Fabrication of Non-Assembly Mechanisms and Robotic SystemsUsing Rapid Prototyping

Constantinos Mavroidis1

Kathryn J. DeLaurentis2

Jey Won3

Munshi Alam4

Robotics and Mechatronics LaboratoryDepartment of Mechanical and Aerospace Engineering

Rutgers University, The State University of New Jersey98 Brett Rd, Piscataway, NJ, 08854-8058

Tel: 732 - 445 - 0732, Fax: 732 - 445 - 3124E-mail: mavro@jove.rutgers.edu, kjdela@jove.rutgers.edu

Webpage: http://cronos.rutgers.edu/~mavro/robot

1 Assistant Professor, Principal Investigator, Corresponding Author2 Graduate Student, NSF Fellow3 Undergraduate Research Assistant4 Graduate Student

Abstract: In this paper, the application ofRapid Prototyping in fabricating non-assemblyrobotic systems and mechanisms is presented.Using the Stereolithography Apparatus SLA190 of the Department of Mechanical andAerospace Engineering of Rutgers University,and the Selective Laser Sintering Sinterstation2000 of DTM Corporation of Austin, TX,prototypes of mechanical joints were fabri-cated experimentally. The designs of thesecomponent joints were then used to fabricatethe articulated structure of experimental pro-totypes for two robotic systems: 1) a three-legged parallel manipulator; 2) a four degree-of-freedom finger of a five-fingered robotichand. These complex multi-articulated, multi-link, multi-loop systems have been fabricatedin one step, without requiring assembly whilemaintaining their desired mobility.

Introduction: It is always desirable to evaluate aproposed robot design prior to full prototyping toensure the swiftest and most cost effective designchanges. Even though powerful three-

dimensional Computer Aided Design and Dy-namic Analysis software packages such asPro/ENGINEER, IDEAS, ADAMS and Work-ing Model 3-D are now being used, they cannotprovide important visual, haptic and realisticworkspace information for the proposed design.In addition, there is a great need for developingmethodologies and techniques that will allow fastdesign, fabrication and testing of robotic andother multi-articulated mechanical systems. Aframework for the feasibility and usefulness ofapplying Rapid Prototyping in fabricatingmechanisms and robotic systems is presentedhere.Rapid prototyping of parts and tools is a rapidlydeveloping technology [1, 2] that provides manyadvantages: a) time and money savings, b) quickproduct testing, c) expeditious design improve-ments, d) fast error elimination from design, e)increased product sales, and f) rapid manufac-turing. Its main advantage is early verification ofproduct designs. Additionally, Rapid Prototypingis quickly becoming a valuable key for efficientand concurrent engineering. Through different

techniques, engineers and designers are now ableto bring a new product from concept modeling topart testing in a matter of weeks or months. Insome instances, actual part production may evenbe possible in very short time. Rapid Prototypinghas indeed simplified the task of describing aconcept to design teams, illustrating details toengineering groups, specifying parts to purchas-ing departments, and selling the product to cus-tomers.The advantages of using Rapid Prototyping tech-niques in mechanism and robot design and fabri-cation are numerous. With Rapid Prototypingtechniques, physical prototypes of the mecha-nism or robot that is being designed can be ob-tained in a very short time, thus making quickdesign evaluation possible. By using these physi-cal prototypes, several properties of the mecha-nisms can be evaluated immediately such as: a)workspace evaluation, b) identification of singu-lar configurations including uncertain configura-tions where the mechanism has internal mobility,c) determination of link interference, and d) visu-alization of joint limits. Evaluating these funda-mental properties of the Rapid Prototypedmechanism can considerably reduce the time andimprove the quality of the design process. Fur-thermore, Rapid Prototyping allows the fabrica-tion of complex three-dimensional structures,which could not be produced with conventionalfabrication processes. Such possibilities make theincorporation and attachment of sensors, actua-tors and transmission elements within the struc-ture and joints of the mechanism easier. Finally,Rapid Prototyping allows one step fabrication ofmulti-articulated, multi-link systems as a whole,without requiring assembly of its structuralmembers and joints after fabrication. This onestep fabrication technique can drastically changethe way that mechanisms and robots are currentlybuilt. This is very important as it can allow rapidfabrication of fully functional and mobile sys-tems. In this paper, these mechanisms and roboticsystems whose multi-articulated structure is builtin one step are called non-assembly mechanisms.

