A History of Collaborative Robots: From Intelligent Lift Assists to Cobots
Kagan Pittman posted on October 28, 2016 |
The evolution of cobots from prototypes at General Motors to a growing market of industrial robots.

The manufacturing industry is entering a global renaissance. 

With all the talk of Industry 4.0, the Industrial Internet of Things (IIoT), cyber-physical systems and collaborative robots (cobots), it’s easy to get lost in all the technologies promising greater levels of efficiency.

Much of the more advanced technologies are still out of reach for SMEs, but collaborative robots are beginning to see a noticeable spike in popularity precisely because of their affordability and emphasis on safety.

Today, humans can work with cobots without requiring safety fences or quarantined factory floor footprints. This is an inherent part of their “collaborative” nature, but it’s not the entire picture. 

Cobots designed for different applications still require unique safety standards as outlined by ISO safety standards and certifications.

Per the International Standards ISO 10218, there are four general safety features for cobots: 

1    Safety Monitored Stop: 

This feature is used in cobots designed to predominately work alone, only rarely having a human enter their workspace. For example, if an employee had to make an adjustment to a part the cobot was handling or remove that part from the cobots space, the cobot will cease movement, but not completely shut down to easily resume its tasks when the human leaves. This is the loosest of interpretations for what “collaborative robot” means, employing advanced proximity sensing technology to turn a conventional industrial robot into a form of cobot.  

2    Hand Guiding: 

Hand guiding is a collaborative feature used for path teaching a robot – literally guiding the robot through a sequence of motions required to complete a task, like pick-and-place applications. These cobots often using end effector technology to sense its position and read forces applied to the robot’s tooling, allowing a cobot to learn through example – literally.   

3    Speed and Separation Monitoring: 

For applications requiring more frequent human intervention with larger cobots, a laser vision system can be installed in the environment to allow the robot to sense a human’s proximity, similar to a safety monitored stop system. With speed and separation monitoring, the robot will slow down more and more as a human approaches, completely stopping when humans get too close. These reactions can be programmed for a robot to perform at various distances between itself and the human operator. The robot will resume its task and slowly accelerate its motion as the operator moves further away. 

4    Power and Force Limiting: 

Robot models like Universal Robot’s UR series, ABB’s YuMi, Rethink Robotics’ Baxter and others, often require little to no safety technology due to power and force limiting safety features. Force limited collaborative robots can read forces in their joints, like pressure, resistance or impacts using embedded sensors. After feeling a disturbance, the robot will stop or reverse its course. The robots soft skin and very fast reaction times dissipate as much of the impact as possible. Their soft skins and rounder designs help to cushion impacts and even eliminate pinch points. These models often have internal motors and wiring to shrink their overall size as much as possible. Even though they can be regarded as “true cobots,” they still require risk assessments and are regulated under the new ISO/TS 15066 technical specification.

With these safety features, cobots take on a variety of shapes and sizes to address pick-and-place, injection molding and assembly applications, just to name a few. But the cobots on the market today aren’t where the technology started.

Today's cobots grew out of an effort in the late 1990s by researchers at Northwestern University, the University of California Berkeley and General Motors (GM), which used what would eventually become cobots in its automotive plants.

Robotics as a Solution for Ergonomic Issues
A floor-based door unloader developed by Northwestern University is tested at a GM plant. (Image courtesy Prasad Akella.)
A floor-based door unloader developed by Northwestern University is tested at a GM plant. (Image courtesy Prasad Akella.)

“In the ‘90s, the Occupational Safety and Health Administration (OSHA) had concerns with the way that GM and the manufacturing industry was handling ergonomic issues in its plants,” said Prasad Akella, who led the effort to develop cobots and related technologies while working as a staff engineer at GM.

Ergonomic problems would escalate into workplace injuries and lost work time for employees. These problems were effecting automotive manufacturers across the United States, but for GM these problems were most significant in the final assembly areas.

“If you tried doing something as simple as picking up and installing a 40-pound car battery, one a minute for eight hours a day, 200 days a year, your back is going to start feeling very sore,” continued Akella, who is currently an entrepreneur-in-residence at SRI International.

