Memorix: A Tangible Memory Game usingiSIG-Blocks

Kiju LeeDepartment of Mechanical & Aerospace Engineering

Case Western Reserve UniversityCleveland, Ohio 44106

Email: kiju.lee@case.edu

Donghwa JeongDepartment of Mechanical & Aerospace Engineering

Case Western Reserve UniversityCleveland, Ohio 44106

Email: donghwa.jeong@case.edu

Abstract—This paper presents a tangible, computerized mem-ory game, called Memorix. Memorix aims to measure the player’sspatial and working memory span by displaying geometricpatterns on the tabletop platform and requiring the player tomemorize and rebuild the patterns. The tangible game technologyused for Memorix consists of a set of iSIG-Blocks with an inter-active tabletop platform and TAPware. iSIG-Blocks are sensor-embedded creation blocks with motion sensing and wirelesscommunication capabilities. The interactive tabletop platformserves as a game display while interfacing with the blocks.TAPware is a tangible programming tool for designing game itemsincorporating a computational measure of geometric complexity(Ci) that takes into account the pattern length, the number ofdistinctive colors, and the number of distinctive shapes used inthe item. This computational measure can potentially serve as aguideline for generating game items with a fine-tuned difficultylevel for each individual or group. Memorix and its associatedtechnology have been evaluated via a small-scale human subjectstudy (N = 12). Accuracy in correctness and manipulation speedmeasurements were over 98%. A high correlation between thegame performance and Ci was also found (r = 0.84, p < 0.01).Parameter sensitivity analysis revealed that the participants’failure rate was more sensitive to colors than length or geometricshape used in the game item.

I. INTRODUCTION

This paper presents a tangible memory game, called Mem-orix, using a set of interactive creation blocks, called iSIG-Blocks, and a tabletop platform (Fig 1). iSIG-Blocks and thetabletop platform serve as a tangible game interface that candetect physical manipulations applied to the blocks and assem-bly configurations among the blocks using embedded sensorswithout requiring any external sensing device. This technologyis suitable for a variety of geometric games, however this paperfocuses on Memorix to demonstrate its preliminary usability.TAPware, the game design software, enables the users to easilycustomize the game items via tangible programming. It alsoincorporates a computational measure of play difficulty thatquantifies the geometric complexity of the game item usedin Memorix. This unique feature allows the user to createthe items with a desired difficulty level and further designdynamic games where the difficulty of the game items can bedetermined by the player’s performance. The results from thepreliminary evaluation of the technology and the play difficultymeasure are also presented in this paper.

Fig. 1. Experimental setup for Memorix using iSIG-Blocks and a tabletopplatform. The player manipulates the blocks to match the geometric patterndisplayed on the tabletop platform (right); and the administrator screendisplays real-time performance and behavior data that is wirelessly transferredfrom the blocks. Display elements and the layout of the user interface can becustomized.

Related Work

Tangible gaming has been receiving growing interest as ituniquely integrates cognitive performance with motor coordi-nation skills. In addition, it often contributes to the “enjoyable-ness” of a game. To realize tangible games in a fully automatedmanner, the system must support detection of human manipula-tion applied to physical objects by using either i) active objectswith embedded sensing and communication capabilities or ii)passive objects with external sensing devices. It may use acombination of the two. Over the past decade, a variety ofactive objects for tangible game control have been developedfor a broad range of education, entertainment, and researchpurposes [1]. Several studies have used such tangible systemsfor measuring three-dimensional (3D) spatial cognitive abilitiesby observing construction patterns and performance [2], [3],[4]. For example, Learning Block is a digitally augmentedphysical block system enriched with a speaker and LED dis-play [5]. It aims to function as a playful learning interface forchildren via embedded gesture recognition. Another interestingapplication is the use of a sensor-embedded block system,called Navigation Blocks, for tangible navigation of digitalinformation through tactile manipulation and haptic feedback[6]. Tangibles is another active object system designed fortangible manipulation and exploration of digital information[7]. AudioBlocks and Block Jam integrated an augmentedsound feedback mechanism into the systems to allow usersto design musical sequences by manipulating the tangibleobjects with visual and sound feedback [8], [9]. Multi-agent

85

Fig. 2. CAD drawings of iSIG-Blocks and embedded electronic components including a wireless module, gyroscopic sensors, triaxial accelerometer, LEDdriver, spring-loaded connectors, piezzo buzzer, and ball magnets.

autonomous interactive blocks and games were developed forbehavioral training of autistic children [10].

