HandJive: A Device for Interpersonal Haptic Entertainment

Transcription

1 57 CHI APRIL 1998 PAPERS HandJive: A Device for Interpersona Haptic Entertainment BJ Fogg Lawrence D. Cuter Perry Arnod Chris Eisbach Stanford University Pixar Animation Studios Triogy Dev. Group IBM P-0, Box W. Cutting 6034 W. Courtyard Dr. User Systems Stanford, CA Richmond, CA Austin, TX Ergonomic Research USA USA USA USA bjfogg@stanford.edu arryc@pixarcom perry~arnod@triogy.com chrisbac@amaden.ibm.com ABSTRACT The paper describes how we designed and prototyped HandJive, a haptic device for interpersona entertainment. Handive is notabe because it reies entirey on haptic input and output. The design process incuded typica steps such as anayzing user needs and performing iterative prototyping 2nd testing. However, deveoping a haptic interface ike HandJive aso presented specia chaenges, such as creating rapid physica prototypes that coud withstand abuse, deveoping a preiminary system of haptic interaction, and testing haptic interfaces through ow-tech prototypes. KEYWORDS Entertainment, interaction design, haptics, product design, rapid prototyping, user testing, ubiquitous computing, tactie feedback, interpersona communication MTRODUCTON For the 1996 University Workshop at Interva Research, we deveoping a system based on the workshop theme remote pay- The chaenge in the design brief was to prototype ways in which peope coud use computer-mediated objects/interfaces together for enjoyment. Our four-person team combined a variety of academic pursuits incuding communication, psychoo,y, drama, product design, mechanica engineering, computer science, computer,mphics, and human-computer interaction. We worked on this project for approximatey three months. Athough we began without biases towards any user group or resuting technoo,v, our eventua goa became to create a haptic-ony, ubiquitous computing device for entertainment and interpersona communication. We hoped to push the computerized aspects of our device into the periphery, making the computer more transparent to the user. This briefing expains the incrementa steps in our design process, as we as the resuting system which we ca?%ndjive. Permission to make digitahrd copies ofa or part ofthis materia for persona or cassroom use is granted without fke provided that the copies are not made or distriiuted for profit or commercia advantage, the copyri&t notice. the tite of the pubication and its date appear. and notice is ~given that copyright is by p&ion ofthe ACM, Inc. To copy otherwise, to repubish, to post on servers or to redihiute to Iii requires specific pprmission an&r fee C3 98 Los Angees CA USA Copyright 199s o-x /9s14..s5.00 WHAT IS HANDJIVE? HandJive is a handhed object that aows remote pay through haptic input and output. Specificay, HandJive is a device that: fits in one hand moves in interesting ways (so you can fidget with it) can communicate wireessy with simiar devices has ony haptic input and output (no visuas, no sound) sends movements to another HandJive device receives movements from another HandJive device RELATED WORK ON HAPTICS Haptic research in the Computer Graphics and HCI communities has bossomed during the past few years, in part due to the deveopment of high-fideity force feedback devices such as the PHANTOM [9]. Much work has focused on simuating our physica environment [S, 14, 161 and on augmenting GUI interfaces with haptic technoogy [7, 11, 121. As an aternative form of haptic feedback, other systems [3, 4, 10, 151 use physica, everyday objects as input devices to provide users with crucia affordances and constraints. Tangibe Bits [6] extends this approach by seamessy integrating these physica objects with the virtua word. However, our project scope differs from a the above systems in that none of these technoogies focus on interpersona communication or on entertainment. At the same time we were deveoping HandJive, other research projects aso began incorporating haptics into interpersona communication devices. The Bed environment [2] connects two peope remotey to aow for intimate and non-verba communication. Biow [ 131 enabes chidren isoated in hospitas to pay with one another. Both of these interfaces combine tactie input with auditory, visua, and/or ambient media. In contrast, HandJive has no audiovisua component: the interaction is competey haptic. Aso deveoped during this timeframe, intouch [] gives the iusion that two peope in separate ocations am interacting with one shared object through a haptic-ony interface. Our work differs in that HandJive aows orthogona or indirect manipuations couped with discrete articuation positions (as opposed to a continuous range of motion). We fee this distinetion eads to a significanty different design space.

