Industrial Robots Fast, Nimble at 30

KEY WORDS

Machine control

Robots

Machine vision

Open architecture

Signal conditioners

Sidebars: Suppliers

FireWire for the factory

Exclusive: Signal conditioner aids motion-control circuits

More on robotics

Mention robots and even engineers may have pictures of Sony’s Aibo pet dog or the Jetson’s household helpers. Robots have also made news recently with products from iRobot Corp. (Somerville, MA) helping troops scour caves in Afghanistan and helping archaeologists search tombs in the Egyptian desert.

Perhaps some are even considering uses for the ‘Roomba’ vacuum cleaning robot.

Many, not-so-far-out applications are solving an ever-growing list of manufacturing problems.

In the 30 years since the invention of the SCARA (selective compliant assembly robot arm) robot, industrial robots have become workhorses of manufacturing. Heavy industrial six-axis robots continue to provide valuable work in automotive applications such as spot welding, de-palletizing heavy objects like engine blocks, and loading parts into dangerous equipment like presses. These include the Fanuc Robotics (Rochester Hills, MI) model pictured on the cover working at the Honda (Anna, OH) engine plant with integrated DVT (Norcross, GA) vision systems. SCARA robots also are found in smaller palletizing and de-palletizing applications for automated assembly equipment and other smaller assembly jobs often in semiconductor machine applications.

For those unfamiliar with SCARA robots, these are articulated pick-and-place units. Where pick-and-place devices typically operate in two axes-in and out x axis and up and down z axis-this robot type combines one or two pivot points yielding wide x-y coverage over an area plus the z axis. Along with a powerful controller, they feature high speed and high accuracy.

Overhead gantry, or ‘Cartesian,’ robots move on fixed x y and z axes, and they can be quite large for moving large components from one conveyor to another or from machine to machine. Use of commercially developed technologies like FireWire networking, improved encoders, and parallel processing controllers, discussed later, are allowing them to be used in smaller, high-speed and high-accuracy applications like component assembly and packaging.

‘Delta’ robots, a more recent development, combine new high-speed control capabilities with space-age materials to form a high speed, yet flexible, solution for picking, sorting, and other similar applications. An example from ABB Robotics (New Berlin, WI) is described in detail below.

Not all innovation in robotics is in glitzy commercial toys and appliances. There’s still a lot of opportunities left in industrial applications.

Kevin Gingrich, Bosch Rexroth (Buchanan, MI) marketing services manager, asserts, ‘Bosch invented the SCARA robot 30 years ago. In that time, we have been working to make the controls and operator interface easier to use. To do this, we went to PC-based control in 1998.

We use Wind River’s VxWorks real-time operating system (RTOS) and Microsoft Windows NT on top. This combination gave us reliable control capabilities, plus the ability to integrate operator interface, vision, and other Windows applications.’

System integrators are always seeking ways to add value to commercial systems for their customers, and this is no different in robotics. Mr. Gingrich points out that affiliated integrators asked for ways to tap into the RTOS so they can access dlls, OPC server, positioning commands, and I/O calls. Ethernet connectivity enables local or remote troubleshooting and program updates.

Further, Bosch has incorporated absolute encoders in its SCARA family, which eliminates the need for an automatic ‘homing’ command on start up. Given proper procedures and safety considerations, developers can build a cell just large enough for the work without wasting space so that the robot can access its entire reach during the homing cycle.

Cartesian coordinates

There are more kinds of robots than SCARA, though. Build a rectangular frame above a conveyor, add motion on the x and y axes, build a z axis with articulated end effectors and control and kinematics, and there is a gantry or Cartesian robot. These can be very fast with high accuracy. Uses include picking parts randomly scattered on the conveyor for palletizing or fixturing or small part assembly.

Challenges involved in building these devices include integration of control, servo amplifiers to the various motion components, component size, and control to servo communications.

Joe Campbell, Adept Technology (Livermore, CA) vp, addresses one challenge, see sidebar. Finding a reliable, yet very fast communication link allows use of distributed amplifiers. He notes that adding intelligence, communication, and power to the servo amplifiers enables design of smaller devices distributed on the machine, thus greatly reducing the control enclosure.

