Modern articulated arm robots are precise, capable of repeatedly making movements within a few thousandths of an inch day after day. That precision is one reason why engineers like them. Precision, however, has a downside too. In the real world, dimensions and surfaces can vary enough from part to part to throw off a robot whose movements have been programmed relative to nominal size or surface. So would finishing or assembly operations that require a sense of "feel," such as assembling finely meshed gears, polishing a metal component, and some types of machining operations.
If you've ruled out assembly or finishing robots for these reasons, it may be time to take another look at them. Over the past couple of years, robot vendors have developed two technologies that improve robots' tolerance of the variability found in real-world assembly and finishing operations. One of these technologies gives robots a sense of "feel" by adding force control capabilities. The other uses machine vision systems to let robots "see" and adapt to variation in part dimensions and locations.
Taken together, force control and vision not only raise the technical capabilities of robots but also improve the economics of using them. Here's a closer look at both technologies and at some applications where they're starting to make a difference:
The addition of force sensors near the end-of-arm tooling and enhanced control software gives some robots the ability to adjust to the forces they encounter as they do their jobs.
ABB Robotics, for instance, has developed a system it calls Advanced Force Control, which can be added as an option to a variety of the company's robot arms. The system uses a force and torque sensor from ATI Industrial Automation at the wrist of the robot. "We feed information from the sensors into our axis controller and adjust force and speed," says Jerry Osborne, vice president & general manager of ABB Robotic Assembly in North America. "There's a lot going on in the controller to make this happen."
According to Osborne, the system can sense forces in six axes with a sensitivity of +/- 2.5 Newtons and with a response time of 4 millisec. The system can also work in conjunction with the robot's speed and position control – for example, by first running a search pattern to locate a feature or object and then switching into force control as the assembly proceeds.
In the past, Osborne says, one option for assembly applications in which the robot's contact force would cause problems was to add a compliant mechanism into a traditional position-controlled robot arm. Mostly, though, these force-sensitive applications literally stayed in human hands or went into complex, dedicated assembly machines.
ABB started to develop its force control system about three years ago, as a way to assemble the spline gear assemblies within automotive torque converters. "This work was manual because you had to feel the meshing of the spline gears," Osborne says, noting that 12 of these systems are now in operation worldwide.ABB is also pitching force control for tricky assembly tasks such as piston stuffing and spark plug assembly. Osborne says the system could make sense "wherever you have a press fit assembly."
Though developed three years ago for automotive assembly, the force control system seems to have even bigger implications for machining and finishing. Osborne says the majority of the systems have gone into finishing operations – such as polishing. "We have about 50 systems involved in finish applications. Some of them are involved in polishing magnesium laptop housings," he says. In these cases, the force control gets the nod for its ergonomic and quality advantages – it can provides a more constant force than a human being and not risk the injuries inherent in a long day of hand polishing.
Controlling forces at the end of the robot arm also have implications for robotic machining. Kuka Robotics Corp., which also has force-torque sensors available for its robots, has delivered systems that perform grinding and milling operations. According to Kevin Kozuszek, the company's marketing director, the force controlled robots are starting to become more popular in "pre-machining" applications--or the use of robots to perform rough machining operations, leaving only a single pass on CNC machine for finish machining. In this case, force control helps the robot close the gap with machine tool feeds and speeds by optimizing the contact forces between the robot-borne tool and the workpiece. Kozuszek says this approach can save significant amounts of money in reduced set-up and fixturing costs as well as in possible avoidance on capital expenses. “If you pre-machine with a robot, you may require fewer CNC machines for a given throughput," he says.
ABB's Osborne makes a similar case for pre-machining and adds that applications without tight tolerances may get away with robotic machining as a replacement for CNC. No one is suggesting that robot arms, which inherently lack the stiffness of a machine tool, will take precision machining operations. But Osborne and others see room for robots with force control to machine to tolerances near robotic repeatability – usually within a few thousandths of an inch
Kuka, meanwhile, currently advocates robotic machining for soft materials--such as aluminum or plastic. "It makes a lot of sense in prototyping and low-volume production applications, especially when you consider the time saved by avoiding complex fixtures and machine set-ups" says Kevin Kozuszek, the company's marketing director.
Still, one of Kuka’s customers has developed a new robotic machining cell that works on a hard material – stone. USMechatronics and the Seis Group, a pair of systems integrators that work on robotic and other electromechanical projects, recently created a stone-cutting robot called RoboJet. This stone cutting robot arm runs an abrasive water jet cutter, rotary saw and 3D milling head. Driven by proprietary control software from USMechatronics and by Kuka's CAMRob robotic machining software, this robot can switch between cutting methods automatically.
The stone industry already uses all three of these cutting methods--but on separate machines. "This is the first time they've had a robotic system like this," says Chris Barbazette, Seis' president. "There's been a tremendous amount of interest in it," he continues, explaining that the system can potentially replace all or some of the conventional, stand-alone cutting machines for a significant savings in capital costs. He adds that there's a utilization advantage associated with the robot. "The robot is always doing something, whether it’s cutting or moving materials into place," he says. That's not always true with stand-alone machines, which would typically have some idle time. And there's an obvious floor space savings, too.
Barbazette foresees the RoboJet approach becoming important in other industries that perform operations on stand-alone machines or processes. He thinks that composites fabrication is one application ripe for more robotic finishing and machining operations.And he's looking at metal machining applications too. "The whole idea of a robot as a machine tool is still in its infancy. I don't see it competing against CNC in precision applications because of the stiffness issue, but there are applications where robotic machining has a bright future," he says.
The other key enabling technology starting to gather steam is vision-guided robotics. These systems add a CCD cameras and lighting to the robot's end effector while specialized software translates images from the camera into move commands for the robot. Braintech Inc. has developed just such a system for ABB Robotics. Called TrueView, this off-the-shelf Windows-based vision guidance system can locate objects in 3D space with "sub-millimeter" accuracy, according to Jim Dara, vice president of Braintech.
Braintech has delivered TrueView systems to Ford, GM and various automotive suppliers. The common thread in many of these automotive applications – and in non-automotive uses too--is that vision can eliminate the cost of putting parts in correct position and orientation for robotic assembly. Dara points out that custom fixtures, precision dunnage and other positioning methods that "cost hundreds of thousands of dollars in a typical automotive plant."Robotic vision systems, by contrast, range from $10,000 to about $100,000, depending on their complexity.
"Robots can be too precise for their own good," Dara says. Vision helps them deal with objects as they are rather than as they should be.