Fabrication of Non-Assembly Mechanisms and Robotic Systems Using Rapid Prototyping

Transcription

1 Constantinos Mavroidis 1 Associate Professor, ASME Mem mavro@jove.rutgers.edu Kathryn J. DeLaurentis Graduate Student, ASME Student Mem kjdela@jove.rutgers.edu Jey Won Undergraduate Research Assistant, ASME Student Mem Munshi Alam Graduate Student, ASME Student Mem Robotics and Mechatronics Laboratory, Department of Mechanical and Aerospace Engineering, Rutgers University, The State University of New Jersey, 98 Brett Rd., Piscataway, NJ Fabrication of Non-Assembly Mechanisms and Robotic Systems Using Rapid Prototyping In this paper, the application of Rapid Prototyping in fabricating non-assembly robotic systems and mechanisms is presented. Using two Rapid Prototyping techniques, Stereolithography and Selective Laser Sintering, prototypes of mechanical mobile joints were fabricated. The designs of these component joints were then used to fabricate the articulated structure of experimental prototypes for two robotic systems: (1) a three-legged parallel manipulator, (2) a four degree-of-freedom finger of a five-fingered robotic hand. These complex multi-articulated, multi-link, multi-loop systems have been fabricated in one step, without requiring assembly while maintaining their desired mobility. The feasibility and usefulness of Rapid Prototyping as a method for the fabrication of these nonassembly type mechanisms and robotic systems is the focus of this work. DOI: / Introduction It is always desirable to evaluate a proposed robot design prior to full prototyping to ensure the swiftest and most cost effective design changes. Even though powerful three-dimensional Computer Aided Design and Dynamic Analysis software packages such as Pro/ENGINEER, IDEAS, ADAMS and Working Model 3-D are now being used, they cannot provide important visual, haptic and realistic workspace information for the proposed design. In addition, there is a great need for developing methodologies and techniques that will allow fast design, fabrication and testing of robotic and other multi-articulated mechanical systems. A framework for the feasibility and usefulness of applying Rapid Prototyping in fabricating mechanisms and robotic systems is presented here. Rapid Prototyping RP, also known as Layered Manufacturing or Solid Freeform Fabrication, is a technique for fabricating a three-dimensional solid model in a layer-by-layer manner through the fusion of material under computer control. Rapid prototyping of parts and tools is a rapidly developing technology 1,2 that provides many advantages: a time and money savings, b quick product testing, c expeditious design improvements, d fast error elimination from design, e increased product sales, and f rapid manufacturing. Its main advantage is early verification of product designs. Additionally, Rapid Prototyping is quickly becoming a valuable key for efficient and concurrent engineering. Through different techniques, engineers and designers are now able to bring a new product from concept modeling to part testing in a matter of weeks or months. In some instances, actual part production may even be possible in very short time. Rapid Prototyping has indeed simplified the task of describing a concept to design teams, illustrating details to engineering groups, specifying parts to purchasing departments, and selling the product to customers. The advantages of using Rapid Prototyping techniques in 1 Corresponding author. Contributed by the Mechanisms and Robotics Committee for publication in the JOURNAL OF MECHANICAL DESIGN. Manuscript received Aug Associate Editor: J. S. Rastegar. mechanism and robot design and fabrication are numerous. With Rapid Prototyping techniques, physical prototypes of the mechanism or robot that is being designed can be obtained in a very short time, thus making quick design evaluation possible. By using these physical prototypes, several properties of the mechanisms can be evaluated immediately such as: a workspace evaluation, b identification of singular configurations including uncertain configurations where the mechanism has internal mobility, c determination of link interference, and d visualization of joint limits. Evaluating these fundamental properties of the Rapid Prototyped mechanism can considerably reduce the time and improve the quality of the design process. Furthermore, Rapid Prototyping allows the fabrication of complex three-dimensional structures, which could not be produced with conventional fabrication processes. Such possibilities make the incorporation and attachment of sensors, actuators and transmission elements within the structure and joints of the mechanism easier. Finally, Rapid Prototyping allows one-step fabrication of multi-articulated, multi-link systems as a whole, without requiring assembly of its structural members and joints after fabrication. This