Skip to content

Automating the Customization Process

By Karen Haywood Queen Contributing Editor, SME Media

Automated customization is moving closer to reality as robots become more affordable, autonomous robots require less programming, and additive manufacturing allows for late-stage changes.

Costs already have come down dramatically. In the mid-2000s, for example, a robot cost $100,000, according to Alastair Orchard, vice president of digital enterprises at Siemens Digital Industries. Now, he noted, they run about $10,000.

The same goes for programming, which in the past required someone to go on the shop floor with a laptop and recode all of a robot’s movements. This isn’t a problem if the robot only did repetitive tasks, Orchard said, but that usually isn’t the case.

“In recent years, as the subject of intense research by Siemens and other companies, autonomous robots require less and less direct instruction,” Orchard continued. “They have vision. They can work out what to do on their own and together.”

Along with better robots, additive manufacturing can meet some of the demands for custom automation, said Tsz Ho Kwok, research chair and associate professor at Montreal-based Concordia University’s Department of Mechanical, Industrial and Aerospace Engineering.

4D-Printing Takes Shape

Alastair Orchard, VP of digital enterprises at Siemens Digital Industries

Moving forward, four-dimensional printing—aka bioprinting or shape-morphing systems—will allow a manufacturer to determine the final shape at the very end of the process, Kwok said. The additional dimension is time. Using smart materials and adding another step at the end, the 3D-printed structures transform into a different shape over time with input from environmental stimuli such as heat, light, or hot water.

This allows for faster production because fabrication starts with a few layers in 2D and then transforms to 3D, Kwok said. This approach saves cargo space during shipping because components and end products can be stored in a compact shape before transformation, he explained. The finished goods have better quality because the stair-stepping effect in 3D printing is minimized. Applications include aerospace, automotive, electronics, fashion, and machinery.

“We try to delay the final shape-setting phase,” Kwok said. “The shape is not finally decided until the last step. We may form the final shape when the customer comes. If we do it in the factory, we use soft robotics to deal with the changes in customization.”


Orchard identified three levels of customization, with increasing amounts of difficulty. The easiest is substituting one attribute of the product for another, notably a feature such as color that has no effect on how the product is made.

“If you’re making a formulated product, everything is the same except you tell the dosing operation to choose one material instead of another one,” Orchard said. “If you have an assembled product, you swap the blue component for the identical green component. There’s no effect on the operation itself because the machine doesn’t have do anything different except pick a blue thing instead of a green thing.”

Other easy wins involve last-stage customizations such as printing a customer’s name on a package, or, slightly more difficult, creating a custom combination of goods in a package such as a razor, shave cream, and deodorant of choice, he added.

“Due to the random nature of the task, the task has to be executed by hand in a dedicated warehouse,” Orchard said. “As the service grows in popularity, the volume of requests become impossible to service and so the company switches to autonomous robots.”

The second stage of difficulty is adjusting an operation in the overall flow, Orchard explained, such as gluing a component instead of soldering. “This requires the machine that is attaching these two things together to operate in a completely different way,” he said.

“Again, that’s where robots come in. They’re ideal for flexibly accommodating changes to individual operations,” he continued. “They can move in almost any way, depending on how many axes they have. We now have nine-axis robots that can articulate in almost any way and do anything a human can do. Robots are really the perfect way of adding flexibility to a single operation within the production flow.”

The third level of customization is to modify the entire production experience, as is done at Siemens’ Amberg Electronics Plant (EWA) in Germany. “Traditionally this has been a highly invasive modification, according to Orchard. “Most factories have a hard-coded sequence, someone presses a button, and the factory goes through 10 steps to make this kind of product.”

Increasing flexibility makes manufacturers more responsive to their customers, and it makes it easier to meet the European Union’s Industrial Emissions Directive (IED) for production and post-production carbon dioxide emissions, Orchard said. For example, instead of a consumer packaging goods company producing products in various countries—liquid soap in France, shampoo in Germany and hand sanitizer in Poland—then shipping the products across Europe, the company can redesign each existing factory to make multiple products, thus cutting down on shipping.

“For one company, 91 percent of its CO2 is generated post manufacturing,” Orchard said. “They don’t have much choice but to invest in these flexible factories.”

Alastair Orchard discusses bridging the gap from where manufacturing capacity exists to where the demand is.

Additive’s Role

Additive manufacturing can meet some of the demands for customized automation, according to Kwok.

“It is not as simple as ‘I have a 3D printer,’” he added. “The challenge is migrating 3D printing into other processes.”

