Growing Autofab
into the 21^st Century

     Automated fabrication started more than 45 years ago with the invention of numerically controlled (NC) machining. In the 1980s, several “additive” techniques were developed. These processes work with photocurable plastic resins, thermoplastic powders, adhesive droplets, and other specially devised materials. Together, subtractive and additive autofab now offer modern manufacturers the ability to prototype and produce new designs faster and cheaper than ever before.

     Future fabricators will go far beyond today’s capabilities to offer higher-resolution and faster object generation. They will expand on the currently available fabrication materials to include metals, advanced composites, and possibly even living tissues. Machine prices will go both up and down, promoting use by both heavy industry and casual tinkerers. Users may someday communicate designs and design changes via three-dimensional “virtual reality” environments, making the machines both easier and fun to use.

Figure 8.1. The triad of important issues in automated fabrication research and development.

     How is all of this going to come about? The necessary work is underway in hundreds of industry, university, and government laboratories around the world. Research and development in automated fabrication may be viewed in terms of the three interconnected fields shown in Figure 1. The first area, process research, looks at methods for manipulating and inducing structure in solid materials. This typically involves issues in mechanical and chemical engineering, such as the power and scanning speed of a photocuring laser, or the temperature and pressure in an extrusion nozzle.

     Materials research studies existing solids and proposed new ones, both from the perspective of their bulk properties (utility after fabrication) and their behavior under manipulative processes (utility as raw material in fabrication). This calls upon materials science and chemical engineering for the identification or synthesis of raw materials best suited to a particular process, such as new photopolymer resins, or plastic powders with a specific range of grain sizes. It may also come up with novel materials whose surprising behaviors suggest new fabrication processes.

     Control research devises mechanisms and procedures to enact the various fabrication processes automatically. This brings together computer science and mechanical engineering to develop mechanisms, algorithms, and protocols.

     None of these fields can be studied in isolation of the others. For example, many current autofab processes involve a change in material state from liquid to solid. Optimal control of such processes requires an intimate understanding of the material properties and behavior in the liquid, solid, and transition states. Conversely, the final solid properties will benefit from various aspects of control that can be brought to bear on the process.

Process research: Learning to manipulate matter

Figure 1. The three fundamental processes are the bases of subtractive, additive, and formative fabrication, respectively.

     There are three fundamental fabrication processes, as illustrated in Figure 2. In a subtractive process a single piece of material is cleaved in two, whereas additive fabrication causes two separate pieces of material to fuse. A formative process contorts a single piece of material into changing its shape. Every fabrication process, manual or automated, comes down to one or more of these basic processes. For example, die cutting is subtractive, masonry is additive, and forging is formative. Spinning pottery on a potter’s wheel is a combination of additive and formative fabrication, with an occasional use of subtractive. The most important processes for modern manufacturing, molding processes such as injection molding and investment casting, are formative.

     Autofab arises when one or more fabrication process is controlled by a computer. Most of today’s fabricators are either exclusively subtractive or additive. For example, a CNC mill or lathe uses only subtractive processes. The SLA StereoLithography devices from 3D Systems are additive fabricators, as is DTM’s Sinterstation. The Helisys LOM, which makes its shapes by bonding and cutting layers of paper, is a hybrid subtractive/additive machine.

     Automation of formative processes, which amounts to molding without molds, has attracted only sporadic attention to date. Although many aspects of modern injection molding are highly automated, the process still requires a special tool (mold) to be created for each shape to be molded. This defeats the basic philosophy of automated fabrication, which requires the ability to make arbitrary shapes without special tooling. There has been some rudimentary investigation of changeable-configuration molds, but the first commercial implementation of formative autofab is in CNC press brakes. A press brake is a machine for placing folds and bends in sheet metal. Coupling this with a CNC punch press yields a hybrid subtractive/formative fabricator which turns flat metal sheets into finished desk drawers, refrigerator panels, or air conditioning ducts. Such machines are currently made by Salvagnini (Sarego, Italy) and Iowa Precision Industries (Cedar Rapids, Iowa, U.S.A.).

     Process research in autofab aims to understand and devise techniques for enacting the three fundamental fabrication processes. The most popular style of cutting breaks off chips with a sharp tool, but other subtractive processes include shearing, abrasion, thermal cutting, and chemical dissolution. Joining may use adhesion or cohesion of solid particles, or it may work by bringing together liquid particles and causing them to solidify together. Formative fabrication requires the application of compressive forces on opposing sides of the material at the same time.

     Examples of process research include work to develop dual-beam laser curing techniques (See Rapid Prototyping Report, July 1993, page 2.) and options for non-photonic selective curing of plastics. Process research can be motivated by specific goals to be achieved in a fabricator; an example would be working on a method that operates in an open space in order to defeat limitations of current fabricators to their fairly small build envelopes. One of the most fascinating fields of process research is in nanofabrication: taking the practical scale of fabrication down to the molecular level. Although outside the realm of direct fabrication processes, one could also include in this category robotic techniques intended to take the output objects from a fabricator and assemble them into working mechanisms.