Robotic systems have been used as part of aRapid Prototyping process [3, 4]. However, theapplication of Rapid Prototyping in robot andmechanism design and fabrication has been verylimited. Professor Gosselin and his group at La-val University, using a Fused Deposition Mod-eling Rapid Prototyping machine, fabricated sev-eral mechanisms such as a six-legged six degree-of-freedom parallel manipulator [5, 6]. Theserapidly manufactured mechanisms required as-sembly after Rapid Prototyping of the mecha-nism parts. Professor Cutkosky and his group atStanford University using a different RapidPrototyping process called Shape DepositionManufacturing developed planar, non-assemblymechanisms and robotic systems with embeddedsensors and actuators [7, 8]. These componentswere inserted in the multi-articulated structureduring its fabrication as opposed to their integra-tion in post-fabrication assembly phases. Thisgroup also proposed methods for performing thesystematic design, error analysis and optimalpose selection for these mechanisms [9-12]. Ad-ditionally, researchers at the Georgia Institute ofTechnology, proposed methods to develop com-plex devices with embedded components usingthe Stereolithography technique [13, 14].During the last two years our group at RutgersUniversity has studied the fabrication of non-assembly mechanisms using two different RapidPrototyping processes: Stereolithography (SL)and Selective Laser Sintering (SLS). Preliminaryresults of our work were presented in [15-17]. Inthis paper our design considerations, results, andprototypes in building complex, spatial, non-assembly mechanisms are presented in detail. Tothe authors’ knowledge, this is the first successfulfabrication of multi-joint, multi degree-of-freedom, spatial robotic systems and mechanismswithout requiring any assembly using the SL andSLS processes.

Using the Stereolithography machine model SLA190, from 3D Systems, CA. a set of joints thatinclude revolute, prismatic, universal and spheri-cal joints, were fabricated. In addition, a three-legged, six degree-of-freedom Rapid Prototype

of a parallel manipulator was built in one step,without requiring assembly. Each leg of thisthree-legged parallel manipulator is composed ofone prismatic joint and two spherical joints,which connect to the two triangular platforms.Further prototype joints, similar to those fabri-cated with the Stereolithography process, wereRapidly Prototyped using the Selective LaserSintering Sinterstation 2000 of DTM Corporationof Austin, TX. Finally, a four-degree-of-freedom finger of a robotic hand with four fin-gers, a thumb and a palm are constructed as non-assembly type mechanisms. Actuation of thejoints of the rapidly prototyped systems is outsidethe scope of this paper and will be studied in ourfuture work.

Rapid Prototyping: Rapid Prototyping or Lay-ered Manufacturing is a fabrication techniquewhere three-dimensional solid models are con-structed layer upon layer by the fusion of ma-terial under computer control. This processgenerally consists of a substance, such as flu-ids, waxes, powders or laminates, whichserves as the basis for model construction aswell as sophisticated computer-automatedequipment to control the processing techniquessuch as deposition, sintering, lasing, etc [18,19]. Also referred to as Solid Freeform Fabri-cation, Rapid Prototyping complements exist-ing conventional manufacturing methods ofmaterial removing and forming. It is widelyused for the rapid fabrication of physical pro-totypes of functional parts, patterns for molds,medical prototypes such as implants, bonesand consumer products [20]. Its main advan-tage is early verification of product designs.Through quick design and error elimination,Rapidly Prototyped parts show great cost sav-ings over traditionally prototyped parts in thetotal product life cycle [21].

