CitationDownload as .RIS
Emerald Group Publishing Limited
Copyright © 2003, MCB UP Limited
Really rapid prototyping
Really rapid prototyping
Jonathan Corneyis with Innovative Manufacturing Research Center, Heriot-Watt University, Edinburgh, UK. Web site: http://www.imrc.hw.ac.uk
Keywords: Rapid prototyping, Rapid manufacturing, Selective laser sintering
Recently we received a part produced by one of the latest selective laser sintering systems. The completely solid, steel component contained a small spiral hole whose walls were covered by intricate shapes, barely visible to anyone peering down it. Considering that, until recently, such parts could only have been manufactured by an elaborate casting process it is impossible not to be impressed by this latest manifestation of rapid prototyping (RP) technology. It is also hard not to wonder “where it will all end?” Will RP systems eventually do to traditional manufacturing (i.e. lathes and mills) what laser printers did to typewriters?
Not necessarily. Indeed rather than causing the extinction of the present automated manufacturing technologies, such as CNC machine tools and automated assembly systems, RP systems could create the pressure needed to force a merger of these previously independent functionalities into a single machine. Such combined assembly/machining centres would be able to form the parts that they need to build complex assemblies. Such systems can be envisaged because of three recent changes in machining hardware and software.
Increased flexibility: although RP systems have been generating the “wows” over the last decade machine tools have not stood still. Additional axes and power tooling have created hybrid, mill/turn, machining centres that can produce, in a single set-up, components that are breathtakingly complex. Unlike RP systems, they can do this in minutes rather than hours. But, and this is the showstopper, creating the CNC code (i.e. the program) to drive these hybrid machine tools is still far from automatic.
Smarter software: 20 years of academic research into automatic feature recognition and computer aided process planning may not have cracked these problems (fully automated part programming systems are still a dream) but it has produced numerous improvements over purely manual methods. These are being steadily incorporated into commercial CAM packages making dozens of small tasks (such as tool sizing and set-up planning) slightly easier with each new release.
Fresh thinking: RP success is built on the fact that, at least in principal, it takes no skill to program them (in stark contrast to CNC systems). They can do this because they divide complex shapes into laminar slices that are so simple that their manufacture can be automated. Part of the brilliance of the first RP machine was the realisation that just because a mechanical component is defined as a single lump it does not have to be manufactured as a contiguous whole.
These elements have allowed several research groups to develop new approaches to manufacturing that merge automated assembly and CNC machine tools into what could eventually become really rapid manufacturing systems. Notably all these new paradigms use forms of subdivision to overcome the process-planning log-jam.
Already commercial systems have appeared from French (http://www.millit.com/) and German (http://www.cirtes.fr/strato/) companies that, given a 3D CAD model, will machine out series of solid cross-sections that can be manual assembled to form the final shape. At first glance it is easy to dismiss these methods as a poor man's RP system. But, what if the parts did not have to be manually assembled? What if a robot picked each precisely machined segment off the machine tool and, with some glue, assembled it onto the stack? Then one would have the flexibility of RP with the speed and precision of machining.
At Heriot-Watt University we are trying to create just such a system: using some clever software we are attempting to take arbitrary objects and divide them into lumps (not laminar) whose shape's allow their automatic machining and assembly. In one way the research challenges are all, the well known, “hard” problems of gripping and machining variable shapes. But in another way it is quite different, because we can change the shape of any component that cannot be machined or gripped (e.g. we can subdivide the shape in a different way). In summer 2004 we hope to demonstrate the close integration of intelligent CAM software, a machine tool and a robot manipulator (http://www.shapesearch.net/RPBlox/index.htm).
So in long term RP may not be the end of machining or assembly, but simply another technology that contributes to their evolution. If intelligent software can be developed to drive such combined machining and assembly systems they could become the really rapid manufacturing systems of the next decade.