics applications from packaging is palletizing. While palletizing is a mature technology, according to Rick Tallian, ABB segment manager for the US consumer industry market, it doesn’t mean processors are doing it right. “We find that people don’t question palletizing technology, and there are failure modes because [processors] are doing the wrong thing.” Palletizing is moving into a new world, and it probably doesn’t have to do with the mechanical equipment and the robot as much, but the software is pushing this stage into a new direction, he adds.

According to Bill Torrens, RMT Robotics director of sales and marketing, in today’s distribution model, the pallet shouldn’t be used to store products before they’re shipped-unless a processor wants to load pallets, unload them to fill an order and restack a pallet to go out on the truck to a customer. Torrens calls his solution “palletizing avoidance,” and uses a staging area serviced by gantry robots for products that typically go out the door within eight hours, especially baked goods. With software connected to the manufacturing system, orders can be filled as “rainbow pallets,” mixed and matched to a customer’s preference. This concept doesn’t eliminate the use of pallets and warehouse space for seasonal goods, but it definitely alleviates the chaos of getting products out the door in a timely fashion.

Process applications

Because food products are not like machine parts, the former’s variability makes it tough to apply robotics to certain process applications. “I define processing as the handling of the raw or unpackaged food products,” says Staubli Robotics Food & Packaging Industry Specialist Bob Rochelle. “This market is not yet mature for industrial robotics and is defined as a growth area for Staubli. Because of this, we have invested in developing our HE model line specifically targeted to these applications. Currently there are minimal options for raw food handling or sanitary applications for traditional industrial robots. The issue is that traditional robotics has been tied to industries like automotive where sanitation is not a requirement. But this is beginning to change,” says Rochelle.

One application is robotic meat trimming, but it hasn’t caught on as fast in the US as it has in Australia and Europe, says Craig Souser, JLS Automation president. There may be several reasons for this. The most important may be ROI, and it’s not good enough to make it a serious replacement for humans in most American meat establishments. Most non-US applications get government funding, and throughputs-while they may be ample in Australia and Europe-are not up to speed enough to compete with humans in US applications.

In Australia, Machinery Automation & Robotics (MAR), a robotics systems integrator, transformed a dangerous work environment at a meat processor with three safe, sanitary robotics solutions. The meat processor wanted a safer work environment for its employees, but also found it extremely difficult to hire and maintain reliable, skilled labor, particularly in the area of brisket cutting.

“Traditionally, the brisket cutting part of meat processing requires a person to operate a heavy, industrial-size hydraulic knife device called a brisket shear,” says MAR CEO Clyde Campbell. “This practice combines a dangerous tool with a physically demanding task that commonly results in wrist and back injury. Our solution eliminated the need to endanger employees altogether.”

MAR’s robotics installation features an ABB Model IRB 4600 industrial robot and an advanced sensing system to ensure accurate, reliable operation. A circular cutting saw, fitted with a dustless micro-toothed knife blade and centering guides, makes one clean cut at the center of the brisket. This has drastically improved quality with fewer instances of internal organ puncturing and other damage, simplifying downstream processing. Moreover, consistent submerging of the blade and guides into a cleaning solution after every cut sterilizes them more thoroughly and prevents cross-contamination.

Machinery Automation & Robotics (MAR), an Australian robotics integrator, designed and built a brisket cutter, which makes one well-placed, sanitary cut. An ABB IRB 4600 robot uses an advanced sensing system to ensure accurate and reliable operation. Source: ABB Robotics.

Touch and sight

Touch (gripper technology) and sight (vision systems) are two key areas that are making it possible to use robotics in new applications. “The area of greatest focus for Adept Technology over the past few years has been in the handling of randomly oriented products that need to be placed into HFFS [horizontal form-fill-seal] or packaging machines,” says Rush LaSelle, director global sales & marketing.

“In the case of natural products typified by those in the proteins market (e.g., fish, poultry, beef and pork), vision systems and tooling alike have had difficulties in managing the variability associated with these products,” says LaSelle. Adept has responded by offering solutions that use vision algorithms to identify products with highly variable dimensions. These vision systems are also capable of conducting various levels of quality control and grading as the systems identify incoming products. Likewise, the company has worked extensively in compiling a suite of intellectual property surrounding the tooling needed to handle not only natural products, but through the use of advanced fluid mechanics, delicate products such as puff pastries at high rates of speed, without compromising the products’ integrity and appearance.