Currently, there are over 30 different types ofRapid Prototyping processes in existence.Some of these techniques are commerciallyavailable while others are still in developmentin research laboratories [21]. Over the years,

major improvements in the overall quality ofprototyped parts have been achieved throughenhancements in accuracy, material choice anddurability, part throughput, surface texture,and alternative RP processes. These improve-ments have led to an evolution of the function-ality of RP prototypes [19, 20]. Evolution ofthe techniques and applications of RapidPrototyping is a continually developing andexpanding field. Current research is leading toa more functional rapidly prototyped part withan increasing number of applications and partfeature enhancements. In this work, two RapidPrototypes techniques are used: the Stereoli-thography (SL) and the Selective Laser Sin-tering (SLS).Stereolithography (SL) is a three-dimensionalbuilding process, which produces a solid plasticmodel. In this process, an ultraviolet (UV) lasertraces two-dimensional cross-sections on the sur-face of a photosensitive liquid plastic (resin). Thelaser partially cures the resin through low energyabsorption of laser light thus producing a solid.The first cross-sectional slice is built on a depth-controlled platform, which is fully submergedunder the first thin layer of resin. This and eachsuccessive thin layer of liquid resin has a depthequal to that of the vertical slice thickness of thepart. After each slice is traced on the surface ofresin, the platform lowers by a depth equal to thatof the slice thickness. Successive 2-D slices arecured directly onto previous layer as the part isbuilt from bottom to top.

Support structures are needed to maintain thestructural integrity of the part and supportsoverhangs, as well as provide a starting pointfor the overhangs and for successive layers onwhich to be built. These supports are con-structed from a fine lattice structure of curedresin. After the part is fully built, the supportstructures are removed and the part is cleanedin a bath of solvent, and air-dried. The pre-pared parts are then flooded with high-intensity UV light in a Post-Cure Apparatus(PCA) to fully cure the resin. The Departmentof Mechanical and Aerospace Engineering of

Rutgers University is equipped with a Stereo-lithography 190 machine from 3D Systems,CA. Available basis materials for this machineinclude photo-polymer resin epoxies withvarious physical properties.Selective Laser Sintering (SLS) is a three-dimensional building process based on thesintering of a metallic or non-metallic powderby a laser. The SLS process involves theheating of the powder using a CO2 localizedlaser beam. This localized heating raises thetemperature of the powder such that solidifi-cation by fusion occurs without melting. Themodel is built on a platform that is situatedwithin a horizontal platen. The platform,which is initially flush with the platen, is low-ered a depth equal to that of the slice thick-ness. A powder is then rolled, scraped or slot-fed onto the platform and then the laser drawsthe two-dimensional cross-section. This low-ering, powdering and lasing process is re-peated until the part is complete.

No support structures are necessary in thisprocess since the part rests on and within thenon-sintered powder. Post-curing is not neces-sary except in the case of ceramic parts. Avail-able materials include polycarbonates, nylons,polyamides, elastomers, sand casting materialsand steels. The fabrication of prototype partsused in the present investigations was througha professional Rapid Prototyping service pro-vider for SLS manufacturing. The machineused is the Selective Laser Sintering Sintersta-tion 2000 of DTM Corporation of Austin, TX.

Joint Fabrication:SL Fabricated JointsThe first step in building robotic systems witha Rapid Prototyping machine is to be able tosuccessfully fabricate joints. Different typesof mechanical joints such as revolute, spheri-cal, prismatic and universal joints were fabri-cated with the SLA machine from the Depart-ment of Mechanical and Aerospace Engineer-ing (Figure 1). These joints were producedwithout requiring assembly.

Through a trial and error process, differentfeatures such as clearance, part size and sup-port structure generation were optimized toproduce working mechanical joints. Of thesefeatures, determination of clearances was veryimportant in successful part fabrication. Theoptimum clearances between two near sur-faces, were determined to be 0.3 mm for flatsurfaces and 0.5 mm for circular surfaces.Also, the size of these parts is in the order of afew centimeters. This was done to determinethe limits of the available apparatus as well asto conserve processing time and material.

