0
Select Articles

Flexible Automation on the Line PUBLIC ACCESS

Advances in Hardware and Software mean Today's Flexible Automation Systems Can Handle a Wide Array of Manufacturing Tasks with Few Configuration Changes and Little Downtime.

[+] Author Notes

Associate Editor

Mechanical Engineering 120(04), 70-72 (Apr 01, 1998) (2 pages) doi:10.1115/1.1998-APR-5

This article discusses developments in flexible automation in the automobile industry. As industry demands growing emphasis on agility, the need for systems with flexibility at scales is becoming critical. Flexible automation is an operational response to this need. Traditional automation systems were extremely limited in scope. Each would do a single task very well over and over again but had little or no ability to adapt to any other task. Although developments in hardware have made automation more flexible, software tends to be the more critical aspect of the system. Much of the flexibility of today’s system comes from the reprogrammability of devices such as tools and material-handling equipment. The area in which flexible automation is most pervasive is the automotive industry. This industry not only has a great need for automation systems but also has the budget to afford such expensive and elaborate systems.

As industry places growing emphasis on agility, the need for systems with flexibility at scales not previously possible is becoming critical. Automakers, for example, traditionally produced only one car model on a particular assembly line, and that model usually would be made continuously for at least several years, so an inflexible system did not present a problem. Nowadays, however, competition is dictating that automakers manufacture several models on one line, so more.., versatile equipment is needed.

Flexible automation is an operational response to this need. "Manufacturing is shifting fro111 attempting to exploit economies of scale to exploiting economies of scope," said Michael Higgins, marketing director of ABB Flexible Automation Inc. in Auburn Hills, Mich. "As a result, the ability to adapt to changing market requirements, product designs, and technological developments is becoming a key factor in the competitiveness of an organization. Viewed from the shop floor, this translates into dealing efficiently with frequent changeovers of parts and small production batches."

The ideal system uses reprogrammable processing, material- handling machinery, and computer coordination of cycles to enable the simultaneous production of different part types with zero on-line setup time and costs. No system has thus far lived up. to this ideal completely. Still, new developments in hardware and software are bringing state-of-the-art systems closer all the time.

Traditional automation systems were extremely limited in scope. Each would do a single task very well over and over again but had little or no ability to adapt to any other task. For example, when an automobile manufacturer needed a robot welder on an assembly line to work on a different part, technicians and engineers h ad to make significant hardware and software changes. Besides being labor- intensive, such change overs typically required extensive downtime.

For years now, industry has successfully implemented computer-driven, flexibly automated manufacturing systems. These systems tend to be small in both size (the number of devices under computer coordination) and scope. They are typically con fined to a single part family or a few similar ones, so there is a restricted part flow within the system. These restrictions are due in part to a lack of underlying models for the operation and control of such systems.

Large-scale flexible manufacturing systems have inherently different problems. Given the high off-line setup costs, large scales of operation cannot be economically mapped to a set of independently functioning small systems. To support manufacturing operations of larger scale and scope, flexible automation systems must be able to scale up. Large systems will also require these devices to be organized into subsystems. The subsystems will have to be dynamically con figured, and their operations must be properly coordinated. This makes system-level flexibility the central problem, rather than device flexibility.

Although a five-axis tool manipulator is more expensive than a less versatile counterpart, it can accommodate a wide variety of part shapes and sizes, enabling a manufacturer to make more products economically.

Grahic Jump LocationAlthough a five-axis tool manipulator is more expensive than a less versatile counterpart, it can accommodate a wide variety of part shapes and sizes, enabling a manufacturer to make more products economically.

A database stores multiple instruction sets that control nozzles on each side of this powder-coating system, which coats parts moving down the overhead conveyor.

Grahic Jump LocationA database stores multiple instruction sets that control nozzles on each side of this powder-coating system, which coats parts moving down the overhead conveyor.

Several advances in hardware have helped nuke flexible automation systems feasible and affordable. The most significant advance is the fast, inexpensive microprocessor. Even the least sophisticated flexible automation system requires a great deal of computing power. Just five years ago, a system with a 20-megahertz processor and 100 megabytes of storage was considered powerful; now even the lowest-cost personal computer offers much more. "What costs under $ 5 ,000 today would have cost $100,000 10 years ago," said Al Harlow, executive vice president of ARTomation Inc. in Cleveland. " State-of-the-art computers have opened up flexible automation to a wide variety of industries. Before, only the very biggest, such as automakers, could afford it."

Another boon to flexible automation has been the advent of more-sophisticated robot arms and manipulators. Early robot arms had almost no adaptability, and they were very ' difficult to maintain. Furthermore, programming was comparatively primitive, requiring an interface through a means like punch cards. By contrast, today's robot arms are much more dependable because they are driven by more-reliable electric servo drives in all but the largest applications. Encoders keep track of positioning, enabling the end effector to be positioned precisely and consistently. Users can interface directly with the robot through a control pad or computer terminal.

Flexible automation also has been aid e d by the proliferation of mechanical devices on the market. Years ago, most automation solutions required a custom-built machine with parts and attachments that had to be custom- machine d by the manufacturer or the user; as a result, the design and construction of an appropriate device was relatively expensive. Today, many vendors sell equipment such as linear actuators, ball-screw drives, and belt drives, so a staff with technical training can put together a basic automation solution fairly easily. Even the most complex machines and configurations are far less expensive than they were just a few years ago.