Delicate food handling requires a special design for vacuum cup grippers, not devices you find at your local Ace Hardware store, says Tallian. Vacuum systems are a very critical link and usually need to be custom designed by integrators who understand the technology and food applications.

With all the vision inputs and motion outputs to the robot, there has been a need not only for a robust robot controller, but also for some common system that can link the robot with connecting equipment-be it packaging systems, conveyors, etc. “One challenge that KUKA Robotics has met is the demand for a common control system on the packaging machine and on the robot associated with that machine,” says James Cooper, vice president sales and marketing. The robot supplier worked with Rockwell Automation to integrate a ControlLogix PLC control system. This allows the processor to use one common control system-one the processor’s operations staff already knows, adds Cooper.

Ease of use

With many vision systems being PC-based and easy to learn and set up via drag-and-drop “programming,” it’s no surprise processors are expecting the same of the entire robotics system-no matter where it’s working in the plant-processing, packaging or warehousing. “Easy-to-program software is what processors want,” says Tallian. “There is still too much engineering content in a robotic system.” Tallian says processors want to be able to make the simple product changes themselves without having to call in suppliers and integrators.

Programming may get a boost from an unexpected source, says LaSelle. “The area where we might see the most advancement over the next five years will be the manner by which automation is deployed, configured and managed for product changeover. I expect there will be a heavy influence from gaming and consumer electronics to where programming of robots will be replaced with easier graphical interfaces and gesture-based programming.”

Performance evaluation

Whether you measure the performance of a computer by GHz or MIPs, a car by horsepower or displacement or a robot simply by cycles per minute, you may not be getting an accurate representation of performance without checking it in an actual application. There are several factors that should be included in the performance evaluation, says Tallian. Raw performance is only one factor. Others include robot motion-how it’s developed and what tools are used with it. Other questions of concern: How suitable is the product for high-speed handling? What vision system is employed, and how can the product or part be located when placed randomly on a conveyor belt? “Just increasing the performance by 25 percent doesn’t mean you necessarily get a 25 percent increase in throughput in a production line,” adds Tallian.

Czuday cites some typical performance expectations. For single pick-and-place applications with delta robots, performance is up to140 picks/min. If more than one product is handled, or if the gripping tool and product are heavier in weight, lower speeds can be expected, like 80 picks/min. Secondary packing applications reach up to 40 cycles per minute. Often layers or complete groupings of two to four packs are handled per cycle. This means big grippers and performance levels of up to 20 cycles/min.

“In my opinion, often the physical properties of the product limit the performance, so I do not expect a strong increase of the performance level [of equipment in the immediate future],” adds Czuday. “Moreover, very fast pick-and-place cycles are perceived by many [processors] as being not as reliable as slower movements. Of course, people will get used to faster operations, but there seems to be some sort of ‘natural threshold,’ above which customers start to get worried about the robustness and reliability.”

“Robotic systems are now capable of providing speeds that in many cases outpace the ability for today’s tooling to acquire and deposit products without damaging them,” says LaSelle.  As such, it would appear that increasing the speeds will be met with diminishing returns for production of items that are susceptible to damage, says LaSelle. “That said, many products and applications still have opportunities for higher rates, so as technology advances, it should be expected that speeds, performance and overall capabilities of robotic automation with improve.”

Benefits and challenges

With all the benefits robots bring to an application, there will be some concerns. According to Steven Hawkins, Stellar Automation Services director of automation, although robots decrease the number of operators, most processors will need to increase their maintenance staffs with people fluent in robotics. While there are fewer operators on staff to contaminate food products, robotic systems are complicated with a greater use of pneumatics and electronics, which demands a higher skill set from the maintenance staff. Hershey Foods needed an automated solution for packaging its Reese’s Peanut Butter Cup wrapped chocolates from six separate wrapping machines into 10-count (2x5) trays. The automated systems needed to handle a continuous flow of wrapped chocolates and paper trays and produce loaded trays for downstream wrapping. Saber Engineering designed and built three robot packaging systems using 3-axis AdeptModules and a multi-pick gripper that can pick 20 chocolates at once. The robots normally picked product from two incoming lanes simultaneously.