one step fabrication technique can drastically change the way that mechanisms and robots are currently built. This is very important as it can allow rapid fabrication of fully functional and mobile systems. In this paper, these mechanisms and robotic systems whose multiarticulated structure is built in one step are called non-assembly mechanisms. Robotic systems have been used as a part of a Rapid Prototyping process 3,4. However, the application of Rapid Prototyping in robot and mechanism design and fabrication has been very limited. Professor Gosselin and his group at Laval University, using a Fused Deposition Modeling Rapid Prototyping machine, fabricated several mechanisms such as a six-legged six degree-offreedom parallel manipulator 5,6. These rapidly manufactured mechanisms required assembly after Rapid Prototyping of the mechanism parts. Professor Cutkosky and his group at Stanford University using a different Rapid Prototyping process called Shape Deposition Manufacturing developed planar, non-assembly mechanisms and robotic systems with embedded sensors and actuators 7,8. These components were inserted in the multi- 516 Õ Vol. 123, DECEMBER 2001 Copyright 2001 by ASME Transactions of the ASME

2 articulated structure during its fabrication as opposed to their integration in post-fabrication assembly phases. This group also proposed methods for performing the systematic design, error analysis and optimal pose selection for these mechanisms Additionally, researchers at the Georgia Institute of technology, proposed methods to develop complex devices with embedded components using the Stereolithography technique 13,14. During the last two years our group at Rutgers University has studied the fabrication of non-assembly mechanism using two different Rapid Prototyping processes: Stereolithography SL and Selective Laser Sintering SLS. Preliminary results of our work were presented in In this paper our design considerations, results, and prototypes in building complex, spatial, nonassembly mechanisms are presented in detail. To the authors knowledge, this is the first successful fabrication of multi-joint, multi degree-of-freedom, spatial robotic systems and mechanisms without requiring any assembly using the SL and SLS processes. The demonstration of non-assembly, mechanism fabrication, including discussions on manufacturing accuracies, joint clearances and the required design modifications that take into account the special characteristics of Rapid Prototyping techniques is the primary scope of the paper. Our future goals, which are outside the scopes of this paper, will include the detailed study of how to use these novel mechanism fabrication techniques for mechanism design evaluation and the development of theoretical and mathematical tools needed in the design optimization and automation. Using the Stereolithography machine model SLA 190, from 3D Systems, CA a set of joints that include revolute, prismatic, universal and spherical joints, were fabricated. In addition, a threelegged, six degree-of-freedom Rapid Prototype of a parallel manipulator was built in one step, without requiring assembly. Each leg of this three-legged parallel manipulator is composed of one prismatic joint and two spherical joints, which connect to the two triangular platforms. Further prototype joints, similar to those fabricated with the Stereolithography process, were Rapidly Prototyped using the Selective Laser Sintering Sinterstation 2000 of DTM Corporation of Austin, TX. Finally, a four-degree-offreedom finger of a robotic hand with four fingers, a thumb and a palm are constructed as non-assembly type mechanisms. Actuation of the joints of the rapidly prototyped systems is outside the scope of this paper and will be studied in our future work. 2 Rapid Prototyping Rapid Prototyping or Layered Manufacturing is a fabrication technique where three-dimensional solid models are constructed layer upon layer by the fusion of material under computer control. This process generally consists of a substance, such as fluids, waxes, powders or laminates, which serves as the basis for model construction as well as sophisticated computer-automated equipment to control the processing techniques such as deposition, sintering, lasing, etc. 18,19 Also referred to as Solid Freeform Fabrication, Rapid Prototyping complements existing conventional manufacturing methods of material removing and forming. It is widely used for the rapid fabrication of physical prototypes of functional parts, patterns for molds, medical prototypes such as implants, bones and consumer products 20. Its main advantage is early verification of product designs. Through quick design and error elimination, Rapidly Prototyped parts show great cost savings over traditionally prototypes parts in the total product life cycle 21. Currently, there are over 30 different types of