Although developments in hardware have made automation more flexible, software tends to be the more critical aspect of the system. Much of the flexibility of today's systems comes from the reprogrammability of devices such as tools and material- handling equipment. Operation of a flexible automation system is specified by a series of device programs-typically numerical-control programs.

Traditionally, any change to a system resulted in a major software bottleneck. If a robot welder needed to learn how to handle a new part, for example, the robot controller would require extensive reprogramming over a series of steps. Operators would enter these steps into the computer, or in some cases an operator would guide the robot through the steps and the robot would remember the path. Once ready, the basic set of instructions had to be tested; programmers then had to refine the instructions to remove any bugs.

Besides taking a lot of time, these steps required well trained, knowledgeable personnel. To be most effective, these workers often had to know the idiosyncrasies of the particular robot and controller, and new employees. tended to have relatively long learning curves. After all the time and effort, the new program would only be valid for one particular part, and another part would require everyone to start again from. step 1. If the manufacturer did not have a spare robot specifically for training purposes, the line had to be shut down completely while the robot was retrained. Few companies can afford that much downtime.

Several software advances have made reprogramming much simpler and quicker. Traditional programming has been done at a motion level-a device like a robot would have its task specified by a series of motions. Such programming is time- and effort-intensive, typically inflexible, and almost impossible to generate and maintain for large flexible manufacturing systems. To support large-scale flexible manufacturing, device programming is now migrating to a task level. For example, instead of a programmer telling a robot to move from point A to point B to point C, an operator would just give one command telling it to move along the length of the piece. Contained within the software is the built-in intelligence to interpret this move command correctly.

"What we are seeing more of is emulation versus simulation," Harlow said. "Rather than having to write code, users now have the ability to characterize the behavior of the process through a series of pull-down menus." GM Nameplate Inc. in San Jose, Calif., uses robots to spray-paint parts for a uniform coating of specified thickness. Although this would seem relatively straightforward, a large nU1nber of variables must be taken into account. In the past, the programmer would have 1;0 specify air-line pressure, atomizer pressure, paint-flow rate, and nozzle speed and direction. After all that work, the end result might still not be satisfactory, because external variables like temperature and humidity can also affect quality.

By contrast, control software today is evolving into a form of expert system. Based on empirical data, GM Nameplate's system learns how paint quality responds to particular variables. Eventually, enough data are in the central data bank for the software to perform actual behavioral analysis. Once this knowledge base has been built, all the programmer needs to do is specify paint thickness, which is what the company is concerned about in the first place. The software will then use its built-in knowledge base to write the appropriate control algorithm. An algorithm for each new part or thickness can be developed within minutes. GM Nameplate uses five-axis machines that typically paint 20 different parts. In a new model year, the company would have needed to write 20 new programs; now it just plugs in 20 new shapes and coating thicknesses.

The interaction between automation systems and computer- aided design systems is also increasing. Given a set of instructions in the traditional way, a computer has no real knowledge of the size or shape of the part it is working on; it just knows that it needs to move the tool along a prescribed path. Interfacing with a CAD system is another shortcut that spares programmers from having to write code. By importing the object geometry from a CAD system, software can automatically write positioning code and process code to achieve the result specified. Systems can handle either two- or three-dimensional drawings. Some companies are evaluating the possibility of simplifying the process even further by configuring software to work with an image from a digital camera instead of a CAD package.

Whereas traditional automation systems could only handle one part, existing technology has made it feasible for them to handle batches of the same part. The systems under development, however, will be so versatile that they can work effectively even if every part is different. Genesis Systems Inc. in Davenport, Iowa, is developing a system that identifies parts in real time using a bar code or magnetic tag. As a part moves down the line, vision devices automatically identify the tag. Contained in the system is a database that specifies the proper control program for a part with that tag. By the time the part makes it to the tool, that tool is ready to perform the appropriate operations on it.

A part cannot be positioned exactly the same way every time, however, so the vision system may not always be able to identify it. Such part-identification systems have improved enormously in recent years, and many can handle position deviations. Furthermore, the software can be configured to adjust the program slightly to account for minor deviations in position.

Eventually, operator intervention will not be needed in the reprogranu11.ing process. Information about a particular part won't even need to be stored in the database, because all the information about a particular part will be taken solely from the vision system. The system will look at the part and write the control program on the spot. Theoretically, the system can then work on any part, as long as the part's dimensions are within the range that the hardware can handle. Few applications currently require this degree of versatility, but with the advent of mass-customization products such as custom-made blue jeans, the need is increasing.

The area in which flexible automation is most pervasive is the automotive industry. This industry not only has a great need for automation systems but also has the budget to afford such expensive and elaborate systems. However, as system prices decrease steadily, flexible automation is beginning to penetrate virtually every sector of manufacturing. "Eventually, anyone on a factory floor will be able to use a flexible automation system just like he or she uses any other tool," Harlow said. "A significant benefit of this is that companies will no longer need to worry about a key employee retiring or changing jobs. The knowledge will be retained in the software and accessible to any authorized user."

Vision-based sensors and leading-edge software allow this spray-painting system to handle different parts with minimal downtime.

Grahic Jump LocationVision-based sensors and leading-edge software allow this spray-painting system to handle different parts with minimal downtime.

Copyright © 1998 by ASME
View article in PDF format.

References

Figures

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In