FIGURE 1. Stereolithography Joint Fabri-cations

In Figure 2 the CAD drawings and pictures of therevolute joints are shown. The revolute joint con-sists of two rings, each with a small stem at-tached. The two rings are connected, through thecenters of the ring holes, by a pin that acts as theaxis of revolution.Spherical joints with varying sizes for the balland socket were built. Two slightly different de-signs of spherical joints were fabricated. In thefirst design, shown in Figure 3, the bottom of thesocket is cut slightly to create an opening for thesupport structure to go through and hold the ballin place, so that the ball would not fuse to thesocket during fabrication. This spherical jointwas oriented as shown in the line drawing suchthat the ball and socket part was built first beforebuilding the link arm. The second type of designfor a spherical joint is shown in Figure 4. In thisdesign, since the ball and socket was built last(the build orientation is shown in the line draw-ing of Figure 4), the ball did not need any support

as it was supported by the vertical link itself. Theapproximate time of fabrication for this joint was5.75 hours.

FIGURE 2. Revolute Joint Built With SL (alldimensions are in mm)

Initially, the prismatic joint or sliding joint wascharacterized by that of a piston-cylinder typeassembly. Poor sliding of the joint resultedfrom a volume of trapped resin in the chamberthat could not be effectively or easily removedprior to final ultra-violet curing. Also evidentin the joint assembly was the presence of un-desired additional degrees of freedom. Thejoint was then redesigned with these two fac-tors as limiting constraints. The final designof the joint is a one degree-of-freedom slidingjoint that does not have any closed cavities.Also, the rectangular shape entirely avoidedthe SLA machine software’s linear approxi-mation of curved lines resulting in closerclearance of parts. This joint was fabricated inthe upright position along its length and isshown in Figure 5. The completion of thisbuild cycle took approximately 6 hours.

FIGURE 3. Type I of Spherical Joint BuiltWith SL (all dimensions are in mm)

FIGURE 4. Type II of Spherical Joints BuiltWith SL (all dimensions are in mm)

FIGURE 5. Prismatic Joint Built With SL (alldimensions are in mm)

The universal joint design was that of a classicalcross-type assembly, which consisted of two dif-ferent constitutive components; the two yokesand the connecting cross hub. One of the twoinitial universal joint designs is shown in Figure6. Using design reference to the revolute joint,the universal joint was also fabricated using the0.5 mm circular surface clearance. This joint wasfabricated in the upright position along its lengthand a picture of the prototype is also shown inFigure 6. The completion of this build cycle tookapproximately 8.4 hours.When building joints at an oblique configuration,instead of a vertical one (upright configuration), aspecial effect appears called the “step effect” or“staircase effect”. This effect can reduce thequality of fabrication of the joint. This is the re-sult of approximating a continuous curved sur-face in the vertical direction with a discrete set ofhorizontal thin layers. Obviously, the thinner thelayer or building the part in an orientation closerto a vertical configuration reduces this effect.

FIGURE 6. Universal Joint Built With SLSLS Fabricated JointsJoints of a robotic finger were built using theSelective Laser Sintering (SLS) Sinterstation2000 of DTM Corporation of Austin, TX. Thebasis material chosen was a polyamide. Clear-ances similar to that established for joints fab-ricated using the Stereolithography SLA 190process were used as trial values. These val-ues are conservative as the Sinterstation 2000has a greater level of accuracy, +0.005"-0.0075" versus +0.0075", and a smaller layerthickness, 0.005" versus 0.006", over that ofthe SLA 190.Two different types of joints were fabricatedwith the SLS machine: a revolute joint and aspherical joint. Both joints are parts of a rapidprototyped robotic hand that is presented inSection 4.2. Because of their use in a robotichand the joints needed to satisfy specific de-sign criteria.The revolute joint, that connects the ends oftwo links, is restricted to approximately 100°of revolution. This limitation on the range ofmotion was accomplished through the round-ing of just one side of the yoke section of thefixed link side. As can be seen in Figure 7,this rounding allows the links to clear eachother in revolution only through the desiredrange of motion.