“OSHA said that the automotive industry had to address this important social problem and that GM, as an industry leader, had to lead the way. Steve Holland, who headed the Robotics Department, was tasked with solving the problem together with Jim Rucker, who headed the General Assembly Center.”

Just as GM had pushed the envelope on the design and use of industrial robots in 1961, working with Unimation, now three decades later GM set out to fulfill a need “to make safe robots that would work with people, not be caged,” said Akella.

Akella and his team brought in experts in robotics from the University of California, Berkeley and Northwestern University as well as other GM employees.

At Northwestern, GM's support went to mechanical engineering professors Michael Peshkin and J. Edward Colgate, whose research resulted in cobots. Homayoon Kazerooni at UC Berkeley received support to work on extenders or human power amplifiers.

The devices that resulted from these efforts, including cobots and extenders, were later collectively called “intelligent assist devices” (IAD).

Where did Collaborative Robots come from? 

A Z-Lift Assist developed by UC, Berkely is tested at a GM plant. (Image courtesy Prasad Akella.)
A Z-Lift Assist developed by UC, Berkely is tested at a GM plant. (Image courtesy Prasad Akella.)

Colgate and Peshkin were seeking a way for robots to improve ergonomics for human workers without introducing new risks from the robots themselves. What they came up with was the idea that robots and people could work in partnership, each contributing what each did best.

For the robot, that meant movement under computer control, without error. To assure safety, the robot could support a load but not cause it to move: All the motive force would come from the human worker.

The team first called their invention a "programmable constraint machine," because the robot contributed constraint surfaces that guided a workpiece under computer control.

Later, as the close interaction of human and machine became more apparent as a key quality, they introduced the term collaborative robot, or cobot, to highlight the interaction. Their first patent on cobots was filed in 1999.

These early cobots could help guide the motion of a worker to ensure greater precision, while supporting a load against gravity, but didn’t have as much versatility as Colgate and Peshkin had hoped.

“Eventually, we saw the value of them having the power to drive themselves around as well,” Colgate continued. “Limits on speed and power and good interactive design helped cobots remain safe even though they had some power of their own.”

Designing Cobots for Final Assembly

The original intent behind cobots was always to have humans and robots side-by-side, working together. In approaching cobot design with independent power in mind, there were several engineering hurdles Colgate and Peshkin had to overcome.

In their research, the pair designed an intelligent hoist system to replace XY rail systems, which are similar in design but limited in mobility as result of fewer axes.

(Image courtesy Prasad Akella.)
(Image courtesy Prasad Akella.)

Colgate and Peshkin’s hoist system used sensors to understand where the user wanted to move the payload and take over much of the lifting. The system incorporated a brushless DC motor and the handle was designed with sensors to allow the device to tell when the user wished to lift. Pushing the handle up brought the hoist up.

“It was a bit of a tricky control system to design because you want it to be really responsive but stable,” Colgate explained. “The thing was, you didn’t always want to use the handle. Sometimes you want to put your hands right on the payload because you’re doing something else at the same time, like putting in a fastener or hooking up a wiring harness. It’s just not helpful to be constrained to the handle.”

Sensors continued to play an integral part in making the device more intelligent and collaborative.

“We used a clever design with sensors that measured the load in the cable and they sensed not only the weight of the object, but the departures from that, which would’ve been due to somebody pushing up or down and driving the system,” Peshkin said.

Peshkin explained that the XY rail system used previously was greatly limited in its mobility due to limited axes, but with the use of their new intelligent hoist system, operators could move much more freely. Thus, the Intelligent Assist Device (IAD) was born.

“Operators could now push on a suspended payload and it would move in the direction they wanted instantaneously, as if it was just floating mid-air,” Peshkin remarked. “A key to that design was in developing something that looked at the angle of the cable that suspended that payload from vertical.”

As soon as the operator began pushing the payload, the cable that supported it would start to tip a bit in that direction. Robotic equipment at the top of the cable would then move the supports in that direction so, to the operator, it would feel as if it were floating.

(Images courtesy Prasad Akella.)
(Images courtesy Prasad Akella.)

