CAD models help designers, but truly sophisticated models incorporating machine vision still need to be developed.
by Andy Wilson
To ease the design of today’s computer-based products, engineers can choose from a number of CAD packages for IC simulation, FPGA programming, schematic design, test, and PCB layout. Using these tools, developers can experiment with various designs before committing them to a specific implementation.
For designers of automated assembly systems that incorporate mechanical, computer, electrical, vision, and robotic systems from a number of OEMs, the task of simulating a specific system is somewhat more difficult. Nowhere was this more evident than at this September’s Assembly Technology Expo in Chicago, where numerous manufacturers displayed their latest robots, conveyor mechanisms, vision systems, and automated manufacturing systems. On many booths, companies showed how vision systems had been integrated into industrial-automation, semiconductor, pick-and-place, and automotive systems.
In developing these applications, designers integrate disparate components, subsystems, and systems from a variety of manufacturers. Wouldn’t it be great if, like electronic engineers, designers of these systems could simulate them as CAD models before committing to specific OEM suppliers of robots, conveyors, and vision equipment? “We believe that in the future, all manufactured goods in the world will be digitally defined, engineered, simulated, produced, and managed throughout their life cycle. And we are convinced that the most appropriate vehicles for achieving this are 3-D systems,” says Bernard Charlès, president and chief executive officer of Dassault Systems (Suresnes, France; plm.3ds.com).
The company’s Delmia Resource Modeling and Simulation software helps create CAD-based models that incorporate tooling, fixtures, machinery, and automation systems. Dassault’s Robot Task Definition software allows users to search a library for more than 700 standard robots and verify that the chosen robot is the most appropriate for the task using reach-analysis, autoplace, and feasibility routines. The programmer optimizes robot layout, programs the robot by teaching specific trajectories, and defines actions at given points in the trajectories.
This all sounds rather good, and it is, as far as it goes. Unfortunately, many automated manufacturing systems use a number of components, such as feeders, conveyors, sophisticated pick-and-place heads, and machine-vision systems, that are not easy to simulate.
John Dulchinos, vice president of sales at Adept (Livermore, CA, USA; www.adept.com), has first-hand experience with this problem. His company’s Production Pilot software was built to evaluate early-stage product designs, assess manufacturing processes and factory flow, and lay out and program factory automation equipment. After marketing the product for seven years, the product was finally withdrawn from the market in 2002.
“Trying to simulate robots, feeders, vision systems, and the cycle timing for these systems is not easy,” Dulchinos admits. “In fact, you can spend more time designing such a system using a simulator than it would take you manually. Because of this, Production Pilot was more often used by our sales people to show potential customers mockups of robotics-based automation equipment. In many respects the software was ahead of its time, and our customers were not ready for it.”