Rapid Prototyping processes in existence. Some of these techniques are commercially available while others are still in development in research laboratories 21. Over the years, major improvements in the overall quality of prototyped parts have been achieved through enhancements in accuracy, material choice and durability, part throughput, surface texture, and alternative RP processes. These improvements have led to an evolution of the functionality of RP prototypes 19,20. Evolution of the techniques and applications of Rapid Prototyping is a continually developing and expanding field. Current research is leading to a more functional rapidly prototyped part with an increasing number of applications and part feature enhancements. The two Rapid Prototypes techniques, Stereolithography SL and the Selective Laser Sintering SLS, used here are described in more detail in the pages that follow. 2.1 Stereolithography. Stereolithography SL is a threedimensional building process, which produces a solid plastic model. In this process, an ultraviolet UV laser traces twodimensional cross-sections on the surface of a photosensitive liquid plastic resin. The laser partially cures the resin through low energy absorption of laser light thus producing a solid. The first cross-sectional slice is built on a depth-controlled platform, which is fully submerged under the first thin layer of resin. This and each successive thin layer of liquid resin has a depth equal to that of the vertical slice thickness of the part. After each slice is traced on the surface of resin, the platform lowers by a depth equal to that of the slice thickness. Successive 2-D slices are cured directly onto the previous layer as the part is built from bottom to top. Support structures are needed to maintain the structural integrity of the part and supports overhangs, as well as provide a starting point for the overhangs and for successive layers on which to be built. These supports are constructed from a fine lattice structure of cured resin. After the part is fully built, the support structures are removed and the part is cleaned in a bath of solvent, and air-dried. The prepared parts are then flooded with high-intensity UV light in a Post-Cure Apparatus PCA to fully cure the resin. The Department of Mechanical and Aerospace Engineering of Rutgers University is equipped with a SLA 190 machine. Available basis materials for this machine include photo-polymer resin epoxies with various physical properties. 2.2 Selective Laser Sintering. Selective Laser Sintering SLS is a three-dimensional building process based on the sintering of a metallic or non-metallic powder by a laser. The SLS process involves the heating of the powder using a CO 2 localized laser beam. This localized heating raises the temperature of the powder such that solidification by fusion occurs without melting. The model is built on a platform that is situated within a horizontal platen. The platform, which is initially flush with the platen, is lowered a depth equal to that of the slice thickness. A powder is then rolled, scraped or slot-fed onto the platform and then the laser draws the two-dimensional cross-section. This lowering, powdering and lasing process is repeated until the part is complete. Unlike the SL process, no support structures are necessary for SLS since the part rests on and within the non-sintered powder. Post-curing is not necessary except in the case of ceramic parts. Available materials include polycarbonates, nylons, polyamides, elastomers, sand casting materials and steels. The fabrication of the SLS prototype parts used in the present investigation was done through a professional Rapid Prototyping service provider for SLS manufacturing. 3 Joint Fabrication 3.1 SL Fabricated Joints. The first step in building robotic systems with a Rapid Prototyping machine is to be able to successfully fabricate joints. Different types of mechanical joints such as revolute, spherical, primatic and universal joints were fabricated with the SLA machine Fig. 1. These joints were produced without requiring assembly. Through a trial and error process, different features such as clearance, part size and support structure generation were optimized to produce working mechanical joints. Of these features, determination of clearances was very important in successful part fabrication. The optimum clearances for the SLA 190 Cibatool SL 5170 resin, between two near surfaces, were determined to be 0.3 mm for flat surfaces and 0.5 mm for circular surfaces. Also, Journal of Mechanical Design DECEMBER 2001, Vol. 123 Õ 517