Tangible gaming can also be realized by employing passiveobjects with external sensing devices. Most of the existingwork on object and manipulation/gesture detection for tangiblegaming has been focused on vision-based approaches [11],[12]. For example, a depth-sensing camera was used to builda height map of the objects on an interactive tabletop platformfor recognizing objects and detecting interaction between theplayer and the objects [13]. PlayAnywhere is a projection-vision system that can detect hover and touch by a humanfinger on the tabletop with a projected image [14]. Anotherinteresting system is called Touch-Space, which is a gameenvironment that combines reality with a virtual game envi-ronment based on ubiquitous, tangible, and social computing[15]. While vision-based systems allow flexibility in the gamedesign and the types of applications, most of these algorithmsare computationally expensive [14]. In addition, sensing islimited to the vision range unless additional sensing devicesare used.

Despite the great potential of tangible game technologies,developing customized games or tests designed for a spe-cific group of people is still challenging due to the gap inknowledge and skills between developers (e.g. engineers andcomputer scientists) and potential end-users (e.g. clinicians,educators, children, and parents). Potential end-users mayhave no experience in computer programming. Likewise, thesetechnologies developed by engineers may not meet the needsof the end-users. In an attempt to ease the programmingprocess, visual programming languages have been developedover the past decades. It allows users to program functions andvariables graphically instead of writing their own code [16].The advantages of visual programming include: 1) a motivatinginterface for beginners; 2) a simplified spatial representation ofcomplex inferences; and 3) time effectiveness. As an example,LEGO-Mindstorms R© is a programmable robotics toy operatedby a box programming language [17]. Scratch, developed byMIT Media Lab, is another educational programming languagewith visual interfaces for controlling images, music, and sound[18]. It is intended for 6 to 16 years, but can be used bypeople of all ages. Alice is an object-oriented educationalprogramming language implemented in Java [19]. It uses adrag and drop environment to create computer animations of3D models. Greenfoot is also an interactive Java programmingenvironment for designing 2D graphical applications, such assimulations and interactive games [20].

Fig. 3. Tabletop platform. A 3× 3 grid structure is used to display patternswith multiple geometric shapes (sub-patterns) for Memorix.

II. TECHNOLOGY DESCRIPTION

The tangible game technology used for Memorix consistsof a set of iSIG-Blocks with an interactive platform andTAPware for tangible programming of game items. TAPwareincorporates a computational measure of play complexity thatcan be used to generate game items with a desired difficulty.

A. Hardware: iSIG-Blocks and Platform

Fig. 2 shows the CAD drawings of iSIG-Blocks and theembedded electronic components. Each block is equipped withsensors for detecting assembly and manipulative motions, awireless communication module for data collection and repro-gramming, and sensory feedback mechanisms. The embeddedsensors include three gyroscopic sensors, a tri-axial accelerom-eter, and electromechanical contact sensors. Sensory feedbackto the user is enabled by reprogrammable LED patterns onthe block surfaces (visual), a piezzo buzzer (auditory), and avibration motor (tactile). Each block is 2.75 inches in lengthalong each side and weighs about half a pound. As shown inFig. 2, most of the electronic components are surface-mounted.An 8-bit microcontroller (ATmega1280) controls and executesthe entire application program and handles sensor data. Theblock is also equipped with 128 KB ISP flash memory card,4 KB EEPROM, 8 KB SRAM, and 86 general-purpose I/Olines. The I/O lines are used to collect inputs from the sensorsand execute sensory feedback functions.