‘We are continuing to move the smart architecture,’ he adds, ‘by relocating power amps across the network and releasing a smart servo kit. The kit is an all-in-one control including full kinematics so that engineers can build their own robot.’

The idea of building a custom industrial robot may sound far-fetched, but with established robotic companies like Adept providing components, it seems much more feasible.

Another company provides a hardware and software platform that would allow an engineer to build a custom robotic solution for an OEM or special machine application. New Micros’ (Dallas, TX) small controller board named IsoPod contains a built-in high-level language and parallel processing operating system.

Its 12 pulse width modulated (PWM) outputs can each individually control a servo, or they can be grouped to control up to two brushless dc motors. It can be programmed to provide coordinated axes of control. The accompanying programming language, IsoMax, is based on the concept of state machines. Programming a real-time task entails describing virtual machines that sense conditions, take actions, and move to new states. ‘Virtually Parallel Machine Architecture’ (VPMA) allows more than one of these state machines to be installed and to run independently in a virtually parallel fashion. The effect is similar to running 20 stamp-like microcomputers, each with a single task, in parallel.

If any concept stirs controversy in the hearts of control engineers (at least among suppliers to control engineers), it is ‘open control.’ The discussion boils down to which platform provides better value to OEMs and end-users, open or proprietary. Open platforms are generally built upon commercial PC technology, which provides interfaces for many suppliers to provide products. In this architecture, the end controls designer builds a system almost from scratch. On the other hand, in a proprietary solution, one vendor assures that all system parts will work together.

In the case of robotics, so much of the control is specific to the entire system that it has been closed. Companies have gradually exposed some internal data for communication to automation systems. Some companies, however, have built open robotic controllers from the ground up.

One criticism of open control built on PC technology is reliability. Steve Sorensen, cto of Cimetrix (Salt Lake City, UT), points to the success of a packaging OEM that has built a high-speed, delta-style packaging robot using its PC-based control with a software PLC logic engine. Success of SigPak’s machine shows that open control can be robust and reliable.

Open control flexibility

Why use this type of control? Flexibility. This system closes motion loops on the robot, runs a software PLC for discrete logic, integrates a vision framegrabber, and accomplishes high-speed conveyor tracking. An integrator can build applications in IEC 61131, C, C++, or a combination of languages, according to Mr. Sorensen, enabling a company to write a custom sorting algorithm in C++, for instance, and expose it as a function block in ladder diagram. This gives a lot of power to OEMs.

Another reason to consider open control on a robotic project lies in the situation where a control retrofit is necessary. Many times, the mechanical parts of the robot are still in good order and the hydraulics are easily refurbished, but the controller may be woefully out of date, or even unsupported. In this case, an engineer must consider going to the original manufacturer for a control retrofit, scrapping the robot, or considering an open control alternative.

Walt Weisel, president of Robotic Workspace Technologies (RWT, Fort Meyers, FL), states, ‘We have seen our open control installed on an older robot and up and running over a weekend. Typically, the open control costs less than a used controller from the manufacturer. Plus the PC uses commercially available components. If one fails, replacements are easier and faster to obtain than with a proprietary controller.’

RWT’s robot controller is designed to be integrated on many robots. Its programming language is a derivative of Microsoft Visual Basic. Using a single language for different pieces of equipment benefits shops with robots from several manufacturers.

Integrating absolute encoders with robotics yielded benefits noted above. Other automation components add power and flexibility to traditional robots, as well.

Fanuc Robotics added 3-D vision with integrated force sensors in the gripper mechanism so that the six-axis robot can pick parts randomly stored in a bin or piled in a tray. By placing a traditional six-axis robot on a controlled rail, the company combines the benefits of six-axis and gantry robotics applications, covering a larger work area? geography than possible before. In addition, an Internet-enabled pendant gives more power to operators and technicians than previously available, to monitor and troubleshoot robots throughout the plant from one location.

The FlexPicker from ABB (see photo) incorporates carbon fiber materials in its arms to make an ultra-fast pick-and-place robot. Joe Crompton, ABB robotics integration project leader, states, ‘This robot is capable of 150 picks per minute with up to 15 g’s acceleration. The best applications for it are lighter payloads, for instance, typically 1 kg (about 0.5 lb). It is ideal for picking up small parts moving rapidly at random along a conveyor and placing in packing. Chocolates, cakes and other foods, electronic and mechanical components, and pharmaceuticals are examples of target industries.’