3 Fig. 1 Stereolithography joint fabrications Fig. 3 Type I of spherical joint built with SL all dimensions are in mm Fig. 2 Revolute joint built with SL all dimensions are in mm the size of these parts is in the order of a few centimeters. This was done to determine the limits of the available apparatus as well as to conserve processing time and material. The tests to determine these clearances were performed only for the SLA 190 SL 5170 resin, at a layer thickness setting of mm The trials were systematic in that the initial clearances for both the flat and circular surfaces began at one millimeter and decreased by 0.1 mm each successive build until the joint was no longer mobile. The clearances were then increased by 0.05 mm until the joint mobility was clear and freely moving. Finally, the optimal joint clearances were found when there was enough space between the surfaces of the joint to allow free movement yet not too great that supports were built between the surfaces to prevent movement. The above tolerances have consistently given the same results for identical parts as well as a variety of different parts made over the past two years. In Fig. 2 the CAD drawings and pictures of the revolute joints are shown. The revolute joint consists of two rings, each with a small stem attached. The two rings are connected, through the centers of the ring holes, by a pin that acts as the axis revolution. Spherical joints with varying sizes for the ball and socket were built. Two slightly different designs of spherical joints were fabricated. In the first design, shown in Fig. 3, the bottom of the socket is cut slightly to create an opening for the support structure to go through and hold the ball in place, so that the ball would not fuse to the socket during fabrication. This spherical joint was oriented as shown in the line drawing such that the ball and socket part was built first before building the link arm. The second type of design for a spherical joint is shown in Fig. 4. In this design, since the ball and socket was built last the build orientation is shown in the line drawing of Fig. 4, the ball did not need any support as it was supported by the vertical link itself. The approximate time of fabrication for this joint was 5.75 hours. Initially, the prismatic joint or sliding joint was characterized by that of a piston-cylinder type assembly. Poor sliding of the joint resulted from a volume of trapped resin in the chamber that could not be effectively or easily removed prior to final ultra-violet curing. Also evident in the joint assembly was the presence of undesired additional degrees of freedom. The joint was then redesigned with these two factors as limiting constraints. The final design of the joint is a one degree-of-freedom sliding joint that does not have any closed cavities. Also, the rectangular shape entirely avoided the SLA machine software s linear approximation of curved lines resulting in closer clearance of parts. This joint was 518 Õ Vol. 123, DECEMBER 2001 Transactions of the ASME

4 Fig. 6 Universal joint built with SL Fig. 4 Type II of spherical joints built with SL all dimensions areinmm fabricated in the upright position along its length and is shown in Fig. 5. The completion of this build cycle took approximately 6 hours. The universal joint design is a classical cross-type assembly, which consists of two different constitutive components; the two yokes and the connecting cross hub. One of the two initial universal joint designs is shown in Fig. 6. The outer diameter of the links of the universal joint shown in Fig. 6 is mm 0.8, the inner diameter is mm 0.6 and the total length is mm Using the revolute joint as a design reference, the universal joint was also fabricated using the 0.5 mm circular surface clearance. This joint was fabricated in the upright position along its length and a picture of the prototype is also shown in Fig. 6. The completion of this build cycle took approximately 8.4 hours. When building joints at an oblique configuration, instead of a vertical one upright configuration, a special effect appears called the step effect or staircase effect. This effect can reduce the quality of fabrication of the joint. This is the result of approximating a continuous curved surface in the vertical direction with a discrete set of horizontal thin layers. Obviously, the thinner the layer or building the part in an orientation closer to a vertical configuration reduces this effect. 3.2 SLS Fabricated Joints. Joints of a robotic finger were built using Selective Laser Sintering. The basis material chosen was a polyamide. Clearances similar to that established for joints fabricated using the Stereolithography SLA 190 process were used Fig. 5 Prismatic joint built with SL all dimensions are in mm Journal of Mechanical Design DECEMBER 2001, Vol. 123 Õ 519