2 PAPERS CHI APRIL Studying Our User Group \ EARLY DESIGN STEPS The initia stages of design can often be the most difficut We then began to study our user group so we coud understand their needs, behaviors, attitudes, and contexts. to quantify and describe. Athough our process for Both at Stanford and at Foothi Community Coege, we HandJive was iterative and recursive, we outine beow the beginning phases of our design, some of which took pace simutaneousy. attended casses with a high percentage of freshmen and sophomores. We aso observed isteners in other reated contexts, incuding a BayCHI meeting. We soon reaized Defining Our User Group: Friends isoated in sience The first step in our research and design process was to put aside a ideas about interfaces and technoogies; instead, we wanted to focus on a user group that had unmet needs. After considerabe brainstorming and ightweight investigations, we found that our team was generay interested in designing for peope who were captive in sience. In other words, we wanted to create entertaining activities for peope who were isoated in a pubic context (see Figure 1). We then began peope watching ; we observed individuas on buses and subways, at airports, in eevators, at cafes, at casses, in church services, and so on. We eventuay discarded situations invoving compete strangers, primariy because we feared that the interactions coud be creepy and/or awkward. After much observation, we decided that our key user group was friends isoated in cassrooms and meetings. For purposes of research, design, and evauation, we narrowed our user group further, focusing on freshmen and sophomores in coege casses. We accepted the specia chaenge of this user popuation: On one hand, we wanted to create a device which aowed students to interact payfuy with each other, but on the other hand we did not want to create something so immersive that it woud distract students from the cass content or interactions. This was an important objective throughout the project. However, in a broader sense, we hoped to deveop a system that woud aow peope to interact payfuy with each other in any context in which they were isoated in sience. Peope isoated in pubic that amost everyone seems to fidget in these situations. As one user ater commented, Fidgeting is a nationa pastime. Some of the behaviors we observed incuded: whispering stretching paying with hair taking notes.eating paying with jewery dooding reading ooking around seeping In addition to making observations, we questioned students about their cassroom behaviors. Through these informa interviews, we earned that some students come to cass prepared to pay games with others (e.g., bringing dice or a squirt gun). So at a ater stage of research, we provided our users with some simpe games, such as Hangman, to pay in cass. We wanted to see how different types of activities fit into the cassroom context. We discovered that interactive games which aowed students to stop and restart were most amenabe to a cassroom setting. Other games were simpy too immersive, either in pacing or in menta effort. From our observations and interviews we created six user profies of fictiona characters who represented key users. These profies were intended to run the gamut of our user popuation. Creating these profies heped us gain a cearer picture of the attitudes and behaviors exhibited by our typica users. Narrowing our Design Space Our investigations and insights into coege students in cassroom settings ed us to narrow our design space in important ways. At this point we had not decided on any device or technoogy, but we reaized that whatever system we buit for remote pay woud have to: be unobtrusive and possiby conceaabe aow stops and restarts aow remote interaction with other peope scae to various eves of cognitive oad These constraints payed a pivota roe throughout the rest of our design process. Figure 1: Our broad user group: peope isoated in pubic contexts. As we observed this user popuation, we became most interested in designing for friends isoated in sience. Scenario Buiding: Leading us to key issues Once we defined our users and narrowed our design space, we began exporing ways to fi our users needs. We created various narrative scenarios that described how our target users coud overcome their isoation in casses by using technoogy. Some of these scenarios we visuay scripted with storyboards (see Figure 2). Others we enacted verbay using improvisationa techniques. 58

3 59 ~-- _----LA 3 I.* - s. - -_I,2,... _., :;:-.-. 1, v:..,~. --- _. _.-- Cf APRIL 1998 PAPERS this user test, we got responses from a convenience sampe of peope we found in pubic paces. On ater investigations, we tested peope who were members of the target user popuation. Testing this prototype ed to key findings and design impications: Fi igure 2: Storyboards brought up key design issues. Baoon Prototype The storyboards and improvisations heped us face key design issues. For exampe, the foowing questions emerged repeatedy as we discussed our scenarios: Shoud we design a gaming device OR a communication cevice? We decided to deveop a gaming device because the task seemed to be a more constrained and chaenging probem. Aso, the gaming device pushed our scenarios in more interesting directions. Shoud our device be excusivey haptic OR muti-moda? We hypothesized that a tactie-ony device might be better suited for the cassroom context in comparison to a mutimoda system. Utimatey, we chose the excusivey haptic route because the possibiities intrigued us more. ITERATIVE DESIGNS AND USER TESTS At this point we buit physica prototypes for some of the sketches in our design notebooks, and we began to test the prototypes with users. The next sections describe the five basic iterations on our design. 