FIGURE 7. Revolute JointThe spherical joint, which serves as the“knuckle” at the finger-palm interface (seeFigure 13), is to have approximately 90° ofrevolution and about +15° of side-to-side free-dom in the fully extended configuration and 0°of side-to-side freedom in the fully contractedconfiguration. This is an approximation on therange of motion present in an average humanfinger. The limitation on spherical range ofmotion was achieved by slotting the socketsection in a shape as seen in Figure 8a. An-other restriction was that of minimizing therange of twist about the extended-finger axis.This was accomplished by removing materialfrom diametrical hemispheres of the inner balland adding material to the inner section of thesocket resulting in a modified spherical joint(Figure 8b). The combination of the modifiedball and the slotted socket (Figure 9) will notfully restrict but will serve to limit the range oftwist to approximately +10°; a value accept-able in preliminary prototypes.

FIGURE 8a. View of Modified Socket

FIGURE 8b. View of Modified Ball

FIGURE 9. Modified Spherical “Knuckle”Joint

As seen in Figure 10, using the SLS process tobuild non-assembly type joints proved suc-cessful. Both parts exhibited good mobilitythrough the desired ranges of motion. Also,the presence of the “staircase effect” was re-duced due to the Sinterstation process’ thinlayer thickness. The fabrication of these twojoints is the first verification step in the robotichand construction.

The improvement in rounded surface qualityhas important consequences in the design ofthese joints. First, the decreased “step effect”means that the clearances between moving sur-faces can be reduced to a lower value. An-other important effect is the diminished im-portance of part orientation during the fabrica-tion process in rounded as well as in flat andsloped components. Selective Laser Sinteredparts do not require any computer or manuallygenerated support structures. This is an im-portant process advantage. Parts do not haveto be designed with the consideration of sup-port structure placement in part orientation

during the build cycle and in addition, the taskof support removal is eliminated.

FIGURE 10. SLS Fabricated Joints

Fabrication of Complex Mechanisms:

Parallel ManipulatorsA three-legged 6-DOF parallel manipulator waschosen as the first mechanism to be fabricatedwith the Stereolithography SLA 190. First, oneleg of the parallel manipulator was built to dem-onstrate the feasibility. The leg of the parallelmanipulator consists of two spherical joints, oneat each end, and a prismatic joint at the middle ofthe leg. The designs of the joints presented inSection 3.1 were used. The fabricated legs areshown in Figure 11.The final platform was built with three identicallegs. The top and bottom platforms are two trian-gular ternary links (links connected to threejoints) having the same dimensions of 1"x1"x1"on three sides of the triangle and the thicknesswas 0.07874" (2 mm). To save fabrication time,the triangular platforms were not completelyfilled. Since the fabrication of each joint wastested separately, the fabrication of the leg wasfavorably completed. In Figure 12, pictures of thefabricated rapid prototype of the parallel ma-nipulator are shown in two different configura-tions. This prototype was built overnight during a12-hour period.

FIGURE 11. Leg of the Parallel Manipulator

FIGURE 12. Three Legged Parallel Ma-nipulator

Robotic HandAnother complex system that has been de-signed and is currently being fabricated is thatof a robotic hand with five fingers and a palm;all constructed as one non-assembly typemechanism (Figure 13). This robotic hand isdesigned with the future purpose of using aRapidly Prototyped robotic hand as a possiblereplacement for mechanically driven prosthetichands. The fingers are composed of three cy-lindrical links connected by two revolute joints(Figure 14). Each of the fingers is to be at-tached to the palm section by modified spheri-cal joints (see Figures 8a and 8b).

FIGURE 13. CAD Rendering of RoboticHand

FIGURE 14. CAD View of A Robotic Fin-ger

Actuation of the robotic hand is to be achievedthrough a combination of cables and ShapeMemory Alloy artificial muscle wires. ShapeMemory Alloy (SMA) wires are characterizedby a reduction in length when resistive heat isgenerated through the length of the wire by theflow of electricity. The SMA wires to be usedin this design have a wire diameter of ap-proximately 150 microns (0.006"). The cableswill be connected close to the pivot points ofthe revolute joints and will run through the in-corporated “pathways” (Figure 15) within thelength of the fingers. The cables will becrimped to SMA muscle wires proximal to thepalm. It is necessary to run the cables throughthe hand rather than the SMAs themselves asthe activation temperature of the SMA is 70 -90°C and the melting temperature of the SLmaterial is 85°C. (The melting temperature isnot a consideration with the SLS material as itsmelting temperature is 185°C.) The use ofSMA wires will be the first attempt at actuat-ing a Rapidly Prototyped system at RutgersUniversity.