5 Fig. 9 Modified spherical Knuckle joint Fig. 7 Revolute joint as trial values. These values are conservative as the Sinterstation 2000 has a greater level of accuracy, versus , and a smaller layer thickness, versus 0.006, over that of the SLA 190. Two different types of joints were fabricated with the SLS machine: revolute joint and a spherical joint. Both joints are parts of a rapid prototyped robotic hand that is presented in Section 4.2. Because of their use in a robotic hand the joints needed to satisfy specific design criteria. The revolute joint, shown in Fig. 7, connects the ends of two links of the finger. The range of motion for this joint is restricted to approximately 100 of revolution. This limitation on the range of motion was accomplished through the rounding of just one side of the yoke section of the fixed link side. As can be seen in the figure, the other side was not rounded to act as a stop. Often, revolute joints provide a full 180 range of motion. Thus, this rounding allows the links to clear each other in revolution only through the desired range of motion. The spherical joint, which serves as the knuckle at the fingerpalm interface see Fig. 13, is to have approximately 90 of revolution and about 15 of side-to-side freedom in the fully extended configuration and 0 of side-to-side freedom in the fully contracted configuration. This is an approximation on the range of motion present in an average human finger. The limitation on spherical range of motion was achieved by slotting the socket section in a shape as seen in Fig. 8 a. Another restriction was that of minimizing the range of twist about the extended-finger axis. This was accomplished by removing material from diametrical hemispheres of the inner ball and adding material to the inner section of the socket resulting in a modified spherical joint Fig. 8 b. The combination of the modified ball and the slotted socket Fig. 9 will not fully restrict but will serve to limit the range of twist to approximately 10 ; a value acceptable in preliminary prototypes. As seen in Fig. 10, using the SLS process to build nonassembly type joints proved successful. Both parts exhibited good mobility through the desired ranges of motion. Also, the presence of the staircase effect was reduced due to the Sinterstation process thin layer thickness. The fabrication of these two joints is the first verification step in the robotic hand construction. Note, that since the fabrication of these parts was through a professional Rapid Prototyping service provider for SLS manufacturing the build duration was not documented. The improvement in rounded surface quality has important consequences in the design of these joints. First, the decreased step effect means that the clearances between moving surfaces can be reduced to a lower value. Another important effect is the diminished importance of part orientation during the fabrication process in rounded as well as in flat and sloped components. Selective Laser Sintered parts do not require any computer or manually generated support structures. This is an important process advantage. Parts do not have to be designed with the consideration of support structure placement in part orientation during the build cycle and in addition, the task of support removal is eliminated. Fig. 8 Views of a modified socket and b modified ball Fig. 10 SLS fabricated joints 520 Õ Vol. 123, DECEMBER 2001 Transactions of the ASME

6 Fig. 11 A leg of the parallel manipulator 4 Fabrication of Complex Mechanisms 4.1 Parallel Manipulators. A three-legged 6-DOF parallel manipulator was chosen as the first mechanism to be fabricated with the Stereolithography SLA 190. First, one leg of the parallel manipulator was built to demonstrate the feasibility. The leg of the parallel manipulator consists of two spherical joints, one at each end, and a prismatic joint at the middle of the leg. The designs of the joints presented in Section 3.1 were used. The fabricated legs are shown in Fig. 11. The final platform was built with three identical legs. The top and bottom platforms are two triangular ternary links links connected to three joints having the same dimensions of 2.54cm 2.54cm 2.54cm on three sides of the triangle and the thickness was 2 mm To save fabrication time, the triangular platforms were not completely filled. Since the fabrication of each joint was tested separately, the fabrication of the leg was favorably completed. In Fig. 12, pictures of the fabricated rapid prototype of the parallel manipulator are shown in two different configurations. This prototype was built overnight during a 12-hour period. 