Teaching Technology: CAD/CAM, Parametric and Interactivity

Martin Bechthold Harvard University, Graduate School of Design http://www.gsd.harvard.edu/people/faculty/bechthold/index.html [email protected]

The paper discusses a project-based approach to technology teaching, and examines the case of the mobile information unit (MIU) for Harvard University and its art museums. A student competition was held to explore design alternatives for this unit. The winning entry proposed an interactive, pixilated fiber-optics display as well as touch screens. Parametric digital modeling was used in the design and design development of the scheme. Research included the study of structural alternatives, fabrication methods and the modes of interaction between users and the MIU.

Keywords: Computer-aided design and manufacturing; fabrication; simulation; prototype; fiber optics.

Integrating new technologies – an has taught the subject over the past two years. The introduction teaching objective is to provide an in-depth under- standing of the technology itself, and how it is ap- Computer-aided design and manufacturing (cad/ plied in practice. The teaching methods include lec- cam) techniques have been used by fabricators since tures, workshops, design exercises, case studies and the 1980s, but only fairly recently have architects field trips. In the second part of the course a more become aware of these technologies, and, more im- comprehensive design and prototyping projects portantly, have adjusted their design paradigms and has successfully allowed students to synergize the methods accordingly. Teaching efforts in this area various aspects of cad/cam technology with archi- differ significantly from teaching traditional CAAD tectural design. A project of this type is the mobile or construction classes, since cad/cam depends on information unit. both worlds. With new and emerging technologies the teaching of technology for students Teaching cad/cam technology: has undergone a profound transformation. fundamentals At Harvard’s Graduate School of Design classes in cad/cam have been offered by a number of in- A major question in teaching cad/cam is whether to structors continuously since the mid 1990s. The au- accept the paradigms of current design practice, and thor has been involved in several earlier classes and teach techniques that are immediately useful in that

Session 17: Digital Fabrication and Construction - eCAADe 25 767 Figure 1 Design proposal for a mobile information unit. Fiber optics panels on the outside create a pixilated display, while inter- active screens on the interior allow access to more detailed information (design: W. Bao, R. Gould and J. Hernandez).

practice context, or teach higher- concepts that CNC processes (for example cutting flat shapes bases might not be immediately applicable, but ultimately on *.dwg or *.dxf files) the use of these software en- lead to a revision of current practice models under vironments is nevertheless far from ideal, and using incorporation of cad/cam technologies. Learning the same environments in teaching cad/cam only experiences and output of the two teaching philoso- emphasizes the limitations of current practice. Limit- phies can be radically different. These differences will ed accuracy, and the cumbersome process of making be illustrated for the three distinct steps present in modifications by essentially erasing and remodeling any cad/cam based project: computer-aided design are particularly felt in the context of CNC manufac- (CAD), computer-aided manufacturing (CAM), and turing where the digital descriptions of parts have to computer-numerically controlled (CNC) manufactur- be digitally ‘watertight’ and without errors. ing environments. The author’s cad/cam teaching emphasizes use of parametric design development environments Entity-based CAD and surface modeler such as Catia or SolidWorks – environments struc- versus parametric design development tured to support the development and detailing of environments a given design. These powerful environments are Contemporary practices use a variety of non- geared towards producing geometry that is suitable parametric CAD environments for drafting (e.g. Au- for CNC manufacturing efforts. Students first are of- toCAD, MicroStation), tools pri- ten irritated by the need to plan and logically struc- marily for surface modeling (e.g. Rhinoceros), as well ture the digital model in order to allow for the types as rendering and other presentation environments. of variations needed in the ongoing design pro- None of these were originally conceived to directly cess, but once the initial hurdles are overcome they support CNC manufacturing efforts. While many of quickly realize the powerful potential of parametric, those environments can be useful in the support of dimensionally-driven modeling and design. The

