The PC controls of the robot were excellent for communicating with the operators. This is vital when people work intimately with machines. One of our early developments was to embed an Internet Explorer window into the cell's user interface. People wanted to control the font size and color of the messages displayed by the interface. Web technology was the easiest way to provide this and much more. For example, it allows us display work instructions on the machine's screen. The robot program determines which Web page to show, so operators have the information they need when they need it. Only relevant information is displayed, so visual clutter and confusion is minimized.

A visual display with text is significantly better than the usual cryptic numeric error message that many machines output when problems arise. Conventional work instructions, which typically consist of a document in a plastic sleeve held by a magnet on the side of the equipment, pale in comparison to what is commonly done now. Operators can be shown what to do with text, audio and photos that have arrows to highlight the task. The display can include links to Web pages, so operators can get more information if they need it. The ubiquitous nature of the Web means that operators don't need to be trained how to use the interface. And, because the Web supports many character sets and languages, the interface can be customized for the site where the machine will be used.

Simplifying Automation

One important feature of the cell is its simplicity, which can be an elusive quality in an assembly machine. Many people believe lean design is incompatible with automation, because the added support costs negate the labor savings. Our challenge was to design a cell in which the equipment is never down and the processes always work correctly. The key was to keep things simple mechanically and shift as much of the functionality to the robot as possible. Robots now are very reliable and will run almost indefinitely with little maintenance. We just had to make sure the application tooling didn't become the Achilles' heel of the machine.

Our philosophy was to make the cells behave similarly to a person, a concept known as "anthropomorphic design." In one application, for example, the robot inspects a few hand-placed electronic components before soldering. In the original assembly line, operators passed pallets between stations on a simple sheet-metal slide. So, we made the robot do that, too. When a pallet arrives, the robot uses a tooling finger to pull the pallet into the cell on fixed rails. It does the inspection and then pushes the pallet out. The chance of a conveyor belt or motor going bad is zero, because they simply aren't there. This illustrates the concept of using the robot for more than just the core process of the machine: You paid for it once, so why not use it as much as possible?

In a number of instances, the robot does the motion typically performed by other actuators. Traditional automated assembly machines consist of motors, belts, pneumatic actuators and other components that perform small motions in some sequence to produce the desired effect. In a robotic workcell, the path of the end effectors in space and time is entirely flexible and is accomplished with fewer components. This approach can be called "four-dimensional design." Looking at the idle machine, it's not obvious how it might work. But, when it runs, it comes alive and the puzzle reveals itself.

In-House Design

Delphi Electronics & Safety is one of the few manufacturing companies to have kept in-house expertise in equipment design and construction. This provides an advantage because we know the needs of our manufacturing operations in great detail. With a short walk down the hall, engineers can talk directly with operators and get immediate feedback. Maintenance personnel can call if there's a problem, so we know all the little things that are needed to keep equipment running well. Engineers on future program teams can stop by to talk informally about what they will need, and we can show them similar machines.

This synergy doesn't come easily with a systems integrator. Many design details are based on input from the skilled tradespeople who build and support our equipment. This input is invaluable for bringing a pragmatic viewpoint to the table when we're designing new machines. We also have a development lab with a robot cell where new ideas for a process can be tested quickly.

Machine design requires trade-offs between many elements. Everything affects everything, so the engineer must simultaneously comprehend mechanics, software, electronics, pneumatics, reliability, maintainability, ergonomics, graphics and logic. You have to keep the arc of the entire project in your head to come up with a truly elegant solution. As one designer says, "The hardest machine to build is the simplest one." It's important to envision the entire system working, and to do that from multiple users' points of view. Otherwise, you end up picking at pieces, and you lose sight of how these pieces are wedded to an entire concept. The development lab is extremely useful in this regard, because cycles of learning happen much more quickly. Ultimately, this is more cost-effective than going back and forth with an integrator.