4.2 Robotic Hand. Another complex system that has been designed and is currently being fabricated is that of a robotic hand with five fingers and a palm; all constructed as one non-assembly type mechanism Fig. 13. This robotic hand is designed with the future purpose of using a Rapidly Prototyped robotic hand as a possible replacement for mechanically driven prosthetic hands. The fingers are composed of three cylindrical links connected by two revolute joints Fig. 14. Each of the fingers is to be attached to the palm section by modified spherical joints see Figs. 8 a and 8 b. Actuation of the robotic hand is to be achieved through a combination of cables and Shape Memory Alloy artificial muscle wires. Shape Memory Alloy SMA wires are characterized by a reduction in length when resistive heat is generated through the length of the wire by the flow of electricity. The SMA wires to be used in this design have a wire diameter of approximately 150 microns The cables will be connected close to the pivot points of the revolute joints and will run through the incorporated pathways Fig. 15 within the length of the fingers. The cables will be crimped to SMA muscle wires proximal to the palm. It is necessary to run the cables through the hand rather than the SMAs themselves as the activation temperature of the SMA is C and the melting temperature of the SL material is 85 C. The melting temperature is not a consideration with the SLS material as its melting temperature is 185 C. The use of SMA wires will be the first attempt at actuating a Rapidly Prototyped system at Rutgers University. Figure 15 shows a cutaway view of the channels within the finger for routing of the cables used for actuation. These path- Fig. 12 Three legged parallel manipulator Journal of Mechanical Design DECEMBER 2001, Vol. 123 Õ 521

7 Fig. 13 CAD rendering of robotic hand Fig. 14 CAD view of a robotic finger Fig. 16 wires Robotic finger built with SLS and actuated with SMA Fig. 15 Cable channels in SLS finger ways, 2.54 mm 0.1 in diameter, were designed to guide the cables through the finger links so as to decrease stress on the cables while maintaining the required tension. One channel through the middle link appears in a diagonal pattern, as seen from this side view, for the above-mentioned purpose. In addition, the cables intended for flexion and extension of the distal link travel over top of the revolute joint between the middle and proximal last links so that the movement of these two revolute joints is uncoupled. The channels through the proximal link fan out around the ball and socket which could be seen from a dorsal or top view. Two cable pathways, one for flexion and one for extension of the modified spherical joint, run through the socket on the palmar and dorsal back of hand sides to allow for 90 rotation during actuation. Currently, abduction and adduction side to side movement at the finger spherical joint is passive. Another important design consideration in this robotic hand prototype is to have the same range of motion and similarity in size to that of an average human hand. The former guiding design feature necessitated the use of modified joint designs to partially restrict some of the degrees of freedom. The design features and fabrication of these joints was presented in Section 3.2. Presently, a single robotic finger has been fabricated using the SLS process as shown in Figs. 16 a and b. The SLS procedure here was similar to that of the revolute and spherical joints described in Section 3.2. The SLS process produced a quality, fully assembled finger with good joint mobility, clearances, and ranges of motion. The SLS glass filled nylon material used for this assembly provided for less joint friction than the previously polyamide fabricated single joints. The clearances are such that there is some additional movement in the joints than desired, so it would be possible to reduce these tolerances in the future. As in the previous SLS constructions, the joints fully reach the designed range of motion. Additionally, all the pathways and cable assembly holes were clear of material. The Selective Laser Sintering process successfully fabricated this multi-joint, multi-degree-offreedom robotic finger. Figure 16 b shows the SLS fabricated finger with the tendons cables attached in a post-fabrication phase. 5 Discussion The joints and systems fabricated using the Stereolithography machine model SLA 190 as well as the SLS Sinterstation 2000 produced quality plastic parts. Joints fabricated using both of these prototyping methods exhibited good overall movement characteristics and near-surface clearing. The two RP process machines used and the main experimental features noted are presented in Fig. 17. Note that the table shown in Fig. 17 is not an all-encompassing description of the machines and their complete 522 Õ Vol. 123, DECEMBER 2001 Transactions of the ASME