1. Proof of Concept Prototype We buit the first prototype as a proof of concept for interpersona haptic interaction. The prototype was simpy two baoons fied with iquid and connected by a tube (see Figure 3). Finding Users found the simpe haptic device compeing Design impication Expore more compex haptic designs Direct hydrauic resistance Expore indirect interactions inspired users to fight for to reduce the impuse to fight contro We found that most users were intrigued by this type of interaction, athough a few aso fet the interaction was too one-dimensiona, which did not surprise us. What did surprise us, however, was that during the tests many users woud squeeze their baoons very hard to force the iquid into their partner s baoon. In other words, the direct resistance of iquid moving back and forth seemed to make peope want to fight for contro. One user commented that she wanted to see if she coud make the baoon pop on the other side. Of course, this type of reaction put considerabe stress on our prototypes (so we had to create various versions of the mode). More important, discovering users incination to fight started us thinking about indirect haptic interactions. We hypothesized that an indirect interaction-one that did not aow direct resistance-woud ikey produce a more cooperative environment than a direct interaction coud provide. (In addition, we knew it woud be difficut to buid a device that coud withstand this kind of stress.) I t * Toward the end of these user tests, we asked peope to imagine themseves with a simiar device in a cassroom setting. Many vounteered that they woud use the device to communicate. A few suggested the possibiity of paying games. We earned that athough this baoon device may not compete with video games for entertainment vaue, users thought it woud be fun in a cassroom setting. In other words, fun is a reative concept, depending heaviy on context. Even a simpe device coud be fim in a situation where peope are isoated in sience. Figure 3: This rap*l physica prototype, made from baoons and a tube, heped us investigate haptic interactions. The next step was to have users pay Three s Company, a game much ike Foow the Leader, with the baoon prototype. Users iked Three s Company, but it seemed to require too much concentration to be usefu in a cassroom setting. Students woud not be abe to pay the game and isten at the same time. Each user hods one of the baoons in hand, and squeezing the baoons moves the iquid back and forth. In -this simpe manner, users coud interact with each other. For

4 PAPERS CHI APRIL Prototypomania Apha With some confidence in our idea for an excusivey haptic device, we began to rapid prototype many possibe soutions, keeping in mind that an indirect haptic interaction woud ikey be better than a direct one. We used Pay-Doh, cay, and Styrofoam to quicky mock up potentia devices. By buiding these 3-D prototypes, we were abe to communicate our ideas better than by just using the drawings in our design notebooks. Our ideas for a handhed haptic device incuded finger rings, a joystick, spinning disks, and a squeeze ba. We expored different kinds of payfu activities that each of the various devices woud aow. And we prototyped some of these ideas with more robust materias, giving them some imited movement Figure 4 shows a variety of these prototypes: Figure 4: Some of the rapid prototypes we tested. We then tested our rapid prototypes with target users. At the beginning of each evauation, we simpy handed the devices to users to get their reactions. Later, we specificay asked them to compare and contrast different devices, hoping to gain insight into what peope did and did not ike. The tests on.these rapid prototypes provided us with many insights, incuding the foowing: handing the devices. In one extreme case, a woman used both hands to rip a neoprene cover off a device, destroying the prototype! In genera, it seemed that users took great iberties in squeezing, bending, and generay manhanding these creations. We suspect that two factors caused this reaction: (1) these were handhed devices, and (2) they ooked payfu. We aso found that some users were suspicious of our purposes. Users sometimes perceived these haptic devices to have sensua undertones, causing them to question our motives. Athough we never intended to design a sensua device, Hofmeester et. a. confirms our experiences: products which fit comfortaby in the hand as we as ones which are soft and fexibe tend to induce a sensua feeing Prototypomania Beta For our next iteration, we deveoped a range of new prototypes based on the design impications from the previous step. During this phase we specificay expored two types of motion: siding (moving in a straight path) and rocking (articuating ike a hinge). We aso compared various form factors such as rings (ike brass knuckes) and within-hand devices (ike a grip from a ski poe). Not ony did these user tests confirm our previous finding that users prefer discrete manipuations over continuous ones, but the tests aso ceary showed that peope preferred devices that fit inside the hand with parts that articuated in a rocking motion. Figure 5 shows an exampe of one such device. Mutipe