Fig.3 shows the tabletop platform that can interact withiSIG-Blocks. This platform functions as a game interface bydisplaying various geometric patterns. Force sensing resistors(FSRs) are installed underneath the top surface to detect theblocks placed on it. Rare-earth magnets in the blocks and theplatform pull each other when the blocks are close to theplatform resulting in the activation of the FSRs. To detect

86

when the blocks are placed on the platform, a pre-calibratedthreshold level from the FSRs together with a neighboringdetection algorithm is implemented.

Fig. 4. An overview of the pattern programming procedure in TAPware.

B. Game Design Software: TAPware

TAPware provides a unique interface for creating cus-tomized sets of geometric patterns for tangible block games. Itcombines the pre-programmed graphical user interface (GUI)application with a tangible programming scheme. TAPwarealso enables the user to create a set of patterns with the desireddifficulty based on the number of images and complexity ofthe geometric pattern. A computational measure of geometricpattern complexity is further discussed in Section II-D. TAP-ware does not require learning specific syntax of programminglanguages, but instead allows programming through physicalmanipulations of the blocks. As illustrated in Fig. 4, the overallarchitecture of TAPware consists of four layers: 1) tangibleprogramming, 2) hardware abstraction, 3) interpretation, and4) verification. Physical manipulation of the blocks involvesrotating, shaking, and assembling. Each of these manipulationtypes can be detected and distinguished by the embeddedsensors, and therefore, can be used as a specific input commandfor game programming. Tangible manipulation protocol fordesigning a game is followed:

• Initialization: Shake any two blocks simultaneouslyfor at least 5 seconds at which point the GUI appli-cation launches TAPware.

• Selection of a game type: Shake a block continuouslyto scroll through the game types and stop shakingwhen the desired game type is selected.

• Design of geometric patterns: Design a pattern byplacing the blocks with desired geometric shapes onthe platform. Once each pattern is completed, shake ablock for 5 seconds. Then, the platform displays thepattern image imprinted from the geometric shapeson the blocks and the GUI also displays the patternwith a corresponding play complexity value. Repeatthis process until the entire set of patterns are created.To revise the pattern, place a block with a differentgeometric shape on the top of previously imprintedone.

• Finalization: Shake any two blocks simultaneously forat least 5 seconds at which point TAPware automati-cally turns off.

The hardware abstraction layer takes the sensor data receivedfrom the iSIG-Blocks and platform, and transforms it into aninterpretable form. The transformed data includes the ID of theblocks being manipulated, the type of physical manipulations,and the positions and orientations of the blocks placed onthe platform. It is then transmitted to the interpretation layer.During this time, gyroscopic sensors in stationary blocks arecalibrated in order to register a threshold value for distinguish-ing shaking motion. Interpretation of tangible programming isbased on Visual C# and OpenGL, and hardware programmingis done in Java. These programming languages are selected be-cause they provide a good balance between performance, crosslanguage compatibility, and rich graphical visualization. Ac-cording to the tangible manipulation protocols, manipulationtypes are distinguished and interpreted into script language byVisual C#. Once interpretation is completed, visual feedbackis provided through the platform for verification that the patternis properly imprinted into the algorithm. If the pattern needsto be modified, the user can redesign the pattern by placingthe blocks again.

C. Sensors and Wireless Network

Each iSIG-Block contains a tri-axial accelerometer, threegyroscopic sensors, and electromechanical contact sensors.These sensors measure angular accelerations, angular veloci-ties, and surface contacts respectively. By integrating multiplesensor data with adequate procedural algorithms, rotation,matching, and overall assembly of the blocks can be deduced.Physical measures such as angular momentum and kineticenergy can be also be estimated using the sensor data. Sinceaccelerometers measure the acceleration with respect to thegravitational force, the angles between the direction of gravityand the accelerometer are estimated by the following equations[21]: θpitch = tan−1[xacc/

√(yacc)2 + (zacc)2] and θroll =

tan−1[yacc/√

(xacc)2 + (zacc)2].