8 Fig. 17 Brief comparison of two RP processes capabilities. The major results of the experimentation as it has been described in Sections 3 and 4 are summarized in Fig. 18. The joints fabricated using the Stereolithography SLA 190 included spherical, revolute and prismatic type joints. These joints were designed with clearances of 0.3 mm and 0.5 mm for flat and circular near surfaces, respectively. An essential design consideration in the SL process is the support structure requirement. In addition to an initial base support structure necessary prior to initial part fabrication, additional supports provide structural stability and a starting point for overhangs and new layers to initiate layering. The SL joints showed good smoothness and evenness in flat vertical and horizontal surfaces. For rounded or oblique sections, the parts showed acceptable surface agreement with CAD models. The joints fabricated using Selective Laser Sintering included spherical and revolute type joints. These were determined to be of a higher quality in overall part prototyping. This can partially be attributed to the Sinterstation 2000 s greater level of accuracy in planar detail and resolution in layer thickness over the SLA 190. These advantages directly lead to a number of machine advantages. Due to a reduced staircase effect, the sliding-friction in the joints was reduced in the SLS fabricated parts over the SL fabricated parts. Also, the SLS process produced joints with smaller clearances and smoother rounded surfaces. These improvements are the result of a more accurate Rapid Prototyping machine in the Sinterstation 2000 over the SLA 190. In contrast, the SL produced parts showed much greater smoothness and regularity over the SL parts. Another important advantage is that the Selective Laser Sintered parts do not require any computer or manually generated support structures. Parts do not have to be designed with the thought of support structure placement in part orientation during the build cycle as well as manually performing the task of support removal. Both Rapid Prototyping processes constructed joints with the desired ranges of motion and size, however the SLS process showed more apparent advantages over the SL process in a number of regards. It is important to note that the majority quality advantages of the SLS parts over the SL parts cannot be solely attributed to the Rapid Prototyping processes alone. To make a more competitive comparison between the two RP processes, a higher-end model of the Stereolithography machine ought be used in SL part fabrication. This would provide a better basis for comparison on areas such as step effect, overall surface quality and minimum clearances. From the comparison conducted using the two somewhat unequally matched RP machines, there are a number of differences between the RP processes themselves, which will be present regardless of the machine model variation. For example, the SL process is based on the photo-polymerization of a liquid resin whereas the SLS process is based on the sintering of powders. These differences will produce parts with varying mechanical properties due to build material choices. Post-processing considerations play an increasingly important role in complicated mechanisms and robotic systems. As previously mentioned, the SLS process does not require the generation of support structures. This provides an advantage in part design freedom and prototype orientation during the fabrication process. Though primarily a comparison between two different RP processes has been made, it is worth mentioning that RP techniques in general have provided benefit. The time saved for building prototypes using one of these methods is significant when compared with traditional building techniques. For example, a Rutgers aluminum finger prototype similar in design to the RP finger shown in Figs. 16 a and 16 b took approximately two weeks to build versus hours to build these prototypes. Also, Fig. 16 shows two different design types that were easily fabricated for comparison. The cable channels shown in Fig. 15, and implemented into the second design Fig. 16 b were an improvement over the previous design with respect to the desired cable movement provided. This type of design feature was easily achieved with the use of the RP process. Additionally, complex systems were manufactured using the techniques mentioned in this paper that are difficult or impossible to do with traditional forms of manufacturing. An example of this is the modified spherical joint Figs. 8 and 9. This spherical joint design is unique; and it is unlikely that these modifications could be easily made by traditional means while still maintaining the quality of mobility presented here. As Rapid Prototyping technologies continue to improve, mechanisms and robotic systems built using this methodology will compete with and eventually surpass those of traditional fabrication techniques. Ideally, the clearances necessary to effectively compete with assembled components and mechanisms need to improve by an order of magnitude. Avenues to approach this need may be currently possible through the use of system technologies, which are currently under development. For example, MicroTEC of Duisburg, Germany has developed a microstereolithography process that can produce parts with layers as thin as 1 m This is a 150-fold reduction over the SLA 190 s minimum slice thickness. Rapid