Figure 15 shows a cutaway view of the chan-nels within the finger for routing of the cablesused for actuation. These pathways, 0.1 inchesin diameter, were designed to guide the cablesthrough the finger links so as to decrease stresson the cables while maintaining the requiredtension. One channel through the middle linkappears in a diagonal pattern, as seen from this

side view, for the above-mentioned purpose.In addition, the cables intended for flexion andextension of the distal link travel over top ofthe revolute joint between the middle andproximal links so that the movement of thesetwo revolute joints is uncoupled. The channelsthrough the proximal link fan out around theball and socket (which could be seen from adorsal view). Two cable pathways, one forflexion and one for extension of the modifiedspherical joint, run through the socket on thepalmar and dorsal sides to allow for 90O rota-tion during actuation. Currently, abductionand adduction movement at the finger spheri-cal joint is passive.

FIGURE 15. Cable Channels in SLS FingerAnother important design consideration in thisrobotic hand prototype is to have the samerange of motion and similarity in size to that ofan average human hand. The former guidingdesign feature necessitated the use of modifiedjoint designs to partially restrict some of thedegrees of freedom. The design features andfabrication of these joints was presented inSection 3.2.Presently, a single robotic finger has been fab-ricated using the SLS process as shown in Fig-ures 16a and b. The SLS procedure here wassimilar to that of the revolute and sphericaljoints described in Section 3.2. The SLS proc-ess produced a quality, fully assembled fingerwith good joint mobility, clearances, andranges of motion. The SLS glass filled nylonmaterial used for this assembly provided for

less joint friction than the previously polyam-ide fabricated single joints. The clearances aresuch that there is some additional movement inthe joints than desired, so it would be possibleto reduce these tolerances in the future. As inthe previous SLS constructions, the joints fullyreach the designed range of motion. Addition-ally, all the pathways and cable assembly holeswere clear of material. The Selective LaserSintering process successfully fabricated thismulti-joint, multi-degree-of-freedom roboticfinger. Figure 16b shows the SLS fabricatedfinger with the tendons (cables) attached in apost-fabrication phase.

(a)

(b)

FIGURE 16. Robotic Finger Built With SLSand Actuated With SMA Wires

Discussion: The joints and systems fabricatedusing the Stereolithography machine modelSLA 190 as well as the SLS Sinterstation 2000produced quality plastic parts. Joints fabri-cated using both of these prototyping methods

exhibited good overall movement characteris-tics and near-surface clearing.The joints fabricated using the Stereolithogra-phy SLA 190 included spherical, revolute andprismatic type joints. These joints were de-signed with clearances of 0.3 mm and 0.5 mmfor flat and circular near surfaces, respectively.An essential design consideration in the SLprocess is the support structure requirement.In addition to an initial base support structurenecessary prior to initial part fabrication, addi-tional supports provide structural stability anda starting point for overhangs and new layersto initiate layering. The SL joints showedgood smoothness and evenness in flat verticaland horizontal surfaces. For rounded oroblique sections, the parts showed acceptablesurface agreement with CAD models.Thejoints fabricated using the Selective LaserSintering Sinterstation 2000 included sphericaland revolute type joints. These were deter-mined to be of a higher quality in overall partprototyping. This can partially be attributed tothe Sinterstation 2000’s greater level of accu-racy in planar detail and resolution in layerthickness over the SLA 190. These advan-tages directly lead to a number of machine ad-vantages.

Due to a reduced staircase effect, the sliding-friction in the joints was reduced in the SLSfabricated parts over the SL fabricated parts.Also, the SLS process produced joints withsmaller clearances and smoother rounded sur-faces. These improvements are the result of amore accurate Rapid Prototyping machine inthe Sinterstation 2000 over the SLA 190. Incontrast, the SL produced parts showed muchgreater smoothness and regularity over theSLS parts. Another important advantage isthat the Selective Laser Sintered parts do notrequire any computer or manually generatedsupport structures. Parts do not have to be de-signed with the thought of support structureplacement in part orientation during the buildcycle as well as manually performing the taskof support removal.