If every block continuously transmits its sensor data tothe host computer, a data traffic jam may occur. To solvethis problem, the event-driven wireless data transfer methodwas selected. The status of the blocks is set to be one of thefollowing three, in manipulation, assembled, or stationary, andonly manipulation or assembly information is collected andtransferred to a host computer for real-time monitoring anddata analysis. This also has the advantage of decreasing thepower consumption of the blocks and increasing battery life.

D. Computational Measure of Play Difficulty

TAPware incorporates a computational measure of playcomplexity that quantifies difficulty associated with a geomet-ric pattern based on the concept of discrete configurationalentropy [22]. This feature helps users create game items witha target difficulty level. Discrete entropy as a measure ofplay complexity of assembly games was presented in [1],[23]. Extending this approach, play complexity of a geometricpattern involving multiple shapes and colors is defined by, forthe ith geometric pattern,

Ci = HLi +HC

i +HSi (1)

87

Fig. 5. The administrator screen of Memorix. The graphical user interface (GUI) displays sensor data, game items, real-time assembly configurations in 3D,correctness notifications, and performance score. The layout and display elements can be customized based on needs.

where

HLi = log2B(N,nLi ); HS

i = log2O(nLi , nSi );

HCi = log2O(nLi , n

Ci ).

N is the total available locations for placing blocks, nLi is thenumber of geometric shapes in the pattern (i.e. the length ofthe pattern), nSi is the number of distinctive geometric shapes,and nCi is the number of distinctive colors used in the ith

pattern. B(n, k) is the binomial coefficient which counts thenumber of ways of choosing k unordered elements from ann-element set, computed as [24]

B(n, k) =n!

(n− k)!k!.

O(n, k) denotes the number of “onto” functions, which countsthe number of ways to distribute n labeled objects among klabeled boxes such that each box contains at least one object,given by [24]

O(n, k) = k!S(n, k) =k∑

i=0

(−1)iB(k, i)(k − i)n,

where S(n, k) is the Stirling number of the second kind.

We note that Ci defined here does not take into account thecomplexity of the global geometry of the pattern. For example,if the pattern contains three geometric shapes along a column,a row, or a diagonal, memorizing such pattern would be easierthan when three shapes are randomly located. Alternatively,complexity associated with locations can be defined by

H̃L = HL +HLP (2)

where HLP captures symmetry properties of the entire patternwhile HL only considers the number of geometric shapes inthe pattern.

III. MEMORIX

Memorix is a tangible memory game created for assessingplayer’s spatial and working memory span. Fig. 5 showsthe graphical user interface (GUI) that allows computerizedadministration of the game and real-time monitoring of theplayer’s performance. Geometric patterns are displayed on theplatform for a specified duration (e.g. 1 to 5 seconds) and aplayer is required to memorize the patterns and rebuild themby assembling iSIG-Blocks. Various types of sensory feedbackcan be provided indicating correctness at each step, such asgenerating melodies or vibrations.

Fig. 6. Launchpad screen of TAPware. When two blocks are shakensimultaneously for 5 seconds, TAPware begins and is initialized with thelaunchpad screen showing the TAG-Game selection table.

A. How to Create Memorix using TAPware

Following the tangible manipulation protocols, two blocksare randomly picked and shaken for 5 seconds to launch TAP-ware. Fig. 6 shows the initial launchpad screen of TAPware.Once this window opens, one of the blocks can be continuouslyshaken to scroll through the choices until a desired game typeis selected. The selection changes every 5 seconds of shakingmotion and is visualized by darkening the color on the image.Selection can also be done by clicking the button on the GUI.After selecting the game “Memory,” the pattern generator isinitiated as shown in Fig. 7. The pattern being designed is

88

Fig. 7. Pattern generator. Any pattern can be created by placing the blockson the platform. Designed patterns can be reset or saved. Also, saved patternscan be loaded. Users can play the assembly game or memorix game withpatterns designed by this pattern generator.

displayed on the left in real time, and the pattern numberand corresponding play complexity are displayed on the right.Previous patterns are also displayed at the bottom so that theuser can review the created items.

