Here Come the Cobots!

Ed Colgates wife says the cobot concept is a special gift -- and thats understandable since it did come to him in church. "The confirmation of the catechumens was happening -- a drier service than usual -- so rather than being riveted to the homily, I allowed my thoughts to wander a bit and suddenly the idea of a computer-steered wheel popped into my head, and that was really the origin of the cobot idea. My wife tells me that it wasnt me at all -- it was handed to me. But almost instantly I realized that we had a really superior way of having a human and robotic-type device interact safely." Colgate, associate professor of mechanical engineering at Northwestern University in Evanston, Ill., began his preoccupation with cobots as the result of a 1995 research grant from the General Motors Foundation. "GM was very interested in helping assembly-line workers do their jobs better," he says, "and we began to research what we initially called operator-assisted devices. In the spectrum of automation technology that stretches from robots to conventional assist equipment, the cobot research at Northwestern targets a yet uncharted middle ground. "We wanted to develop and adapt robotic-type technology to facilitate automotive assembly in a new way," Colgate explains. "The goal was to find a way to make robots or robot-like equipment safe enough to team with people. The approach: use some of the intelligence that we associate with robots, but make it work with a human operator. Rather than fully automating and replacing human operators, the goal with cobots is to try to exploit the strengths of both the automation and the human operator." Although robots have revolutionized some jobs in manufacturing, the implementations consistently seem to share a characteristic -- they are almost always isolated from production workers by fences and signs that warn them away. Together with co-inventor Michael Peshkin, a fellow associate professor of mechanical engineering, Colgate wants to change that. They are pioneering the development of a new way of enhancing the capabilities of a line operator with robotic-type technology. The name cobot, a contraction of collaborate and robot, reflects the goal of the technology -- machines with robot-like features that are able to safely collaborate with assembly-line workers. Although robotic automation has become widely used in automobile plants during the last 25 years, a close examination reveals that in most facilities the units perform little more than welding and painting where they excel in isolated efficiency. Until now robotic technology has done little for vehicular-type assembly operations. "For example, in a typical plant, we might have 300 welding robots, perhaps another 50 or 100 painting, and maybe one in the general-assembly area," says Prasad Akella, the senior project engineer leading GMs cobot-application-development thrust at the companys Technical Center, Warren, Mich. He says GM is interested in applying cobot technology throughout its assembly operations. "The primary intent is to improve ergonomics and safety of the hundreds of people who install various subassemblies into a vehicle," adds Steve St. Angelo, director of the General Assembly Center at GMs Technical Center. But the implications are broader and extend beyond improving GMs ability to compete. "Our workforce is getting older, and weve got to alleviate ergonomic stresses by giving them the tools to perform efficiently. Cobots are about safety, ergonomics, quality, throughput -- just about every metric that is important to manufacturing management in an assembly plant, automotive or not." St. Angelo says the technology could bring benefits to companies in other manufacturing industries as well as such organizations as Federal Express and the U.S. Postal Service. A prototype is under evaluation at Ford Motor Co. Working with Colgate and Peshkin, St. Angelo and Akella are targeting the basic tasks common to all of GMs assembly plants. "One is the job of lifting components and guiding them into cars for installation," says Akella. "Any task that requires a human being to move a component into the car or truck during assembly is a job where cobot technology could be beneficial. The examples vary from components like batteries to entire subassemblies like a cockpit or instrument panel." Adding urgency to any automakers interest in cobots is the trend toward larger modules (subassemblies) that assembly workers must handle, adds St. Angelo. "Not only are the modules getting bigger, but as vehicles get more aerodynamic and the door openings smaller, it becomes more difficult to maneuver and insert components to finish the assembly process." Akella identifies two key functions of a cobot: "One is to provide gravitation and force compensation so that the effort of lifting components and assemblies is taken from the worker. The second is to help the operator guide the components and assemblies into place without damaging the product." His analogy for force compensation is the way power steering assists drivers. "The person is still driving, but the power assist makes it easier." To illustrate guidance, he alludes to drawing a straight line with and without a straight edge. "Its obviously easier with a straight edge. So thats really what were trying to do on the guidance side. Were trying to use devices that through hardware and software create those straight edges -- actually virtual surfaces -- in a programmable way." The cobot researchers at the General Assembly Center have two prototypes under test in their development laboratory. Some of the initial focus is on facilitating the installation of complete instrument panels -- large cumbersome assemblies that barely fit through the door openings of vehicles. By providing virtual surfaces, the cobot would guard the door frame and interior surfaces of the cab during the assembly process. Scraping or gouging can result in costly quality problems. These surfaces are defined and controlled by the computer software that operates the cobot. The virtual surfaces defined by the cobot software are invisible, but their presence cannot fail to be noticed. "The virtual surfaces extend out of the cab door like an invisible funnel," says Peshkin. "Workers can maneuver the instrument panel down the middle of the funnel if they wish, but more likely they will prefer to push it up against a virtual surface and just slide it along into the cab." The virtual surfaces must be strong enough to deflect and redirect the motion of an auto part pushed by a human worker. Cobots do not use motors for this purpose, which would require ratings of at least human strength. Instead, cobots use free-rolling wheels. The inventors have found in-line skate wheels to be ideal. "[These] wheels, appropriately oriented, can change the direction of motion of a 200-lb skater in a fraction of a second," says Colgate. "Thats exactly what a cobot does -- it redirects motion by steering." Although cobots use robotic technology and are designed to help people with assembly and material-handling tasks, they look like neither robots nor the human form. Consider Scooter, a nickname given to a three-wheeled cobot prototype at Northwestern. Resembling a low, sturdy triangle with an in-line skate wheel at each corner, Scooter remains firmly in contact with the floor. The wheels are independently steered, under control of a computer. True to the passive nature of cobots, Scooter cant move on its own. "You can push it in any direction until a virtual surface is reached," says graduate student Witaya Wannasuphoprasit, who built Scooter (and who is now chief engineer at Colgate and Peshkins start-up, Collaborative Motion Control Inc., Evanston, Ill.). As the cobot approaches a virtual surface, the control computer steers Scooters wheels so that only motion parallel to the virtual surface is possible. "It feels like running into a wall," says Peshkin. "Its an invisible wall -- and a slippery one. Suddenly you and the cobot are just going in a new direction, yet nothing is physically there." The future may hold cobots of other shapes. The researchers are designing an arm-like cobot that could be used in computer-assisted surgery.