Prototyping has been shown to be a viable means of simple and quick fabrication of prototypes for the articulated structures of robotic systems. Several joints and robotic systems have been fabricated using this framework. The successful fabrication of the robotic hand gives further confidence in this Rapid Prototyping framework. In the future, actuation of rapidly fabricated prototypes will be investigated through the use of Shape Memory Alloy artificial muscles or other types of smart materials. Additionally, the entire compilation of the robotic design, fabrication and actuation of the hand will be the topic of a future paper also, see 17 for further information. Fig. 18 Summary of experimental results Acknowledgments This work is supported by a CAREER grant DMI from the National Science Foundation. The Center for Computational design of Rutgers University, The Center for Advanced Information Processing CAIP, and the NASA Space Grant Con- Journal of Mechanical Design DECEMBER 2001, Vol. 123 Õ 523

9 sortium provided additional support. Kathryn DeLaurentis is supported by a National Science Foundation Graduate Research Fellowship. References 1 Ashley, S., 1995, Rapid Prototyping is Coming of Age, Mech. Eng. Am. Soc. Mech. Eng., 117, No. 7, pp Ashley, S., 1998, RP Industry s Growing Pains, Mech. Eng. Am. Soc. Mech. Eng., 120, No. 7, pp Tse, W. C., and Chen, Y. H., 1997, A Robotic System for Rapid Prototyping, Proceedings of the 1997 IEEE International Conference on Robotics and Automation, Albuquerque, NM, pp Vergeest, J., and Tangelder, W., 1996, Robot Machines Rapid Prototype, Ind. Robot, 23, No. 5, pp Laliberté, T., Gosselin, C., and Côté, G., 1999, Rapid Prototyping of Mechanisms, Proceedings of the 10th World Congress on the Theory of Machines and Mechanisms, Oulu, Finland, Vol. 3, pp Laliberté, T., Gosselin, C., and Côté, G., 2000, Rapid Prototyping of Lower- Pair, Geared-Pair and Cam Mechanisms, Proceedings of the 200 ASME Mechanisms and Robotics Conference, Baltimore MD, September 10 13, 2000, Paper DETC2000/MECH Cham, J. G., Pruitt, B. L., Cutkosky, M. R., Binnard, M., Weiss, L., and Neplotnik, G., 1999, Layered Manufacturing with Embedded Components: Process Planning Issues, Proceedings of the 1999 ASME DETC/DFM Conference, Las Vegas, NV, Sept 12 15, 1999, Paper DETC99/DFM Bailey, S. A., Cham, J. G., Cutkosky, M. R., and Full, R. J., 2000, Biomimetic Robotic Mechanisms via Shape Deposition Manufacturing, Robotics Research: The Ninth International Symposium, John Hollerbach and Dan Koditschek Eds, Springer-Verlag, London. 9 Rejagopalan, S., and Cutkosky, M. R., 1998, Tolerance Representation for Mechanism Assemblies in Layered Manufacturing, Proceedings of the 1998 ASME DETC/DFM Conference, Atlanta, GA, Sept 13 16, 1998, Paper DETC98/DFM Rajagopalan, S., and Cutkosky, M. R., 1999, Optimal Pose Selection for In-Situ Fabrication of Planar Mechanisms, Proceedings of the 1999 ASME, DETC/DFM Conference, Las Vegas, NV, Sept 12 15, 1999, Paper DETC99/ DFM Goel, P., Rajagopalan, S., and Cutkosky, M. R., 2000, Error Analysis for the In-Situ Fabrication of Mechanisms, Proceedings of the 2000 ASME DETC/ DFM Conference, Baltimore, MD, Sept 13 16, 2000, Paper DETC2000/DFM Binnard, M., 1999, Design by Composition for Rapid Prototyping, Kluwer Academic Press. 13 Kataria, A., and Rosen, D., 2000, Building Around Inserts: Methods for Fabricating Complex Devices in Stereolithography, Proceedings of the 2000 ASME Mechanisms and Robotics Conference, Baltimore MD, September 10 13, Paper DETC2000/MECH Geving, B., and Ebert-Uphoff, I., 2000, Enhancement of Stereo-Lithography Technology to Support Building Around Inserts, Procedings of the 2000 ASME Mechanism and Robotics Conference, Baltimore MD, September 10 13, Paper DETC2000/MECH Alam, M., Mavroidis, C., Langrana, N., and Bidaud, P., 1999, Mechanism Design Using Rapid Prototyping, Proceedings of the 10th World Congress on the Theory of Machines and Mechanisms, Oulu, Finland, 3, pp Won, J., DeLaurentis, K., and Mavroidis, C., Rapid Prototyping of Robotic Systems, Proceedings of the 2000 IEEE International Conference on Robotics and Automation, April , San Francisco, CA, pp Won, J., DeLaurentis, K., and Mavroidis, C., Fabrication of a Robotic Hand Using Rapid Prototyping, Proceedings of the 2000 ASME Mechanisms and Robotics Conference, Baltimore, MD, September 10 13, Paper DETC2000/MECH Dolenc, A., An Overview of Rapid Prototyping Technologies in Manufacturing, http// ado/rp/rp.html, Helsinki Univ. of Technology, Crawford, R., and Beaman, J., 1999, Solid Freeform Fabrication: A New Manufacturing Paradigm, IEEE Spect., February 1999, pp Bylinsky, G., 1998, Industry s Amazing Instant Prototypes, Fortune, 137, No Wohlers, T., Rapid Prototyping and Tooling State of the Industry: 1999 Worldwide Progress Report, Wohlers Associates, Õ Vol. 123, DECEMBER 2001 Transactions of the ASME