Both Rapid Prototyping processes constructedjoints with the desired ranges of motion andsize, however the SLS process showed moreapparent advantages over the SL process in anumber of regards. It is important to note thatthe majority quality advantages of the SLSparts over the SL parts cannot be solely attrib-uted to the Rapid Prototyping processes alone.To make a more competitive comparison be-tween the two RP processes, a higher-endmodel of the Stereolithography machine oughtbe used in SL part fabrication. This wouldprovide a better basis for comparison on areassuch as step effect, overall surface quality andminimum clearances. The model SLA 190 isno longer in current production. 3D Systemsproduces models ranging from the SLA 250 tothe fifth-generation design SLA 7000. Thethird-generation Sinterstation 2000 is currentlyone of two systems offered by DTM Corpora-tionFrom the comparison conducted using the twosomewhat unequally matched RP machines,there are a number of differences between theRP processes themselves, which will be pres-ent regardless of the machine model variation.For example, the SL process is based on thephoto-polymerization of a liquid resin whereasthe SLS process in based on the sintering ofpowders. These differences will produce partswith varying mechanical properties due tobuild material choices. Post-processing con-siderations play an increasingly important rolein complicated mechanisms and robotic sys-tems. As previously mentioned, the SLS proc-ess does not require the generation of supportstructures. This provides an advantage in partdesign freedom and prototype orientationduring the fabrication process.

As Rapid Prototyping technologies continue toimprove, mechanisms and robotic systemsbuilt using this methodology will competewith and eventually surpass those of traditionalfabrication techniques. Ideally, the clearancesnecessary to effectively compete with assem-bled components and mechanisms need to im-

prove by an order of magnitude. Avenues toapproach this need may be currently possiblethrough the use of system technologies, whichare currently under development. For exam-ple, MicroTEC of Duisburg, Germany has de-veloped a micro-stereolithography process thatcan produce parts with layers as thin as 1 µm(0.00004") [21]. This is a 150-fold reductionover the SLA 190’s 0.006" minimum slicethickness.

Rapid Prototyping has been shown to be a vi-able means of simple and quick fabrication ofprototypes for the articulated structures of ro-botic systems. Several joints and robotic sys-tems have been fabricated using this frame-work. The successful fabrication of the ro-botic hand gives further confidence in thisRapid Prototyping framework. In the future,actuation of rapidly fabricated prototypes willbe investigated through the use of ShapeMemory Alloy artificial muscles or other typesof smart materials.

Acknowledgments: This work is supported bya CAREER grant (DMI-9984051) from theNational Science Foundation. The Center forComputational Design of Rutgers University,and the NASA Space Grant Consortium pro-vided additional support. Kathryn DeLaurentisis supported by a National Science FoundationGraduate Research Fellowship.

References[1] Ashley S., 1995, “ Rapid Prototyping is Com-

ing of Age”, Mechanical Engineering, Vol.117, No. 7, pp. 63-68.

[2] Ashley S., 1998, “RP Industry’s GrowingPains,” Mechanical Engineering, Vol. 120, No.7, pp. 64-67.

[3] Tse W. C. and Chen Y. H., 1997, "A RoboticSystems for Rapid Prototyping," Proceedingsof the 1997 IEEE International Conference onRobotics and Automation, Albuquerque, NM,pp. 1815-1820.

[4] Vergeest J. and Tangelder W., 1996, "RobotMachines Rapid Prototype," Industrial Robot,Vol. 23, No. 5, pp. 17-20.

[5] Laliberté T., Gosselin C. and Côté G., 1999,"Rapid Prototyping of Mechanisms", Pro-ceedings of the 10th World Congress on theTheory of Machines and Mechanisms, Oulu,Finland, Vol. 3, pp. 959-964.

[6] Laliberté T., Gosselin C. and Côté G., 2000,"Rapid Prototyping of Lower-Pair, Geared-Pairand Cam Mechanisms" Proceedings of the2000 ASME Mechanisms and Robotics Confer-ence, Baltimore MD, September 10-13, 2000,Paper DETC2000/MECH-14202.