In many cases, manufacturing customers strongly desire incremental change as opposed to a thorough revamping. “If we change something, there is always a risk,” Kwok said. “You don’t know how the market will respond. You don’t know if this is a feature you will keep forever. You want to test that feature, see if it will bring a benefit. Manufacturers say, ‘We still need to make sure this is something our customer needs.’ Bottom line, we always want gradual change.”

At the beginning, end customers may not even know what they want. After their first need is met, Kwok said, customers often realize what else they want.

Expense is a significant challenge and a limiting factor for low-cost, mass-produced items. “We know customization very well,” Kwok stated. “But because the customization cost is very high, it will be used only for high-end applications,” he continued. “We haven’t achieved mass customization yet. If we can automate for mass customization, use the same technology without needing to change anything, everyone will be able to enjoy this benefit.”

Amassing Challenges

Tsz Ho Kwok, research chair and associate professor at Concordia University’s Department of Mechanical, Industrial and Aerospace Engineering

Kwok identified several other challenges including:

Meeting increasingly diverse customer needs;

Developing scanning technology that is portable and efficient;

Designing free-form shapes as opposed to fixed geometric shapes; and

Meeting these needs quickly.

Kwok’s Concordia research team has developed a customization tool, DesignForFab, that is said to slash design time from days to just five minutes. What’s more, the latter can be achieved by an engineering student, according to Kwok, rather than requiring an experienced designer to five minutes. One requirement is that the reference model is represented by high-quality mesh.

Mass customization isn’t just about one thing, Kwok noted. It’s not only a manufacturer’s problem, he explained, it involves the customer, scanning, computerization, visualization, and computer graphics. And it requires planning, management, engineering, functionality, then automaton and economics. Morever, Kwok emphasized, “The scanning technology needs to be portable and efficient.”

Flex Factories

The ideal end state is a factory that combines the flexible nature of aerospace manufacturing with the efficiency of automobile manufacturing, Orchard opined.

“An automotive factory is incredibly efficient but not flexible,” Orchard said. “It can’t pivot from a car to a truck to a personal scooter. It costs hundreds of millions of dollars to build a factory like that and you have to be certain to be able to make cars like that for years and years and years to pay down the investment in the factory. If you’re at the halfway point of paying down your factory but no one wants your car anymore because they want a scooter, you can go bankrupt.”

Meanwhile, aircraft/aerospace factories are highly customized but not as efficient, Orchard said. “Each side looks at the other’s model with terror,” he said. “But both of these types of business models need to meet in the middle. All you need is one company to solve this problem.”

Siemens’ EWA factory, which makes 1 million items per month (one every second), was redesigned to achieve the third level of customization. As of 2010, the factory’s automation level tripled from 25 percent to 75 percent. Productivity is up 1,400 percent and production quality is now at 99.9988 percent, the company claimed.

Workers and robots at the plant make 9,800 switch gears, programmable logic controllers (PLCs), and human machine interfaces (HMIs) on one set of equipment, Orchard said. These different families of products are not traditionally made on the same manufacturing lines or even in the same factory, Orchard noted.

This diversity is possible because instead of proceeding numerically in order from station to station as in an automotive plant, each product has a unique journey through the factory based on its specifications and equipment available at that moment. The process is designed so that each product made can query stations within the factory at each step of the manufacturing process, Orchard said.

For example, a liquid product tells the system it needs to be mixed at a particular temperature and quantity that prompts available, qualified industrial blenders to essentially raise their hands and note that they are open. An automated guided vehicle (AGV) takes the product to the correct station.

“The factory isn’t engineered to make this product,” Orchard said. “The product is engineered to use the flexible factory to get itself made.”

The U.S. is better positioned than Europe for this new wave of customization, according to Orchard. “Over the last few years, European factories painted themselves into a corner by highly automating their processes. They are very efficient. But injecting flexibility into these factories is very difficult. In the United States, many factories were built, not in the 1950s, but in the 1850s. That provides a great opportunity, a little bit like African nations who never had a fixed telephone network and went straight to mobile.”

Looking ahead, robots can play an even larger role.

“In extreme cases in our labs, if you give a robot a CAD file, it will work out on its own how to move in to get the job done,” Orchard said. “It has vision. It can find the parts itself in XYZ space. One robot can pick up a part and hand it to another robot. That’s the future of robotics. It’s not that the robot can achieve more customization but that it is far more autonomous in its flexibility.”

  • View All Articles
  • Connect With Us

Always Stay Informed

Receive the latest manufacturing news and technical information by subscribing to our monthly and quarterly magazines, weekly and monthly eNewsletters, and podcast channel.