Materials research: Getting intimate with solids

Figure 8.8. Three types of solid crystals. In a metal (left) outer electrons float freely about the atomic cores (nuclei and inner electrons represented by black dots). A covalent crystal (center) shares individual electrons between pairs of atoms. In an ionic material (right) outer electrons are shared disproportionately by the different species of atoms present. [Reprinted with permission from Automated Fabrication by Marshall Burns.]

     The properties of a solid material depend on its composition and structure at various levels. The most basic level involves the types of atoms and the ways they are arranged (see Figure 3), while at higher levels, a polycrystalline material, for example, is affected by the sizes, shapes, mutual orientations, and packing density of the crystal grains. Modern chemistry and materials science have done a great deal to identify and explain many types of materials, but surprises continue to arise. Atomic clusters, self-organizing liquid crystals, and spherical fullerene molecules are relatively new artificial structures that have forced researchers to rethink our understanding of matter. Of the literally infinite variety of possible composite materials, only a few of the most obviously promising ones, such as fiberglass and reinforced concrete, have been created. Decades, if not centuries, of challenging work remain to be done in exploring the potential for designing new materials for use in autofab.

     A very intriguing field at the intersection of materials and process research is that of “smart materials.” This refers to material systems that respond reversibly to various mechanical, thermal, chemical, or other kinds of influences in their environment. The behavior may involve changes in color, chemical reactivity, physical structure, or other properties. The best example of a smart material to-date is an engineering marvel given to us by nature, called muscle. Our developing ability to create new materials that expand and contract on demand portend a possible future with dynamic, or “real-time” fabrication processes. Such magical processes would create structures whose internal and external shapes are not permanently determined at their time of fabrication, but are subject to a range of movement and reconfiguration.

Control research: Incorporating intelligence in machines

Figure 8.9. Control aspects of automated fabrication. The key elements, shown here with underlined labels, are the representation of abstract geometry in computer code and the creation of machine instructions to direct the fabrication process.

     Control is the essence of automated fabrication. It links the mind of the user with the physical processes that create the desired object.

     The two most important elements of control are the representation of the desired geometry in computer code and then translation of this code into instructions to guide the fabrication process. (See Figure 4.) The geometry may arise from a human design, from scanning the shape of an existing object, or from another mathematical source. In the course of fabrication, a good system will monitor its ongoing results and feed them back to the control computer. After the object is made, the user may evaluate the results and decide whether to make certain changes to the design or to the process parameters.

     One very interesting challenge in the arena of control research involves the coordination of multiple process sites. The ability to process material in several places at once offers opportunities for dramatic improvements in speed and efficiency. It is possible that the best way to control such simultaneous processing will use computation devices very different from today’s serial and parallel computers. Neural networks, fuzzy logic, and other new computational techniques may find important applications in controlling advanced fabricators.

     At the interface between the user and the fabricator, we can expect to see dramatic improvements in 3-D CAD systems. Computer screens, keyboards, and mouses will give way to 3-D displays with interactive fingertip point indication and voice instruction. All of these technologies are currently under development. Developments in “virtual reality” technology are making important contributions here. Another element, automatic mechanical property prediction, will allow the user to read the strength of a material in a design in the same way as drawing programs today indicate color on a monochrome screen with various patterns of shading and hatching.

     For the more serious fabricator user, there will be matter programming languages. To understand this concept, notice that a fabricator is a machine that does with matter what a computer does with information: it takes it in in one form, performs some operations on it, and sends it back out in a different form. A matter language is a system of coded communication that gives a person intimate control over those operations without the person needing to know the details of how the operations are performed. This provides the same advantages that a computer programmer gets from using C or Pascal to achieve efficient, robust processing without needing to be concerned about computer registers and accumulators. Instead of drawing an object graphically in CAD, the matter programmer “writes” the object in process code. While casual users will enjoy the convenience of a 3-dimensional interactive design environment, professional fabricator engineers and fabricator “hackers” will appreciate the increased control and versatility available from writing matter programs.

     Work underway today in laboratories around the world is making great strides in understanding and improving the processes, materials, and control strategies at work in automated fabricators. While the improvements seen in commercial systems in just the last few years are certainly dramatic, these developments have only begun to explore the potential for generating 3-dimensional, solid objects under computer control.

     If you found this interesting, you’ll also want to read:

• The Origins and Direction of the Fabricator Revolution, the farthest-reaching discussion of future fabbers, including “accretive fabrication,” modeled on the growth of biological systems.
• Using Fabricators to Reduce Space Transportation Costs, explains how fabbers are the critical element that will finally make space habitation feasible for the first time in history.