Once viewed as a threat to American workers, robots are fast becoming essential productivity tools for keeping U.S. factories running at a healthy clip.
In one of the biggest increases in years, North American-based robotics companies saw orders surge by nearly 25 percent in 2007, according to the Robotic Industries Assn. The industry now has grown to the point where some 178,000 robots are at work in U.S. factories, making this country second only to Japan in robot applications.
Though vastly different in design and purpose, such robotic systems, say engineering experts, share one thing in common: They're the perfect test beds for mechatronic design. The engineers who develop them blend a wide range of engineering disciplines and harness the latest modeling and simulation tools before committing their creations to hardware. That approach was evident in robot applications Design News recently examined from two leading robot companies, Adept Technologies and Motoman.
Adept Technology: Fastest Robot in the West
Based in Livermore, CA, Adept Technologies has earned a global reputation for vision-guided robots. Applications range from assembly and testing to life sciences and medical fields, including a new partnership with UK-based Prosurgics to develop a next-generation robotic surgical system. Earlier this year, the company reached a major milestone, having shipped a total of 25,000 robots.
In terms of potential applications, the company's new Quattro robot looms especially large, since it satisfies an overriding customer requirement: speed. “Quattro offers performance that is twice that of conventional robots,” says Adept President John Dulchinos, “giving companies twice the productivity out of the same amount of factory space.”
Adept is targeting Quattro for fast-paced packaging and handling applications, such as loading wrapping machines, cartons, trays and boxes, as well as conveyor-to-conveyor transfers. Featuring integrated vision and tracking, the robot can pick parts from moving conveyor belts at the rate of up to 180 per min.
As Project Manager Matt Bjork points out, Quattro addresses an unmet market need for extremely fast cycle times across a 1.3m diameter work space and 2-kg payload. He says SCARA and other 6-axis robots on the market typically have “dead spots” that they cannot reach. Previous parallel robots also have displayed poor performance at the edges of the workspace on payloads of more than a half kilogram.
Quattro, with its four fixed axes, typically mounts on a rigid overheard frame above one to four conveyor belts. Its integrated vision system locates the parts on moving belts and the robot decides when to pick up the part by tracking it through a belt pick-up window.
A typical application, according to Jeff Baird, director of engineering, might involve randomly placed cookies coming out of an oven. The robot first correctly orients the cookies on a conveyor belt, then places them into trays on a second conveyor before the package is sealed.
Finding Creative Solutions
To achieve Quattro's ambitious performance goals, the design team needed to address the high forces in the robot's outer arms and mounting platform. “We had to measure these G forces and achieve materials fatigue strength over 500 million cycles to meet our design criteria,” says Lead Mechanical Engineer Daniel Norboe. “An extensive amount of research went into finding and characterizing the material and design requirements.”
Among other considerations, the team had to minimize the moving mass of the robot, as well as keep heavy motors and gears from having to be “carried” by the kinematic chain. This involved specifying lightweight components, such as carbon fiber, for robot links. The result: A very stiff design with little structural oscillation. What's more, Quattro achieves 15 m/s/s accelerations and a top speed of 10 m/s across the center of a typical long move.
Quattro also features a rotation clocking mechanism capable of 360 degrees. Built into the moving platform of the robot, this device addresses mechanical design issues associated with achieving reliable rotation in parallel robot designs. Project Manager Bjork explains the basic math equation involves a robot with four independent motors and arms. Mathematics, combined with solid ME design, allow for motion of a single joint while the flange remains parallel (no tip or tilt). For smooth linear tool flange motion, all four joints must be tightly coordinated in a real-time control system. The four independent actuators provide four degrees of freedom: X/Y/Z position and Theta part rotation.
Still another mechanical design consideration was a requirement that the robot be sealed water-tight for wash-down after packaging meat or fish. All external surfaces had to stand up to caustic cleaning agents and any wear particulates needed to be food-compatible. Rated for Clean Room ISO 6 or better, the design demanded novel coating requirements, since casting porosity can lead to early fatigue failure.
Meanwhile, electronics engineers tackled the challenge of combining high speed and acceleration with precise positioning in a compact electronics package. Adept's high-performance “Amplifier-in-base” (AIB) technology, originally developed for the company's Cobra robots, provided the perfect answer, according to Lead Electronics Engineer Deron Jackson.
The AIB, located within the robot's base, is coordinated by Adept's compact SmartController CX hardware, running industry-proven V+ software. The controller can coordinate multiple Quattro robots, using Adept's SmartServo distributed architecture. Customers can link several SmartControllers via Ethernet to form integrated robot lines of any size. The overall production line determines interface requirements for the Quattro. Some production systems prefer to interact through Digital I/O on fieldbus technologies like DeviceNet; others use customer-developed PC interfaces or PLC architectures. To facilitate these interfaces, Adept offers products like Adept ePLC Connect and Adept Packaging Manager.