768 eCAADe 25 - Session 17: Digital Fabrication and Construction introduction to Catia and SolidWorks usually starts files from the design environment to the CAM envi- with simply model assignments, but soon introduces ronment and eventually fabrication can certainly put more advanced techniques that are distinctly differ- this myth into perspective. ent from conventional digital modeling. Techniques Tolerances are a particular issue that cannot be such as patterns, dimensional constraints, equation- overestimated. Students that are used to working driven dimensions, model configurations, design and designing in CAD often tend to assume that the tables, advanced replication tools (e.g. power copies physical world is as accurate and precise as is sug- in Catia) and macros can be introduced fairly early. gestive of the CAD world. Even simple modeling These and other digital modeling strategies facilitate exercises and 2D cutting operations of CNC lasers or a mental re-tooling process that broadens the stu- water jets quickly bring up these issues when parts dent’s design techniques. are supposed to snap into a tight fit but the width of the laser or water jet has not been figured into the Computer-Aided Manufacturing (CAM) and design model. Robust dimensionally-driven mod- Computer-Numerically Controlled (CNC) eling tools allow for tolerances to be incorporated manufacturing into the design model, as well as permitting interfer- Is there value of having students learn to generate ence checks and allowing students to test assembly their own G-code to eventually run CNC machining processes virtually before cutting and assembling operations, or generate files that can be sent directly materials. to a CNC laser cutter? In practice few if any fabricators will accept machine instructions from anyone out- CAD/CAM and design side the company, and few will eventually be doing these kinds of operations. Many teaching The temptation is great to have students soon gen- institutions support the ‘service model’ where parts erate and produce models and prototypes are not made by students, but by shop operators. using the computer-controlled devices available. In The myth of ‘press the button’ fabrication, often the author’s teaching experience it has been fruitful evoked by terms such as ‘digital fabrication’, still per- to build up understanding and skills incrementally. vades the design community, and teaching institu- Early fabrication exercises are typically extremely tions are no exception here. Having students actually limited, exploring specific devices usually begin- experience first hand the complexities of migrating ning with 2D cutting operations, for example of a

Figure 2 Student project of a plywood chair designed parametrically and produced on a CNC rout- er. Scale model tests triggered the of eliminat- ing the cantilever still present in the first iteration (Design: S. Panata, C. Talbott).

Session 17: Digital Fabrication and Construction - eCAADe 25 769 CNC router, water jet or laser. Typical design assign- unit for Harvard (see next section) demonstrates a ments may be to produce a piece of furniture from more comprehensive design project that begins to a single sheet of plywood, ideally preceded by scale address the real complexities of cad/cam in practice. model studies based on the same parametric digital model. Class design project: interactive mobile The constrained assignments encourage exper- information unit imentation because the restrictions and possibili- ties of the available technology are readily under- The interactive mobile information unit (MIU) was stood. To contextualize these design experiments conceived as a mobile, trailer-based urban kiosk case studies as for example included in [Schodek set up to inform the Harvard community as well 2004] are important. Field trips to fabricators can as the surrounding communities about the many give students real insights in the problems and op- public events at Harvard. A particular focus of the portunities of the technology. Questions such as MIU was to disseminate information about exhibi- how owns the digital model, liability and schedul- tions, events and lectures at the Harvard University ing issues become alive once raised by practitio- Art Museums (http://www.artmuseums.harvard. ners that are actively involved in the application of edu), one of the largest private art collections in the technology. the United States with more than 260,000 artifacts. Intermediate project assignments usually re- From the perspective of the Graduate School of De- main small in scope, but introduce more complex, sign the project was conceived as a multi-disciplin- often indirect processes (e.g. casting, molding). ary research that involved expertise in interaction Comprehensive design projects have been used by design, information technology and fabrication the author as a way to allow student groups to ex- technologies. plore the full complexity of cad/cam technologies. A detailed brief for the MIU provided the basis of The design problem is usually simple as far as site a course-based student competition that took place and program are concerned, but should encourage over a 5 week period in the fall semester of 2006, as students to creatively re-think design approaches part of a course taught by the author, entitled “CAD/ by starting with the opportunities that parametric CAM 1: Introduction to Applications in Architecture”. digital modeling and CNC manufacturing offer. The 29 students were asked to form groups of not more case study on the project of a mobile information than 3 individuals, and prepare a design proposal for

Figure 3 Student entry with projection screens and SMS-based inter- activity (design: S. Panata, C. Parlato, J. Rule)