Haptic Prototypes Finding Users iked devices that cicked into discrete positions Three stacked cyinders fet good in their hands Design impication Expore different types of discrete motion Expore designs with handfitting objects One important finding was that most users preferred the devices that coud be manipuated into discrete positions (i.e. parts that snapped into pace) more than the devices that had a continuous range of motion (such as a disk they coud spin). Athough we are sti not certain why users preferred discrete over continuous manipuations, we suspect it has something to do with the greater certainty and feedback a discrete movement provides. In contrast, perhaps our continuous-motion prototypes offered too many degrees of freedom, creating a ess compeing interaction. In this round of testing, as in others, we found that users exerted a surprising amount of force in manipuating and Figure 5: Users iked this haptic prototype for various reasons. This became our basic form factor for HandJive. Athough other types of form factors and articuations were possibe within our design space, because of time constraints we setted on the above prototype for further refinements and testing. We made these basic devices with superbas, two eectrica switches (rocker switches with three positions), and various kinds of putty. This kudge of materia had a strong appea for our users. In fact, reactions to this simpe toy hinted at how much aduts ike to pay, though they often hide that fact. 60

5 61 We refined this basic three-part device with further usertesting, earning that most peope iked a squishy center with a combination of one arge and one sma ba at each end. We aso found that one size did not fit a; users were sensitive to moderate variation in the size of these devices. 4. Deveoping Interactions for HandJive As we went through the previous steps, we simutaneousy expored the interactions users coud have with the various prototypes. A. Creating an indirect, orthogona intetzxtion One important chaenge was to deveop an indirect way of interacting through the devices. Our first prototype test bad shown that dir& resistance woud ikey inspire fighting for contro, so we discussed a number of aternatives. The interaction scheme that seemed most viabe was to give each user contro over a separate and orthogona pane of movement. More specificay, a user coud manipuate each ba on their device forward and backward into three discrete positions, ike articuating a hinge forward and backward. The bas on the device coud aso ean from side to side in three discrete positions; however, this side-to-side movement woud be a resut of the forward and back motion on the partner s device. Figure 6 attempts to show how this,orthogona interaction works. (Top view of Han&ive) m I User manipuates device forward and backviard Side-to-side movement is a resut of partner s movements Figure 6: A scheme for orthogona interactions. Using words and pictures to expain this scheme of orthogona and indirect interaction was a considerabe chaenge throughout the project. In contrast, whenever we had the opportunity to demon&ate the interaction with a physica prototype, peope readiy understood the interaction scheme. This difficuty iustrates how words and pictures are often insufficient to communicate ideas about movement, and it shows the vaue of quick physica prototypes, especiay for haptic-ony devices. EL Deveoping patterns and routines With our orthogona interaction scheme in pace, we then began deveoping our first iteration of a haptic anguage for HandJive. We hypothesized that HandJive woud aow a new type of interaction not found anywhere ese. The HandJive interaction coud be something ike socia dance, jazz improv, and conversation. We examined these three activities to gain some insight into how to create interactions for HandJive, noting interesting simiarities among them: sma units ordered combinations of units guiding principes for how units fit together jazz improv rifts scaes conversation words utterances grammars socia dance positions patterns routines Because these three types of interactions were buit of smaer units, we suspected that a haptic interaction anguage for HandJive shoud rey on putting smaer units together to create a arger possibiity of interactions. At first we caed these sets of movements grammars. Athough this term worked we for us, we ater reaized that grammar was not a word our users associated with fun, so we changed the terminoogy to patterns and routines. After various attempts at defining and user testing Tactiese, our haptic anguage, we eventuay setted on the foowing: Tactiese consists of singe units caed positions, simpe movements caed patterns, and more compex movements caed routines (see Figures 7 and 8). A user can move the HandJive into one of nine discrete positions. Patterns combine these positions and are the buiding bocks for the routines. For exampe, the Two- Step Pattern is embedded in the Mambo Routine beow. We hypothesized that these patterns and routines coud be the buiding bocks for both coaborative activities (much ike dances) and competitive activities (perhaps ike thumb wresting or a shared Rubik s cube). To become a competent HandJive payer, one woud need to earn a certain number of patterns and routines. Both types of interaction woud invove computer mediation. Figure 7: An exampe of a HandJive pattern which consists of two positions, repeated twice. We caed this simpe movement the Two-Step Pattern.