As for vision options, Adept offers an extensive vision library embedded in the SmartController CX, in the PC platform (AdeptSightTM) or in smart camera technologies, says Travis Armstrong, Quattro lead systems designer. The vision library includes advanced vision operations and such features as the Object Locator, a contour-based tool for locating parts. Adept also supports third-party vision hardware, including IEEE 1394, Line-scan and Gigabit Ethernet cameras.
“The tight integration of Adept's motion and vision allows for accurate location and tracking of customer parts at any instant of time and at high conveyance speeds,” says Armstrong.
Mechatronics and More
Blending the various engineering disciplines to develop Quattro was perhaps the easiest part of the project, according to Engineering Director Baird. “Mechatronics is Adept engineering,” he says. “Our product offering has always involved solutions that use mechanical design, electrical design, embedded software, advanced servo/motion control and application software tools.”
Besides plenty of interaction among engineering functions, the Quattro team tapped a number of software tools. Among the most valuable, recalls Lead Kinematician Paul James, was Adept Digital Workcell. The 3-D simulation and off-line programming software package optimized the design before it went to hardware. For example, the software helped solve questions involving the forward kinematics of the Quattro, including the proper positioning of the robot's tools. The mathematics involved iterative schemes and solutions of quartic equations. Engineers developed, simulated and tested the algorithm on Digital Workcell prior to implementation in the Adept SmartController CX.
The mechanical team used SolidWorks CAD to lay out the entire robot in 3D and perform clearance and work envelope studies. The software also helped determine the correct positioning of the motors and drive train inside the robot base. In addition, FEA tools, such as ANSYS and Cosmos, determined material fatigue life.
“ANSYS and Cosmos analyzed stiffness and stresses in all the critical areas of the robot,” says Mechanical Engineer Norboe. “For example, we analyzed the stresses inside the platform to make the platform lightweight, yet strong enough to handle the high forces encountered during the Quattro's amazing accelerations and speeds.”
Ultimate Goal: Satisfied Customers
The Quattro's design achievements are all the more significant, given the team's tight schedule. “We started with a basic concept and some research from a university,” says Project Manager Bjork. “Within three months, we had the robot 80 percent redesigned and prototypes fabricated.”
Adept exhibited a Quattro prototype at a trade show, moving billiard balls around the workspace at eye-popping accelerations and speeds. Then came more validation testing and second iterations on initial design choices. In all, from concept to the first production shipments, the project took just nine months. “This was a tremendous effort that is only possible in an environment of highly skilled engineers where all disciplines of engineering work together closely, minimizing design turns and lengthy decision processes,” says Bjork.
In short, adds Bjork, Quattro's ramp-up was the fastest in the company's history, with more than 200 units sold in the first nine months. Early customers include companies in the food, consumer goods and pharmaceutical markets. In March, Germany-based Solar Line Saxony selected the system for solar cell production. “We chose Adept and the Quattro robotic because they enable our systems to achieve unprecedented performance,” says Karsten Barth, CEO of SLS.
Motoman Robots: They're Team Players
Among the world's robot manufacturers, Motoman can lay claim to being one of the most prolific. A subsidiary of Japan's Yaskawa Electric Corp., Motoman's product line offers more than 175 robot models and 40 pre-engineered solutions.
With U.S. headquarters near Dayton, OH, Motoman maintains application and project engineering teams that can design, build and integrate customized robot systems. Among the many examples of tasks these robots perform:
Sorting up to 950 blood sample tubes per hour at a testing laboratory.
Deburring an extruded aluminum component in just 23 sec.
Loading automotive filters on pallets at the rate of 16 cases per min.
Increasingly, Motoman engineers are designing cells that feature two or more robots working together — a trend the robotics industry describes as “cooperative robots.”
Case in point: A new material handling and welding system for Pennsylvania-based New Holland Agriculture. The farm equipment manufacturer needed a new welding solution for handling 18 different hay baler roll assemblies. These components consist of tubes with trunnions on each end that are tack-welded in place, plug-welded and then final-welded around the entire circumference of the end seam. Parts are made of 6.0-6.6-mm thick ASTMA500 steel.
Although the previous welding system for these assemblies used robots, parts had to be manually loaded, unloaded and clamped using traditional weld fixtures. Achieving good welds on smaller diameter rolls was especially difficult. Operators had to check 100 percent of assemblies to be welded and then straighten them to maintain run-out requirements. Even so, misalignment of components resulted in extra operations to grind and repair welds manually. If undetected, misalignment can lead to vibration, which causes wear and potential weld failure in the field.
Together with Motoman engineers, New Holland tackled these problems with a new, fully automated cell that features two robots. A Motoman HP200 material handling robot loads and unloads a mechanical slide fixture mounted on a headstock positioner and a Motoman HP50 performs the arc welding. Holding enough parts to support up to two hours of unattended production, the system improves part run-out tolerances and weld quality, while reducing labor related to loading and unloading parts. It also eliminates the need for checking roll run-out 100 percent and straightening rolls. Misalignment between the tube slots and trunnion discs is rare, so grinding and weld repairs have been virtually eliminated.