ABB’s FlexPicker is available integrated with Cognex (Natick, MA) vision systems. For nearly 20 years, vision companies and robotics companies have improved coordination of the two product areas to ease set up and use, as well as to open possibilities for new applications. George Blackwell, Cognex director of product marketing, adds that Ethernet connectivity has greatly aided this integration.

Vision sensor company DVT (Norcross, GA) provides a coordinate system within its ‘blob tool’ that can convert location into real-world values and transfer to the robot in x-y coordinates.

For those who need to move parts, whether loading engine blocks onto a conveyor or chocolates into a tray, there is most likely a robotic product that will provide a fast, accurate, cost-effective solution.

Comments? E-mail gmintchell@reedbusiness.com .

FireWire for the factory

FireWire (IEEE 1394) is gaining popularity in distributed control architectures, particularly for high-performance, complex motion control. Why consider FireWire over other motion buses like SERCOS or fieldbus derivatives?

In spite of its consumer heritage, FireWire brings significant advantages to industrial automation. It is low cost, extremely high speed, and offers functional signaling modes particularly well suited to advanced motion control. And it is unique in combining these features in a single architecture.

IEEE 1394-A delivers 400 megabit/second performance that is three orders of magnitude faster than most industrial serial buses and almost 1,000 times faster than various fieldbus networks. In fact, the latest implementation, 1394-B, competes with widely available backplane solutions, and offers control system designers an opportunity to truly distribute processing. In systems integrating machine vision and motion control, only FireWire will have the bandwidth to support multiple camera image transfer, side by side with complex motion control.

Another feature is Isochronous (equal interval) Transfer Mode. Advanced motion control, particularly coordinated motion control of a complex mechanism, demands strict time-based processing. This is critical for all closed-loop servo control as well as data acquisition and integrated machine vision and motion control. 1394 offers a highly deterministic network, guaranteeing not just delivery times, but the delivery order of each message.

Meanwhile, Asynchronous (not simultaneous) Transfer Mode enables response to external events with an acceptable latency. FireWire allows asynchronous event messages to be generated by any device on the network within any bus clock cycle, and those messages can propagate with a known and predictable latency. 1394-B has an asynchronous time window to support the most demanding applications: 62.5 microseconds, which is expected to drop 16

Unlike host-centric networks, such as Ethernet and USB, FireWire supports full communications directly from node to node. By bypassing varying delays due to central processing loads, 1394 is fast and deterministic in node-to-node communications. This peer-to-peer feature may also enable high-performance embedded applications without any PC presence.

Synchronizing a process start or stop over multiple nodes, or advising of safety violations, are common requirements for complex motion control systems. FireWire’s Broadcast Mode allows a message to be sent from any point in the system to all other nodes.

For more information,

EXCLUSIVE:

Signal conditioner aids motion-control circuits

‘Microanalog’ dc/dc analog signal conditioner from Weidmüller (Richmond, VA) provides important electrical protection in a slim 6.1 mm, DIN rail-mounted package. This analog signal conditioner has 3-port isolation, which is especially important in applications where the reduced isolation voltage is high enough to eliminate earth and ground loops with alow power level.

Positioning systems-the backbone of robotics applications-require elimination of noise from ground and control signals. Induced noise in a control loop will reduce accuracy and system performance. This signal conditioner can easily be used for isolation and/or conversion of analog signals, while saving valuable panel space.

One module fits most standard functions. Inputs and outputs are DIP-switch selectable for 0-10 V, 0-20 mA, or 4-20 mA. Further, the module is undamaged when DIP switches are changed with power applied.

Cross connection for power supply and ground connections eliminates redundant power and ground connections to each module, speeding installation.

Both screw or tension clamp versions are available. Low power dissipation of 0.6 W allows ‘side-by-side’ installation in panels.

More on robotics&/NAME>

For additional coverage of robotics see Control Engineering Online at

Do you have experience and expertise with the topics mentioned in this content? You should consider contributing to our WTWH Media editorial team and getting the recognition you and your company deserve. Click here to start this process.