6 _ L....^-..,.. -,,:... _./ _c - PAPEES CHI APRIL 1998 opponent might receive a sma eectric shock. Aternativey, a user can compete in singe-payer mode against the computer: Since Sweet Spot requires significant computing and microeectronic technoogy, we coud not user test this tactie game with the mechanica prototypes we buit. Instead, we deveoped a software version with HyperCard and tested the game in a visua manner (see Figure 9). Figure 8: An exampe of a HandJive routine. We caed this series of movements the Mambo Routine. Notice that the Two-Step Pattern is embedded in steps 6 and 7. In conducting user tests, we found that peope coud easiy perform the patterns. However, these users coud not master the routines during the timeframe of a testing session, giving us itte hepfu feedback As a resut, we asked five users to practice these more compicated movements. We found that after two hours of practice, our experienced users coud perform our most compex routines enjoyed manipuating the device coud manipuate the top ba more easiy than the bottom one preferred the compex routines over the simpe ones preferred routines that offered variety: both sma and dramatic moves improvised their own routines Of course, we reaize that Tactiese is an artificia anguage. If HandJive were an actua product, users woud ikey evove their own interaction patterns. We were curious how this haptic anguage might evove but did not have the time or resources to answer this question. C. Deveoping potentia games In addition to deveoping a tactie anguage, we aso expored different games peope coud pay with HandJive that did not rey on Tactiese. One chaenge in creating a game was finding activities that were rewarding a aong the earning curve. In other words, we wanted a game that was easy to earn but sti interesting after paying it for some time. One of the games we invented was Sweet Spot, which somewhat resembes the chidhood favorite Batteship. Initiay, the computer secrety assigns each payer two sweet spots. The two payers take turns moving their own HandJive, which in turn affects the position of the opponent s HandJive. The objective is to position the top ba into one of the sweet spots and the bottom ba into the other. The device might send a victory signa to the winning payer (perhaps a vibration) whie the osing Figure 9: We tested a tactie game idea using a aptop computer with a divider that separated the screen. We found that users iked this strategy game. If HandJive became a commercia product, one coud easiy envision appication deveopers buiding game software for HandJive, much ike Sweet Spot, that users woud downoad. 5. Interactive prototype testing During this period, we aso created three interactive prototypes. These prototypes aowed the bas on the HandJive to rock side to side in the user s hand, whie the user manipuated the bas forward and backward. These devices simuated our proposed HandJive movements amost exacty and woud aow for Wizard of 02 testing. We buit our prototypes using bike cabes that ran to a contro board (see Figure 10). We tested the interactive prototype with our five experienced users. We had two key questions going into these interactive user tests: (1) Woud users be abe to sense the side-to-side position of the bas? and (2) Woud users have difficuty manipuating the bas as they moved from side to side? We took each user into a cassroom and aowed him or her to interact via the HandJive with our wizard at the contro board. These tests showed that users: found the movement in their hand compeing but difficut to describe coud sense the orthogona movement of the bas fairy we 62

7 63 CH 98 e APRIL 1998 PAPERS found that manipuating the moving bas was simper than they expected These user tests dispeed our main fears. Users coud indeed sense the position of the moving bas, with the top ba being easier to sense than the bottom. Furthermore, users coud manipuate the bas whie they were moving orthogonay. One unanswered question is whether users woud adopt the patterns and routines we created. We think not; users woud ikey invent their own anguage and games. Another question is whether Ha&ive woud be used as we have intended. We cannot be sure, though we suspect that peope woud use Handhve in unanticipated ways. Because this device does not rey on visuas or sound, it coud be a significant communication or entertainment medium for the deaf or bind. KEY LEARNINGS FROM THIS PROJECT Whie creating this new interface, we gained insights that are common to many design probems, such as the benefits of rapid