“We looked at many alternatives, including machining of the tubes and trunnions and re-design of the rolls, but these approaches would have further increased part cost,” says Bob Burkholder, operations specialist, Manufacturing Processes and Technology for New Holland. “Working with Motoman, we developed the current concept that provides even better part quality and higher savings then we initially thought we could achieve.”
Inside the Work Cell
Tubes enter the robot cell via a gravity feeder device, while the trunnions enter through a Motoman MR-300 rotary positioner. Two fixtures (one per side of the positioner) each hold up to 24 trunnions. The 6-axis HP200 robot, with a 200-kg (441-lb) payload and 2,651-mm (104.4-inch) horizontal reach, performs the handling operations within the cell. Custom end-of-arm tooling transfers one tube and two trunnions. Two large grippers simultaneously grasp the tube, while two sets of smaller gripper jaws hold individual trunnions.
“The use of self-centering jaws on both the weld fixture and robot manipulator to hold the trunnions and the tube significantly reduced run-out variances,” says Clint Hinkle, the welding/manufacturing engineer in charge of the robot cell. “This method assures that the trunnion will always be inserted into the center of the tube, despite material and part variances.”
Myke Dziengel, Motoman's lead engineer on the application, adds that trunnion run-out was the most important factor for the project, with weld quality second. “The tooling provided the necessary alignment for the trunnions, with the horizontal weld positioning providing the weld quality,” says Dziengel.
Also important to system accuracy, says electronics engineer Len Brezinski, is a laser sensor mounted on the gripper. Together with a sophisticated software search routine, the laser finds holes and slots in the curved tube surface. The tube is then rotated to locate the holes or slots in the proper position for welding. To eliminate any misalignment errors due to damaged slots or dirt in the slot, the sensor also measures slot length. If the slot is not the proper length, the sensor moves to the next slot. The sensor also verifies the part number, as well as the tube length and diameter, based on the location and presence or absence of specific part details.
Prior to being loaded onto the fixtures, another laser sensor at an inspection station checks each trunnion part to determine the length of the welded trunnion shaft and its position relative to the gripper. This information determines exactly how far the mechanical slide portion of the fixture will insert each trunnion into the end of each tube.
The robot controller commands the headstock to reposition the parts during welding to facilitate robot access to both sides of the tube for the tack-welding, plug-welding and final-welding operations. The slide fixture has the same type of Yaskawa servo motors as the robots and these motors are also controlled by the robot controller. The ends of the slide fixture open and close, allowing the HP50 robot to perform its welding operation. Once welding is complete, the HP200 robot unloads assemblies onto a gravity outfeed chute. Changeover between part models is simple, typically involving selection of another robot program.
Controls and Design Tools
In the New Holland application, both Motoman robots are controlled by two NX100 DR2C robot controllers, interfaced as a single controller with one programming pendant. Should New Holland decide to reconfigure or redeploy the robot cell, this controller configuration provides additional layout flexibility that would allow the two robots and controllers to be separated. One NX100 controller can control up to 36 axes of motion (up to four robots with a single programming pendant). For overall cell control, the application uses an Allen-Bradley SLC 5/05 Programmable Logic Controller (PLC), along with a MotoHMI user interface.
As for software tools to design the system, Dziengel says the team relied on Solid Edge for overall system design and turned to the Solid Edge FEA tool for analyzing stresses in the most critical areas of the tooling. He adds that MotoSim software was instrumental in situating the two robots and the positioner precisely within the cell, since part sizes can be quite long (up to 77 inches for final welded assemblies). “Robot reach was critical,” says Dziengel. “Placement of the robots a half inch either way was the difference between success and failure.”
Project success also depended on harnessing a mix of engineering disciplines on the Motoman team, along with close cooperation with New Holland. “Some members of the team had prior experience welding trunnions and tubes and provided insight on the difficulties of meeting the design requirements,” says Dziengel, “while others provided the mathematical skills needed to calculate the forces required to withstand weld distortions, grip forces, frame and slide sizing.
A Leap Forward on Quality
In the end, the Motoman system exceeded New Holland's requirements for quality improvement. It consistently holds run-out tolerances of less then 0.2 mm (0.008 inches) on trunnions and 0.75 mm (0.030 inches) on tubes. That's twice as good as the project's requirements. Higher-quality parts also have reduced the need for extra finishing operations and associated direct labor costs.
In addition, the new cell surpasses project goals that the system support one hour of unattended production. What's more, the cycle time of five to six min per finished part is expected to improve as the plant implements additional tooling and part changes.
“The cell was installed in late August 2007 and has been running without any significant problems,” says Hinkle of New Holland. “Except for weld gun maintenance, the cell runs the entire shift, through breaks and the lunch period. It was a real team effort between Motoman and New Holland to get us where we are today with this cell.”
Motoman Staff Writer/Editor Mary Kay Morel contributed technical details and engineer comments for this case history.