Designing Your Products for Robotic Automation

Success depends on attention to details

Designing products for robotic automation requires switching from a linear to a concurrent thought process and attention to details.

'If you don't have a product that's designed for automated assembly, you're not going to realize the full potential of it,' warns John Rueping, associate division director, Manufacturing Systems Technology Division at Eastman Kodak, Rochester, NY.

That means involving automation personnel at the concept stage so the right concurrent re-engineering and system engineering decisions can be made. 'Nothing will give technology a bad name faster than deploying it without being able to support it adequately,' explains Rueping.

Planning automation concurrently with product design will help avoid what many robotic automation specialists see as the biggest pitfall in transitioning from manual, semiautomatic or hard automation processes. 'People try to apply new tools to old methods and processes and haven't thought out what it is that they are really trying to do,' says Arthur Adlam, director for virtual prototyping at the U.S. Army Tank Automotive Armaments Command, Warren, MI. 'It's essential to look at re-engineering the way you do business or the promise of robotic systems may not be realized,' he adds.

Inevitably tradeoffs are necessary. For example, says Adlam, 'Design for repairability may not be particularly compatible with robotic assembly.'

When designing for robotic automation, there's a host of details beyond product function and cost, which should be considered simultaneously. 'Products must be designed with an understanding of the manufacturing systems and processes in mind,' says Brian Huff, associate professor, Industrial and Manufacturing Systems at the University of Texas at Arlington. For example, if a SCARA robot is to be used, then the product must accommodate top-down assembly because this type of unit is not particularly compatible with off-axis construction.

Other concerns include parts feeding and fixturing, sequence of operations, future maintenance and repairability requirements and part variability. Some forethought enables designers to 'include features that can significantly improve feeding and not affect fit or function,' says Huff.

It's also possible to include safeguards to mistake-proof the system. This could be something as simple as different color parts for different models to simplify changeover and allow quick confirmation that the correct product is being built.

If more than one product is involved, flexible fixturing helps speed changeover and can make the system easier to reconfigure for a new product.

Since robotic units are not particularly tolerant of dimensional variations, part variability must be controlled within tight tolerances or accommodated.

On the robotic system side, it's important to build in flexibility and modularity, so the unit can be reconfigured easily and reused for the next generation of product. Since it takes about two years to design and build an automated assembly system and product life cycles have compressed to the point they are measured in months rather than years, it's essential an automated system be flexible and modular so it can be used for a succession of products. This reduces capital costs, improves return on investment and shortens time to market.

Another system consideration is balance. 'If the elements or cells in a system are not roughly equal in terms of reliability, the system will be paced by its slowest element,' says Rueping.

Finally, for the smoothest running project, avoid temptation. 

• Don't increase the cost and complexity of a project by introducing more technology than is absolutely necessary. 
• Don't overpredict the ability of the automation to work correctly and eliminate all back-up options for manual assistance. 
• Don't overconstrain designs with too many positioning points.

With so many details requiring simultaneous attention, it can be difficult to define needs clearly. 'The first thing we try to do is identify what our customer, the soldier, wants to do,' says Tom Mathes, associate director for design and manufacturing at the Tank Automotive Research Development and Engineering Center in the U.S. Army's Tank Automotive Armaments Command. Next is a study of the environment the robot will work in, space, weight and power limitations, budget and time constraints, cost, commercial availability of components, degree of sophistication required for the controller and human/machine interface requirements. 'There's also a need to plan for [future] product improvement, maintainability and repairability,' says Mathes.

Eastman Kodak, where assembly of one-time use cameras is automated, uses phases and gates processes to ensure a new technology is robust and then to commercialize it.

Simulation can help
Simulation and modeling also are widely used tools. Such programs can evaluate performance of capital investments before the purchase order is signed and also to evaluate process and methodologies before implementation on the shop floor. This provides an opportunity to eliminate any problems with fixturing placements and interference in a virtual environment before the system is actually installed and turned on. Once engineers are satisfied with the simulation, the data can be used to program the robot.

Simulations also are valuable because they allow users to see systems in action and permit experimentation with multiple iterations. 'It's much cheaper to simulate rather than create a physical mock-up,' notes Al Hufstetler, vice president, Deneb Robotics, Inc., Troy, MI.

'We like to use simulation because it saves time and money in the long run,' says Mathes.

Relying on data imported from a CAD program, simulation software like that provided by Deneb can optimize a collision-free path for the robot, as well as jigging and fixturing and end-of-arm tooling.

On automotive production lines, for example, original equipment manufacturers and, increasingly, Tier 1 suppliers use Deneb's Assembly program to ensure the part will fit through the opening in the chassis and determine the type of tooling required to move and attach it. An offline simulation program is used to determine how to mount the tooling. As a result, robots have moved beyond painting and welding areas into trim, chassis and final assembly. 

One company, which has done considerable work in simulation is Adept Technology, Inc., San Jose, CA, and its SILMA division. Its new Production PILOT virtual factory suite includes four classes of interconnected simulation tools based on 3D CAD: assembly design and feasibility, yield prediction, cycle time prediction and line flow analysis and optimization. The interactive programs share a single operator interface and simulation environment and seamlessly transfer data between tools. 'The goal is a quantitative measure of the entire process,' says Joe Campbell, vice president of marketing at Adept.

Designers can begin in any of the suite's three modules: Pilot Yield, Pilot Cell and Pilot Line. However, beginning in Product Yield, the product-oriented module, 'allows manufacturing and design to work together at the earliest conceptual stages to create products, which are easily assembled,' says Eric Jacobs, simulation product manager at SILMA in a paper entitled 'Production Ramp Up: The Key to Automated Assembly.' As might be inferred from their names, Pilot Cell examines cycle times and process feasibility, and Pilot Line produces a system model demonstrating line flow.

Embedded in Pilot Yield is DAC assembly process knowledge software sublicensed from Sony Corp. of America, New York. 'Sony uses it internally to determine automation feasibility and approach,' reports Campbell. Built into the Pilot DAC (design for assembly/disassembly cost effectiveness) program are manufacturing rules and process knowledge for assessing assembly ease.

The software allows engineers to evaluate assembly processes at the earliest design stage when manufacturing investment is low and product and process design flexibility is high. This is accomplished by examining design criteria such as features of parts to evaluate ease of feeding and pickup, features of assembly and features of processing such as subassembly constraints for the next station. Once data on features of parts, assembly and processing are input, the program evaluates the product's suitability for automated assembly/disassembly, assigns a rating and validates assembly steps and sequences.

Thus, engineers can begin process planning before physical parts or automation hardware exist. It also gives the engineer the flexibility to analyze various automated and manual assembly scenarios for feasibility and financial benefits. As a result, time to volume production is shortened, an essential factor in 'establishing market share, staying ahead of competitors and generating return on capital/ROI,' says Campbell.

Future prospects
While some manufacturers have resisted robotic automation it's only a matter of time before it will mean the difference between success and failure.

'People will always be more flexible,' admits Huff. However, he says, 'We are getting to the point in a lot of our product development where people are no longer capable of doing the work.' Many parts are too small to handle manually and human operators cannot be steady enough or repeatable enough to provide the precision needed. 'Automated processes are 10 times as precise with less variability than manual processes,' notes Huff.

In addition, 'Design for flexible automation not only helps build a better product physically, but helps shorten time to market,' he concludes.