prototyping, user tests, and iterative design. However, the nature of our device aso ed to ess-common insights. They are as foows: Designing devices for pay Fun is a reative term: What is fun in one context may not be fun in another. A major chaenge for a pay device is providing rewards a aong the earning curve. Meeting in ocations ike toy stores and parks inspires design. Having toys at design meetings is aso hepfu. Pay is never free-form; it reies on some form of imposed structure. We aso tested the interactive prototype on five novice users-peope who had never used HandIive before. Athough these users had not practiced any of the patterns or routines, they seemed fairy adept in manipuating the interactive device and enjoyed the moving sensation in their hands. Even novice users coud manipuate the bas forward and backward, despite the fact that the bas were simutaneousy moving side to side. FUTURE DIRECTIONS FOR DESIGN Athough HandIive presents a new type of computermediated technoogy and pushes forward the possibiities for haptic interactions, the design process is not yet compete. The future steps for deveoping and understanding this device incude: testing fuy functiona devices in a cassroom setting having users spend extensive time with the device and noting how the interactions evove exporing other user groups, such as the bind or those in corporate cubices Designing handhed devices Peope exert considerabe force on handhed objects. Fidgeting is surprisingy pervasive. Handhed objects easiy become objects of desire and can often be charged with sensua undertones. The technica aspects of handhed devices (battery ife, mechanica functioning, etc.) are chaenging. Handhed objects are not one size fits a. Designing haptic devices It is difficut to communicate haptic events in words or pictures. Rapid prototyping is more effective than anguage in conveying haptic ideas. Peope can earn haptic skis fairy quicky. A simpe haptic device can support compex interactions. Our users iked discrete manipuations over continuous ones. Haptic interactions can be compeing and powerfu, making this an exciting area for research and design. ACKNOWLEDGMENTS We woud ike to thank the foowing peope for their many insights and contributions to our project: A Cho, David Keey, Chris Rohrer, Een Tauber, and Terry Winograd. This work was supported by Interva Research and Stanford University.

8 PAPERS CHI APRIL 1998 REFERENCES U PI 131 t41 PI E [71 PI Brave, S. and Dahey, A. intouch: a medium for haptic interpersona communication. Edended Abstracts of CH1 97, Atanta, Dodge, C. The bed: a medium for intimate communication. Extended Abstracts of CHI 97, Atanta, Fitzmaurice, G., Ishii, H., Buxton, W. Bricks: aying the foundations for =Oraspabe user interfaces. Proceedings of CHI 95, Denver, Hinckey, K., Pausch, R., Gobe, J., Kasse, N. Passive rea-word interface props for neurosurgica visuaization. Proceedings of CHI 94. Boston, Hofmeester, G., Kemp, J., Bankendaa, A. Sensuaity in product design: a structured approach. Proceedings of CH1 96, Vancouver, Ishii, H. and Umer, B. Tangibe bits: towards seamess interfaces between peope, bits and atoms. Proceedings of CH1 97, Atanta, MacKenzie, I. and Oniszczak, A. The tactie touchpad. Extended Abstracts of CHI 97, Atanta, Mark, W., Randoph, S., Finch, M., Van Verth, J., Tayor, R. Adding force feedback to graphics systems: isues and soutions. Proceedings of SIGGRAPH 96, New Oreans, PI WI [111 tw [I I31 Massie, T. and Saisbury, J. The PHANTOM haptic interface: a device for probing virtua objects, Dynamic Systems and Contro, Chicago, Pausch, R., Bumette, T., Brockway, D., Weiben, M. Navigation and ocomotion in virtua words via fight into hand-hed miniatures. Proceedings of SIGGRAPH 95, LA, Ramstein, C., Arcand, J., Deveaut, M. Adaptive user interfaces with force feedback. Video Proceedings of CHI 96, Vancouver, Rosenberg, L. and Brave, S. Using force feedback to enhance human performance in graphica user interfaces. Extended Abstracts of CHI 96, Vancouver, Rueb, T., Wardzaa, J., Mistone, J. Biow: networked hospita payspace for chidren. Extended Abstracts of CHI 97, Atanta, Ruspini, D., Krasimir, K., Khatib, 0. The haptic dispay of compex graphica environments. Proceedings of SIGGRAPHP7, LA, Sachs, E., Roberts, A., Stoops, D. 3-draw: a too for designing 3D shapes. IEEE Computer Graphics and Appications, 11(6), S-26. Saisbury, K., Brock, D., Massie, T., Swarup, N., Zies, C. Haptic rendering: programming touch interaction with virtua objects. Proceedings 1995 Symposium on Interactive 30 Graphics, Monterey,