770 eCAADe 25 - Session 17: Digital Fabrication and Construction the unit using a parametric, dimensionally-driven MIU: development process CAD environment such as CATIA or SolidWorks. The brief asked for a solution that included innovative Students of the winning team (Wei Bao, Russell aspects of fabrication and possibility of interactively Gould, Jaime Hernandez) proposed a series of fiber- displaying information. Robustness and ease of op- optics panels for a pixilated display on the outside of eration were also to be considered, and a budget the unit. On the interior several touch screens and of $25,000 was given for the fabrication of the unit. flat screen displays allowed for more detailed infor- Both night and daytime use was to be considered. mation to be displayed in interactive ways. Each pan- Submission requirements included a parametric dig- el box measured approximately 1.5 x 2.4 m and 0.45 ital model and a full size of a typical portion m in depth. A total of four interactive panels were to of the scheme. be mounted on a trailer, with an interstitial ramp and As a first step student groups were charged with elevated platform system that allowed access from three different tasks that included site documenta- the ground to the unit. tion using photogrammetry, research on projection and interactive technologies, and the parametric Design development – a group approach modeling of utility flat bed trailers on several para- A group of 8 students further developed the winning metric cad platforms. The results of this phase were scheme as part of a research seminar taught by the au- submitted to the course web platform for use by thor in the spring of 2007. The group had one or two all groups in their design efforts. Student groups weekly group meetings, with other meetings and work worked largely independent on the design. There sessions in between. Two subgroups were set up, one were two scheduled class pin-ups where feedback concentrating on the physical aspects of the design was given by the author as well as by other mem- (five students), and the other focusing on developing bers of the GSD faculty. At the final presentation an interactivity concept and use concept for the MIU a large variety of schemes were presented by stu- (three students). To emulate a professional practice sit- dent groups. A jury consisting of faculty members, uation students within each of the two groups chose to guests, the museum director and deputy director take on responsibility for certain aspects of the scheme then selected the winning entry that was to be fur- (e.g. trailer and wheel assembly, ventilation system, fi- ther developed. ber optics, structural frame, electronic devices etc.).

Figure 4 The revised scheme allowed for easy access of the interior, but posed challenges for wheel and suspension sys- tems. Structural issues were addressed early with compu- tational simulations (center). Different configurations of operable panels adjusted the unit to a variety of locations (right).

Session 17: Digital Fabrication and Construction - eCAADe 25 771 The group meetings were the primary mode of mounted assembly that includes a raised floor. The coordinating work between individuals, and students original scheme had not featured a roof, but for the took turn in documenting the meetings and writing sake of structural stability a roof was adopted early meeting minutes that were posted on a shared web- in the semester. The roof also had the advantage of based course platform. Meeting minutes laid out shading the displays in bright sunlight, as well as action items that needed to be resolved. A schedule creating a more intense, enclosed environment that was developed that laid out a time line for major de- enhances the experience of the users. cisions and development steps. The goal of the se- mester was to complete the design development of MIU development – fiber optics the unit and produce a full-size working prototype of A main innovative aspect of the proposal was the use one of the interactive panels. of large fiber optic panels as an interactive, pixilated display. Fibers were threaded from an extremely MIU development – trailer and frame dense end plate (onto which the image was project- The group felt that stepping onto the ramp and plat- ed) into a less densely spaced array spread out over form could be an obstacle that would prevent many the size of the outer acrylic panel. people from casually interacting with the MIU. A de- Much of the research seminar was dedicated to sign change was studied that proposed an inverted optimizing the spacing and type of fiber, as well as U-shape with the wheels contained in each of the testing brightness and legibility under a variety of vertical panels, but no horizontal trailer bed at the lighting conditions that ranged from bright daylight lower level. to a night scenario. The ability to read the display from Structural simulations of the revised frame indi- the side (angular reading) was also investigated. To cated that the revised scheme was indeed feasible. answer these questions multiple smaller prototypes Lateral loads were considered at their maximum were produced on a CNC router. Optical glass fibers when the unit was close to tipping over as might oc- with diameters between 1 and 3 mm were evaluated cur under driving conditions in a tight curve. The as- during this phase. The experiments showed that the sumed maximum permissible weight of the unit was different fiber diameters did not significantly affect 1350 kg (U.S. regulations for road trailers). A center of the brightness of the display. For cost reasons 1 mm gravity was calculated based on the digital models. fibers were chosen for the full-size mockup. During Maximum deflections for a frame made from welded bright daylight only strong contrast (white/yellow – hollow rectangular tubes (50 mm square) were in the black) could be perceived on the display, while lower order of several millimeters. Forces from driving over ambient light levels allowed for more detail to be a major pothole resulted in a maximum deformation legible. Text of various sizes was tested for legibility. of 15 mm – requiring a flexible joining of the acrylic Having graphics move while being displayed much envelop to the structural frame. increased their legibility for any ambient light con- A major question remained the availability of dition. Movements were ideally slow and steady to an axle-less wheel/suspension system. Design and improve legibility. fabrication of custom wheels was excluded due to time and budget reasons (quoted cost was $5,800). A walk-in computer – interactivity The only ready-made wheel and suspension system Understanding the way individuals would interact available for the revised design was a torsion axle set with the unit was a major aspect of the research proj- that far exceeded the available width of the side pan- ect. The MIU was conceived as a walk-in computer. els. Difficulties of integrating those wheels into the Clearly it had to allow different types of ultimately lead the scheme back to a trailer- to be displayed, and permit a variety of modes of