Pattern design is made by “stamping” the geometric shapesof iSIG-Blocks on the platform. The image on the top surfaceof the block is imprinted on the platform as demonstrated inFig. 8. Designed patterns are saved by shaking an iSIG-Blockfor 5 seconds or clicking “Save All Patterns” on the screen.TAPware also has template patterns so that a user can simplyload predefined sets of patterns for the game. This functionis enabled by clicking “Load Patterns.” “Reset All Patterns”deletes created patterns and allows the user to start over. Onceall patterns are designed, shaking two blocks simultaneouslyturns off the TAPware application. Otherwise, by clicking“Play Assembly Game” or “Play Memorix Game,” the gameusing the designed patterns can be played.

The 3 × 3 platform can generate more than 2.8 × 1017

distinct patterns. Each pattern consists of multiple geometricshapes arbitrarily placed on 9 possible locations and eachof the geometric shapes may be differently colored. For ourpreliminary evaluation, 12 geometric patterns were designedas shown in Fig. 9. Three groups of items each containingfour patterns resulting in 12 patterns total. Group I displaysuni-color and uni-shape patterns, Group II shows multi-colorand uni-shape patterns, and Group III flashes multi-color andmulti-shape patterns.

B. Play Complexity of Memorix

Fig. 10 shows the corresponding play complexities forthe designed patterns using (2). Computational steps for threeselected patterns are provided below:

• Pattern 1:

C1 = HL1 +HS

1 +HC1 = 5.17, (3)

whereHL

1 = log2B(9, 2) = 5.17;

Fig. 8. Tangible programming of geometric patterns using TAPware. Itdemonstrates the design process for a pattern (P12) used in Memorix by“stamping” the geometric shapes of the block on the platform.

Fig. 9. Geometric patterns used in Memorix. Group I patterns (P1-P4)with mono color/shape (top); Group II patterns (P5-P8) with mono color andtwo distinctive shapes (middle); and Group III patterns (P9-P12) with threedistinctive colors and two different shapes (bottom).

HS1 = log2 1 = 0; HC

1 = log2 1 = 0.

• Pattern 6:

C6 = HL6 +HC

6 +HS6 = 8.98, (4)

whereHL

6 = log2B(9, 3) = 6.39;

HS6 = log2O(3, 2) = 2.59;

HC6 = log2 1 = 0.

• Pattern 12:

C12 = HL12 +HC

12 +HS12 = 19.12, (5)

whereHL

12 = log2B(9, 5) = 6.98;

89

Fig. 10. Play complexity of patterns.

HS12 = log2O(5, 3) = 4.91;

HC12 = log2O(5, 2) = 7.23.

In the last pattern (P12), the complexity induced bymemorizing the locations of geometric shapes is computedby the entropy on the probability of choosing one from allpossible ways of picking 5 locations among 9, i.e., B(9, 5).The complexity associated with geometric shapes is computedbased on the number of different ways to populate the 5different locations with 2 distinctive geometric shapes such thateach shape is used at least once, which is counted by O(5, 2).Lastly, there are O(5, 3) ways for coloring 5 elements using 3different colors, such that each color is used at least once.

IV. PRELIMINARY EVALUATION

A small-scale human subject study (N = 12) was con-ducted for preliminary evaluation of the technology and com-putational measure of play complexity of Memorix. Thispreliminary evaluation focuses on the following: 1) technicalfeasibility of the developed technology (e.g. reliability ofthe entire system and accuracy of sensor-collected data); 2)score distribution; 3) validity of the proposed play complexitymeasure; and 4) the effect of the three pattern parameters (i.e.shapes, colors, and length of the geometric pattern) on theplayer’s performance.