When a wrapper went down, the robot could dynamically switch to picking product from only one incoming lane. Saber installed a vision inspection system for each wrapper to inspect package quality before it entered the robot packaging systems. An Intellution SCADA system was installed to provide a single operator interface for all three packaging systems as well as providing data for production reports and IT.

Food safety requirements

When this kind of data from SCADA systems is available, track and trace gets much easier. According to Hawkins, while robots handle food, they’re also checking barcodes for traceability, tracking batch and lot numbers and checking for two or three different products coming down the same line-all this while not contaminating food with human illnesses. With a vision system, robots easily handle different sizes and types of products. In the pharma industry, tracing individual serial numbers is becoming more frequent, and this technology can be applied to the food industry if required.

“Traceability has become a critical mandate for many who occupy a place in the food and beverage supply chain,” says LaSelle. “We are increasingly being asked to have the images of products fed to a database where they are recorded and tagged with the packaging and ultimately the lot with which they leave the facility. So, where robots are implemented with vision systems, opportunities exist to utilize the images used for product location to also provide product traceability.”

Tallian has put in systems that do a form of lot tracking. They process date codes with a specific barcode number and automatically record the time of this processing, so if the processor has a recall, it’s possible to pull all affected food down to lot and batch numbers out of the supply chain based on the processor’s SCADA data. There isn’t a possibility for human error. “One of the big issues we discussed with pharma companies is they would have to track serialization into cartons and up the line,” says Tallian. “It wouldn’t surprise me if it’s going to fall into food, as well. That’s the virtual elimination of human intervention because humans can make [a wrong] decision where a robot is better because it does what it’s told to do.”

Plan ahead

If you’re new to robotic applications, there are three basic issues to keep in mind. According to Rochelle, one of the most common mistakes is that processors select a complicated application first. “I would recommend that the first-time robot user select a simple pick-and-place application to learn the process of integration, the trials of maintaining the system in running condition and as a base line to advance to more complicated applications,” says Rochelle.

Souser makes a second point. Processors that approach systems integrators saying the main goal of their first robotics project is to replace two dozen people in the packaging area should not look at this as a short-term goal. Instead, make it a five-year plan, which at least allows processors to get accustomed to robotic applications-how workflows are affected and how maintenance is conducted.

Closely allied to replacing people is the question of ROI for a robot installation. Czuday says that in general his customers aim at ROI of one to two years. However, the applications and product margins are very different. Often the performance and the flexibility (reduced setup time) are key arguments. But there are more processors that focus on investment costs, rather than operational and lifecycle costs.

“We still battle ROI every day,” says Tallian. It’s still the major criterion for companies making decisions on whether to use a robot. Tallian feels processors are making more conscious decisions to help their employees by removing them from at-risk situations. “But we still have to get a lot closer to the hard numbers of being able to pay back [in replacing people’s jobs with robots].” 

Third, processors that plan robotic applications often reap benefits they never considered in the design phases of a project. Motley points to a processor in the baking industry. In the baking of specialty breads, the processor decided to place a robot at both ends of the band oven because of the ergonomic issues involved-reaching low to unload stacks of dough to place in the oven and to remove the loaves after baking-all in a rather hot environment. When the process was automated, the robots were able to keep up and load and unload every available slot going into the oven, where their human predecessors left several empty slots. Because the oven was no longer “baking air” in the empty slots, throughput went up 80 percent. But an even bigger win was in energy savings per loaf of bread, which yielded nearly a 50 percent improvement. An improvement in throughput reduced the time needed to run the oven. “So while setting out to solve an ergonomic problem, the processor had no idea it would save this amount of energy,” says Motley.

For more information:
Brian Huse, Robotics Industry Association, 734-994-6088,
Rick Tallian, ABB Robotics, 248-391-9000,
Roland Czuday, Bosch Packaging Systems GmbH, 49 (7151) 7007-0;
Dick Motley, Fanuc Robotics, 248-377-7522,
Bob Rochelle, Staubli Robotics, 864-433-1980,
Craig Souser, JLS Automation, 717-505-3800,
Rush LaSelle, Adept Technology Inc., 925-245-3400,
James Cooper, KUKA Robotics Corp., 586-465-8817
Steven Hawkins, Stellar Automation Services, 904-899-9398,
Bill Torrens, RMT Robotics, 905-643-9700, ext. 273,