772 eCAADe 25 - Session 17: Digital Fabrication and Construction Figure 5 Full-size prototype of the fiber optics panel (left). Zones of interactivity range from medi- um range to close range, with touch screens on the interior functioning as additional input devices (Interactivity Study by D. Tsigaridi and A. Gonzalez).

interactivity. The needs of the museum were taken detailed information on the museums using the as a research microcosm to understand large issues touch-screens (mode 3). of interactivity in urban-scale furniture and kiosks Precedent for interaction with screen-displayed such as the MIU. information in modes 2 and 3 are plentiful, so the Early on three different levels of interaction research focused on developing interfaces that were identified, ranging in scale and distance would allow for relevant information about the art from the most distant interaction when viewing museums to be displayed. The touch-screens were the fiber-optics display (mode 1), to a medium dis- designated to allow interested individuals to browse tance mode when viewing large flat screen moni- a Macromedia Flash site with content pulled over tors mounted on the hinged-out doors (mode 2), urban wireless communication directly from the to the most intimate interaction with the touch- museum’s database. The two large flat panels would screen panels on the interior underneath the roof normally display movies or animated presentations element (mode 3). Each of the three modes al- about the museums or, as the case might be, other lowed for a distinct amount of detail and level of events at the University. interaction. The fiber optics display was seen as a way to Interactive game catch people’s attention and allow for a casual, play- Significant development time was dedicated to un- ful interaction with the unit through. Together with derstanding and prototyping modes of interactiv- the large flat screens (mode 2) they provided avi- ity with the fiber optics display. The basic concept sual incentive to enter the unit and browse for more was to record a viewer’s position with sensors, and

Session 17: Digital Fabrication and Construction - eCAADe 25 773 Figure 6 Ultrasound sensors detect distance of up to two indi- viduals. Right: The signals are processed through an i/o board, and the output is converted into an interac- tive game using ‘processing’ programming (Design and Prototype: R. Urbano- Gutierrez).

translate changes in position and proximity into an was a useful test bed to understand the relation animated graphic display. To test this concept an between the physical and the virtual aspects of interactive game was developed, whereby the po- the MIU. sition of up to two viewers standing in front of the fiber optics panels would controlled the position Conclusions of two circular shapes that were being followed by smaller graphic elements. Position changes of the In teaching cad/cam technologies it is important for players created a virtual chase of the graphic ele- students to gain a deep understanding of the un- ments on the display. derlying principles, including advanced parametric The final technical set up used four ultrasound modeling techniques as well as hands-on knowl- sensors whose signals were processed through an edge of manufacturing processes, G-code program- Arduino i/o board. The signal output – essentially ming and CAM applications. Small-scale design constantly changing numbers that represented the explorations are useful as long as the broader prac- distance of a viewer from a set of two sensors – were tice context is provided by the instructor. For more converted into the graphics display using the ‘pro- advanced students more cromplex projects are ex- cessing’ programming environment. The ultrasound tremely useful, but it is highly recommended to limit sensors tracked distances up to 6.4 m. scale and scope such that an in-depth investigation Tests with individuals unfamiliar with the is possible. workings of the game showed that its rules and As a case study for a more advanced design proj- possibilities were not immediately obvious for us- ect the mobile information unit (MIU) was presented. ers, and the final version as well as other applica- In a broad approach to current as well as emerging tions will be designed for easier and more intui- fabrication and interaction technology it provided tive access. As a prototypical application the game a test bed for parametric modeling, simulation, and

774 eCAADe 25 - Session 17: Digital Fabrication and Construction furthered the understanding of urban interaction of small urban interventions. As a means of teach- ing technology to graduate students the project has proven highly valuable. At the time of writing of the paper the project was in continuous development, with completion scheduled for the end of 2007.

Acknowledgements

The author would like to acknowledge the help of Allan Sayegh and Prof. Kostas Terzidis in the design and prototyping of the interactive technology. Much appreciated was Prof. Dan Schodek’s input on many levels. The following students participated in the design development of the unit: Wei Bao, Behrang Behin, Ana Gonzalez, Russell Gould, Jaime Hernan- dez, Christopher Parlato, Dido Tsigaridi, Rosa Urbano Gutierrez. Fabrication of the MIU is supported by a grant from the Harvard University Provost.

Reference

Schodek, D., Bechthold, M., et al: 2004, Digital Design and Manufacturing: CAD/CAM Applications in Ar- chitecture and Design, Wiley, Hoboken.

Session 17: Digital Fabrication and Construction - eCAADe 25 775 776 eCAADe 25 - Session 17: Digital Fabrication and Construction