A. Design and Method

University students aged between 22 and 34 years (M =27.8;SD = 3.4) were recruited and participated in this study.Among the total 12 participants, 2 were female and 10 weremale students. Due to the small sample size, we did notevaluate the effect of gender or age. The test involves 12 itemsshown in Fig. 9. Each participant was a given brief introductionof Memorix and how to play the game. Two sample itemswere introduced to the participants to demonstrate the game.The platform displays each item for two seconds and thenthe player is asked to manipulate and place the iSIG-Blocksto match the pattern as demonstrated in Fig. 11. At each

assembly step, the blocks provided sensory feedback to theplayer by blinking LEDs and vibrating when each pattern iscompleted. The Memorix game was scored based on whetherthe player memorized each item completely or not within the30 second time limit. Partial completion was not counted. Thetotal available raw score for 12 items was 98 where scores forindividual items were (2, 3, 4, 5, 4, 6, 8, 10, 8, 12, 16, 20).It was then divided by 9.8, such that the maximum availablescore is 10.

Fig. 11. Memorix: The platform displays a pattern consisting of multiplegeometric shapes for 2 seconds. When the platform turns off the LEDs, theplayer recites the pattern by choosing blocks with matching shapes and colorsand placing them (facing up) on exact locations.

B. Results

1) Technical evaluation: There were some technical prob-lems raised during the evaluation study. The pulling forcebetween the rare-earth magnets within the blocks and theplatform are stronger than desired. This occasionally resultsin a large impact force on the blocks when they are placedon the platform. This force causes a jump in noise within theblocks, which results in the LEDs on the blocks to blink andthe patterns on the block faces to get disrupted. The magnetsin the blocks and the platform may be replaced with smallerones to reduce the pulling force. Alternatively, by resending thepatterns to reset the LED patterns on the block, this problemcan be fixed.

To evaluate accuracy of the automatic assessment, wecompared the wirelessly transmitted sensor data with manuallyrecorded data. Reference data from manual recording involvescorrectness at each block placement and completion time foreach pattern. Accuracy of detecting correctness of assembledblocks was 98.2%. Accuracy in time measurement was also98.2% within a 1.5 second tolerance. These errors were alsocaused by the high pulling force. This occasionally resultedin a large impact force on the blocks when they were placedon the platform. We expect that replacing the current magnetswith smaller ones would fix most of the technical problemsidentified during this preliminary evaluation.

90

Fig. 12. Number of participants who successfully memorized each pattern.

Fig. 13. Failure rate and pattern complexity variations for patterns (P1-P12).

2) Evaluation of Ci: The average score from 12 par-ticipants was M = 6.29 (SD = 1.61; max = 7.96;min = 2.65). Fig. 12 shows the number of participants whocorrectly memorized each item. All participants successfullymemorized patterns with two geometric shapes, i.e., P1, P5and P9. However, none of them were successful in P12.The Spearman’s rank correlation test was conducted on playcomplexity of Memorix, Ci, and performance by the failurerate. A strong correlation was found between the two, such thatr = 0.8376 where p < 0.01. As shown in Fig. 13, an itemwith a higher value of Ci tended to have a higher failure ratein memorizing the pattern. Despite the small sample size, thisis an interesting and important finding that leads us to furtherevaluate the relationship between geometric complexity andindividual performance in geometric reasoning, processing,problem solving, and execution.