Although this advancement helped address ergonomic issues, leaving the design at that would have allowed the risk of damaging vehicles on the assembly line. If an operator didn’t have total control, a lifted seat, for example, could connect with the side of the vehicle on the line.

To address this issue, Colgate and Peshkin designed what they called “virtual walls.”

“These were not walls that were designed just to prohibit motion, but in fact exploit them,” Colgate said.

Colgate pictured the concept as a funnel.

“You push up against the wall and slide along it,” Colgate explained. “It gave the operator a great confidence that they knew where they were going. The system had to be really good at sensing where the vehicle was and so we had to create the software infrastructure for designing those things, but it worked well and a lot of applications were able to take advantage of that capability.”

The inventions introduced by Northwestern and UC, Berkeley had first put GM’s employees on edge. Workers were concerned that their jobs might get replaced by the robots Colgate, Pehskin, Akella and their colleagues were developing.

However, as experimentation continued the early cobots began to change their perception for the better.

“The assembly line workers really appreciated that it was not a robot looking to replace them,” Peshkin said. “Because it was collaborative and because their human skills were going to continue to be needed to work with this cobot, it helped them do their jobs with less risk of ergonomic injury. They were smooth, quick, responsive and agile, so they were appreciated.”

Some of the prototype cobots quickly became recognized as essential by assembly line workers.

“I vividly remember an event that occurred at the Fort Wayne, Indiana plant,” said Akella.

“The line workers were using a prototype that we had planned to use, test and learn from over three months. When we went back to dismantle the device, Otilio, a line worker said, ‘Prasad, I have to thank you.’ I asked why and he replied, ‘My dream is to retire to my farm in good health. I didn’t see how that was going to happen until you installed this device.’ When we tried to pull that equipment out, he and several others went to the union to retain the use of the device!”

(Image courtesy Prasad Akella.)
(Image courtesy Prasad Akella.)

From Development to Commercialization

With GM’s initial mission success, the company looked to bring the technology and its benefits to manufacturers across the United States.

By using grant dollars to fund its academic partners, GM enabled the universities to own and commercialize the intellectual property. This spawned the founding of Cobotics Inc. by Peshkin and Colgate. Several years later, Kazerooni's technology was adopted by Gorbel Inc.

From that point, GM continued to invite the robotics research, vendor and user communities to join in advancing cobot and assist device technology.

Ford, then led by Tom Pearson, joined in 1996 to help drive the technology while delivery companies like UPS explored its use a couple of years later. Companies like FANUC Robotics and Gorbel also started developing their own technologies in this area.

“We continued the R&D process and then in 1999, took what we had learned to the entire robotics community at the IEEE International Conference on Robotics and Automation,” Akella said.

GM and Ford also approached the Robotics Industry Association (RIA) and the American National Standards Institute (ANSI) to write safety standards for the technology, to enable its broader use. This led to the creation of the first safety standard for “Intelligent Assist Devices — Personnel Safety Requirements,” T15.1 in March, 2002.

Today, companies like ABB, Rethink Robotics and Universal Robots have developed new products and have changed what “collaborative robot” means.

“Cobots and intelligent assist devices have transformed how people view, interact with and use robots today,” say Akella, Colgate and Peshkin.

“What we see happening today is person and cobot side-by-side, each doing their own thing toward a common objective,” Colgate said. “Their key challenge to overcome is that these newer devices need to be considered safe enough to be around without a fence and once you overcome that, you can work with them on the same part, side-by-side. I think it’s a really great thing.”

Goldman Sachs projects that the collaborative robot market will be attain a value of USD$3 billion by 2025, in a report published in April 2016.

Collaborative Robots Today

Manufacturers who need to invest in cobots today have a growing list of options when it comes to which brand to choose for their applications.

Before approaching an integrator for advice, it’s important for the manufacturer to do some of their own research first, to understand their application, know what they’re capable of investing in and understand the technology they’re looking for.

Our e-book below outlines the most popular collaborative robots on the market today and includes technical specifications as well as mentioning which applications each cobot is best suited for.

To learn more, download The Collaborative Robot Buyer’s Guide E-Book.

Create your own user feedback survey

Recommended For You