[7] Cham, J.G., Pruitt,B.L., Cutkosky, M.R., Bin-nard, M., Weiss, L., Neplotnik, G., 1999,"Layered Manufacturing with EmbeddedComponents: Process Planning Issues," Pro-ceedings of the 1999 ASME DETC/DFM Con-ference, Las Vegas, NV, Sept 12-15, 1999, Pa-per DETC99/DFM-8910.

[8] "Bailey, S. A., Cham J., G., Cutkosky, M. R.,Full, R. J., 2000, "Biomimetic Robotic Mecha-nisms via Shape Deposition Manufacturing,"Robotics Research: the Ninth InternationalSymposium, John Hollerbach and DanKoditschek (Eds), Springer-Verlag, London.

[9] Rajagopalan, S. and Cutkosky, M. R., 1998,"Tolerance Representation for Mechanism As-semblies in Layered Manufacturing", Pro-ceedings of the 1998 ASME DETC/DFM Con-ference, Atlanta, GA, Sept 13-16, 1998, PaperDETC98/DFM-5726.

[10] Rajagopalan, S. and Cutkosky, M.R., 1999,"Optimal Pose Selection for In-Situ Fabricationof Planar Mechanisms," Proceedings of the1999 ASME, DETC/DFM Conference, Las Ve-gas, NV, Sept 12-15, 1999, PaperDETC99/DFM-8958.

[11] Goel, P., Rajagopalan, S., and Cutkosky, M.R.,2000, "Error Analysis for the In-Situ Fabrica-tion of Mechanisms," Proceedings of the 2000ASME DETC/DFM Conference, Baltimore,MD, Sept. 13-16, 2000, PaperDETC2000/DFM-14019.

[12] Binnard, M., 1999, Design by Composition forRapid Prototyping, Kluwer Academic Press.

[13] Kataria A. and Rosen D., 2000, "BuildingAround Inserts: Methods for Fabricating Com-plex Devices in Stereolithography," Proceed-ings of the 2000 ASME Mechanisms and Ro-botics Conference, Baltimore MD, September10-13, 2000. Paper DETC2000/MECH-14206.

[14] Geving B. and Ebert-Uphoff I., 2000, "E n-hancement of Stereo-Lithography Technologyto Support Building Around Inserts," Pro-

ceedings of the 2000 ASME Mechanisms andRobotics Conference, Baltimore MD, Septem-ber 10-13, 2000. Paper DETC2000/MECH-14207.

[15] Alam M., Mavroidis C., Langrana N. and Bi-daud P., 1999, "Mechanism Design usingRapid Prototyping", Proceedings of the 10th

World Congress on the Theory of Machinesand Mechanisms, Oulu, Finland, Vol. 3, pp.930-938.

[16] Won J., DeLaurentis K. and Mavroidis C.,“Rapid Prototyping of Robotic Systems,” Pro-ceedings of the 2000 IEEE International Con-ference on Robotics and Automation, April 24-28 2000, San Francisco, CA, pp. 3077-3082.

[17] Won J., DeLaurentis K. and Mavroidis C.,"Fabrication of a Robotic Hand Using RapidPrototyping," Proceedings of the 2000 ASMEMechanisms and Robotics Conference, Balti-more MD, September 10-13, 2000. PaperDETC2000/MECH-14203.

[18] Dolenc A., "An Overview of RapidPrototyping Technologies in Manufacturing",http://www.cs.hut.fi/~ado/rp/rp.html, HelsinkiUniv. of Technology, 1994.

[19] Crawford R. and Beaman J., 1999, "Solid Free-form Fabrication: A New Manufacturing Para-digm," IEEE Spectrum, February 1999, pp. 35-43.

[20] Bylinsky G., "Industry's Amazing Instant Pro-totypes", Fortune, Vol. 137 No. 1, 1998.

[21] Wohlers T., Rapid Prototyping and ToolingState of the Industry: 1999 Worldwide Prog-ress Report, Wohlers Associates, 1999.