3) Parameter sensitivity: Parameter sensitivity analysis wasalso conducted to evaluate the effect of three parameters(shapes, colors, and length of the geometric pattern) on theplayer’s performance. High-level patterns in each parameterare selected and compared to each other. For example, P1 and

TABLE I. PARAMETER SENSITIVITY INDEX

ai Parameter Sensitivitya1 Length 0.1232a2 Shape 0.3164a3 Color 0.8333

P4 are compared to evaluate the length effect because both useidentical geometric shapes with a single color. P4 and P8 areselected for differentiating the shape effect. Also, P8 and P12are compared for the color effect. The following formula isused for incremental sensitivity of each parameter [25], [26]:

Si =∆F

∆ai(6)

where Si is sensitivity coefficient, ∆F is finite difference ofoutput and ∆ai is scaled small perturbation of input values.a1, a2, and a3 refer to the three parameters of length, shape,and color respectively. Table I shows that the failure rate ismore sensitive to colors than length or geometric shape usedin the pattern.

V. CONCLUSION AND DISCUSSION

This paper presented a tangible memory game, Memorix,and its preliminary evaluation results. The developed tech-nology consisting of a set of iSIG-Blocks with the tabletopplatform and TAPware for tangible programming serve as aplatform technology that can be used for various research,entertainment, and educational applications. TAPware alsofeatures a special embedded function that computes play com-plexity based on geometric properties of the pattern allowingthe user to create a game or test with proper difficulty fortarget players. Preliminary evaluation of Memorix focusedon technical feasibility, validity of computational complexitymeasure, and sensitivity of pattern parameters. In order forMemorix to be used as a cognitive and intellectual assessmenttool, additional reliability and validity evaluation must beconducted. To do so, our future work may involve evaluationson internal consistency, test-retest reliability, and correlationalanalysis with other existing measures as well as upgradingthe hardware and software. While the game presented in thispaper is based on 2D configurations, the system can alsosupport 3D games. In addition, development of interestingblock games with enhanced graphics (e.g. combining a virtualenvironment) and interactive rules would broaden the useof the developed technology. Since the presented technologyallows programmable sensory (i.e. vision, auditory, and tactile)feedback to the users, it can be used for studying the effect ofdifferent user feedback on performance and also for designingmore fun or engaging games using these mechanisms.

ACKNOWLEDGMENT

This work was supported by the National Science Founda-tion under Grant No. 1109270.

REFERENCES

[1] Jeong, D., Kerci, E., and Lee, K. 2010. TaG-Games: Tangible GeometricGames for Assessing Cognitive Problem-Solving Skills and Fine MotorProficiency, IEEE International Conference on Multisensor Fusion andIntegration for Intelligence Systems, 32-37.

91

[2] Buechley, L., and Eisenberg, M., Boda Blocks: A Collaborative Toolfor Exploring Tangible Three- Dimensional Cellular Automata. CSCL’07Proceedings of the 8th International Conference on Computer SupportedCollaborative Learning.

[3] Watanabe, R., Itoh, Y., Asai, M., Kitamura, Y., Kishiro, F., and Kikuchi,H. The Soul of ActiveBlock: Implementing a Flexible, Multimodal,Three-Dimensional Spatial Tangible Interface. Proceedings of the 2004ACM SIGCHI International Conference on Advances in ComputerEntertainment Technology.

[4] Anderson, D., Frankel, J.L., Mark, J., Leigh, D., Ryall, K., Sullivan,E. and Yedidia, J., Building Virtual Structures with Physical Blocks.Proceedings of the 12th Annual ACM Symposium on User InterfaceSoftware and Technology, 1999.

[5] Terrenghi, L., Kranz, M., Holleis, P., and Schmidt, A., A Block to Learn:a Tangible User Interface for the Design of a Learning Appliance. PersUbiquit Comput, 153-158.

[6] Camarata, K., Yi-Luen Do, E., Johnson, B., Gross, M.D., 2002. Nav-igational Blocks: Navigating Information Space with Tangible Media.International Conference on Intelligent User Interfaces. ACM Press, SanFrancisco, pp 13-16.

[7] Gorbet, M. G., Orth, M., and Ishii, H., 1998. Triangles: Tangible Interfacefor Manipulation and Exploration of Digital Information Topography.Proceedings of CHI’98, 18-23.