As the saying goes, "hindsight is 20/20," so we strive to get hindsight as quickly as possible. Building a working prototype of a robotic cell allows us to distill the essence of an assembly process. It's almost impossible to do this without some iteration, discussion and banter with those involved. The process may not be predictable, but it allows invention to thrive. As we have taken on more projects, our engineers have developed an intuition about what will work and what won't. This is advantageous, because new mechanisms can be tested quickly before much money is spent. Bigger risks can be taken since the cost of failure is low. In many cases, this has led to leaps in process efficiency. As an added benefit, we can demonstrate working examples to managers to secure funding for production machines.

Building Blocks

There are many reliable components with which to build a machine today. Choosing the right one is not so much a matter of which component is best, but of maximizing functionality while minimizing part numbers and ensuring worldwide support. With about 150 approved components, Delphi engineers can build nearly any machine. Standard platforms are used extensively in surface-mount equipment, and the benefit applies just as well to final assembly and test operations. We now have machines that dispense, drive screws, inspect, and perform test handling. All share the same user interface, software environment, machine base, electrical structure, utilities and overall approach to design.

There's a misconception that innovation requires each machine to be different from the previous one. In fact, there's a certain art to using a small component set with the right degrees of freedom in the right places. There's an art to using these components over and over again for new applications.

The CIMPLE cell is a narrow machine for manual or semiautomatic lean stations. Made entirely out of aluminum, the monocoque base has front, back and side plates of aluminum to provide a rigid box for the tabletop. The machine is light enough that two people can pick it up, yet it's rigid enough for the robot to run at full speed and acceleration.

The CIMPLE cell program has necessitated breaking some of the traditional rules for automated assembly systems. In these robotic systems, the design trade-offs between the mechanical, pneumatic, electronic and software portions of the system are different. Sometimes, actions that used to require specific mechanisms can be done entirely with software. This flexibility hasn't been available until now.

The PC's flexibility is easy to harness. Our motto now is, "If you can think it, you can do it." Because of the consistency of the platform approach, it's easy to add programming for new ideas and apply it to machines quickly while spending little or no extra capital. Also, because of the PC, the cell can do things that traditionally are thought of as IT-related functions. E-mails can be sent when the material supply in a dispenser gets low. Statistics programs can analyze process data and show the results graphically. When any software changes are made, the PC automatically zips up and copies all the code to a server, so there's always a backup. Machines and testers now can handshake via Ethernet instead of discrete I/O, eliminating many wiring problems. Web pages can show drawings, spare parts, PDFs, and the Web sites and contact information for component suppliers. You can't do these things with PLCs or proprietary robot programming environments, but you can with software and a PC.

Systems Are the Future

We believe manufacturing equipment can be designed to perform complex activities while keeping the mechanics simple enough that they're reliable and cost-effective. This approach yields simplicity and sophistication, but not complexity. It is broadly applicable, and not just a set of prescriptive techniques that is forced onto all situations. It hasn't been codified into a formula, but the elegance of the results is readily apparent.

Innovative machine designs like our CIMPLE cell rarely happen in rigidly systematic environments. Rather, serendipity must be allowed to exist. In the course of discussion and mock-ups of assembly systems, we often find that unexpected ideas pop out. Inventions rarely happen in a meeting room or office.

Good automation components are readily available on the market, but the winners of tomorrow will be the companies that can combine them together most effectively. Like a talented musician whose gifted musical interpretation takes notes and rhythms to produce music that delights the ear, machines that are designed holistically have a certain beauty to them. As David Gelernter writes in Machine Beauty: Elegance and the Heart of Technology, "The beauty of a [mathematical] proof or machine lies in a happy marriage of simplicity and power-power meaning the ability to accomplish a wide range of tasks, get a lot done. The power and simplicity criterion applies to birch bark canoes, suspension bridges, programming languages, scientific theories, and machines of all kinds."

It's what we call CIMPLE!