[8] Schiettecatte, B., and Vanderdonckt, J., 2008. AudioBlocks: a DistributedBlock Tangible Interface based on Interaction Range for Sound Design.Proc. of the 2nd International Conference on Tangible and EmbeddedInteraction (TEI’08)

[9] Newton-Dunn, H., Nakano, H., and Gibson, J. Block Jam: A TangibleInterface for Interactive Music. Proc. of NIME’03.

[10] Alers, S., Barakova, E.I., Multi-Agent Platform for Development ofEducational Games for Children with Autism. Games Innovations Con-ference, 2009. ICE-GIC 2009. Intl. IEEE Consumer Electronics Society.

[11] Wachs, J.P., Kolsch, M., Stern, H., and Edan, Y., Vision-based Hand-gesture Applications, Communications of the ACM, 54(2): 60-71, 2011.

[12] Kato, H., Billinghurst, M., Poupyrev, I., Imamoto, K., and Tachibana,K. Virtual Object Manipulation on a Tabletop AR Environment. IEEEand ACM International Symposium on Augmented Reality, 2010.

[13] Wilson, A.D., Depth-Sensing Video Cameras for 3D Tangible Table-top Interaction. IEEE International Workshop on Horizontal InteractiveHuman-Computer Systems, Newport, RI, October 2007.

[14] Wilson, A.D., PlayAnywhere: A Compact Interactive TabletopProjection-Vision System.

[15] Cheok, A.D., Yang, X., Ying, Z.Z., Billinghurst, M., and Hirokazu, K.(2002). Touch-Space: Mixed Reality Game Space Based on Ubiquitous,Tangible, and Social Computing. Personal and Ubiquitous Computing.6(5-6): 430-442.

[16] Garcia, M.A. and Patterson-McNeill, H., 2002. Learn How To DevelopSoftware Using The Toy Lego Mindtorms, 32nd ASEE/IEEE FrontiersIn Education Conference, 7-10.

[17] Baum, D., Definitive Guide to LEGO MINDSTORMS, Second Edition.APRESS.

[18] Resnick, M., Maloney, J., Hernandez, A. M., Rusk, N., Eastmond,E., Brennan, K., Millner, A.,Rosenbaum, E., Silver, J., Silverman, B.,and Kafai,Y. (2009). Scratch: Programming for All. Commun. ACM,52(11):60-67.

[19] Cooper, S., Dann, W., and Pausch, R. Alice: a 3-D introductoryprogramming concepts. CCSC ’00 Proceedings of the Fifth AnnualCCSC Northeastern Conference on the Journal of Computing in SmallColleges.

[20] Kolling, M. Greenfoot: a Highly Graphical IDE for Learning Object-Oriented Programming. Proceedings of the 13th Annual Conference onInnovation and Technology in Computer Science Education.

[21] Wong, W. Y. and Wong, M. S., Detecting Spinal Posture Change inSitting Positions with Tri-axial Accelerometers, Gait & Posture 27:168-171, 2008.

[22] Shannon, C.E., A Mathematical Theory of Communication, Bell SystemTechnical Journal, vol. 27, pp. 379-423, 623-656, July, October, 1948.

[23] Jeong, D, Kerci, E., and Lee, K., Sensor-Integrated Geometric Blocks:Towards Interactive Play-Based Assessment of Young Children, Interna-

tional Workshop on Interactive Systems in Healthcare (CHI-WISH 2010),Atlanta, GA, April 10-11, 2010.

[24] Erickson, M. J., Introduction to Combinatorics, Wiley, New York, 1996.[25] Hamby, D. M., A Review of Techniques for Parameter Sensitivity

Analysis of Environmental Models. Environ. Monitoring and Assessment32(2): 135-154, 1994.

[26] Downing, D.J., Gardner, R.H., and Hoffman, E.O., An Examinationof Response-Surface Methodologies for Uncertainty Analysis in Assess-ment Models, Technometrics, 27, 151-163. 1985.

92