first steps in creative computational thinking with ... · 1) promote computational thinking —...

First steps in creative computational thinking withnatural language programming and Lego

MINDSTORMS

Geoffrey FalkBirkbeck Knowledge Lab

Department of Computer Science andInformation Systems

Birkbeck University of LondonLondon WC1E 7HX UK

Keith Leonard MannockBirkbeck Knowledge Lab

Department of Computer Science andInformation Systems

Birkbeck University of LondonLondon WC1E 7HX UK

Email keithdcsbbkacuk (Contact Author)

Keywordsmdashe-Learning technologies and tools Case studiesand emerging applications e-Learning design and methodologiesAuthoring tools and Social impact and cultural issues in e-Learning

AbstractmdashWe describe the design and implementation of aprototype Android application to help promote the developmentof creative computational thinking in the UK Key Stage 1and 2 classroom The system is based on the combination ofa natural language interface and Lego MINDSTORMS robots[1] focussing on engaging children and teachers who wouldnot naturally be interested in computer programming Theprototype was tested with groups of children at local schoolsWe describe the stages of development for the application andtest results The paper concludes with a brief comparison ofother programming applications and suggests ways forward forthe further development of this work

I INTRODUCTION

The aim of this work was to develop an application thatcan aid the introduction of creative computational thinkinginto the UK Key Stage 1 and 2 classroom (KS1 and KS2)A new Computing programme of study was introduced intothe UK National Curriculum from September 2014 in broadrecognition that computational skill is now a basic skill thatis required by everyone It has meant that for the first timeprimary school teachers (5-9 years old) will be required toplan and deliver lessons that teach basic programming skills[2]

There are a growing number of products on the marketwhich can be used to help teach such skills including Bee-bots floor turtles that follow in the tradition of Logo [3]and a range of tablet applications However for the beginnerthe overriding bias is placed on developing skills rather thaninspiring creativity (see Section V) and this may not beenough to engage the imaginations of teachers and pupils whoare less enthused by the process of programming itself than bythe creative things that can be done with it This applicationaims to help redress that balance

In the next section we outline the basic approach andobjectives of our work In Section III we describe how we

determined the requirements for the work In the followingsection (Section IV) we outline the basic design and imple-mentation of the prototype system including the integrationof the Lego robot and Android application In Section V wereview and analyse the results of our experiments with theprototype application Then in Section VI we provide a briefcritical comparison with two other products targeting this agegroup Finally in Section VII we provide some conclusionson the work and thoughts on future work

II APPROACH

Our system is inspired by the work of Seymour Papert andothers at MIT Media Lab who since the 1980s have repeatedlyshown how computational thinking can be encouraged inchildren A recurring approach has been to connect abstractprogramming concepts with the physical world via Logoturtles in the 1980s physical micro-worlds and latterly LegoMINDSTORMS robots [5] [6] [7] As a consequence manyschools and after-school clubs have starting running roboticscourses and there are competitive leagues to motivate chil-dren to become better programmers (eg ROBOCUP JUNIOR[8]) However these advances are taking place against thebackground of declining numbers of children studying STEMsubjects at higher levels [9]

There is a concern however that taking a isolated skills-based approach to teaching computing even if it uses roboticsmay not have the desired effect for some students [10] Rusket al proposed that for many learners instead of focussing ondesign challenges like building cars or moving robots throughobstacle courses structuring workshops around shared themesand storytelling would be more effective at engaging a widerdiversity of students from early childhood onwards Their workwith the programming tool SCRATCH [12] has shown this tobe true at least for older learners Scratch allows users to createstories art and games via a block programming interface andnow hosts several million projects [13]

Scratch has been a major success however according tostatistics from MITs site and shown in Figure 1 the vast

10 Intl Conf e-Learning e-Bus EIS and e-Gov | EEE18 |

ISBN 1-60132-474-X CSREA Press copy

majority of users are of middle school age and above peakingat age thirteen [14] No programming environment has yetappeared for younger children that can inspire the creativityand dedication of Scratch (even considering the popularity ofMindCraft [11])

Fig 1 Age distribution of Scratch users

Young pupils have been found to engage easily with robotspossibly because they are comfortable with playing withLego bricks and doing so reminds them of toys rather thanassignments [19] They are also very powerful with complexbehaviours made available through coordinated use of themotors and sensors In general young children feel verycomfortable using mobile handsets with a huge variety ofgames and educational applications available on Google Play(and the Apple App Store) for four year olds and upwardsThe technology for communication between Android devicesand the Lego MINDSTORMS robot via Bluetooth is wellestablished and has been integrated into a diverse set ofproducts and platforms

There were four main objectives for this work1) Promote computational thinking mdash targeted at the cur-

rent UK KS1 and KS2 computing curriculum2) Possess a fast learning curve mdash the application should

provide an interface that young students (and teachers)can pick up and use instantly without the need for askills-based tutoring system

3) Appeal to the dramatists mdash [20] discusses how youngchildren start out by being ldquoboth little storytellers andlittle engineersrdquo but a division happens later on inschooling as a some children lose confidence in newabstractions that they cannot tie to concrete experiencethe aim was to engage both the dramatists and the so-called patterners

4) To provide an extensible architecture which can easilybe adapted to a changing Internet-of-Things world

The approach has remained closely aligned to using story-telling as a method for engaging children in programmingby integrating an Android based application with the LegoMINDSTORMS product utilising a novel natural languagestyle interface

III REQUIREMENTS GATHERING

Three methods were used to gather requirements for theproject mdash- Interviews Document Analysis and Prototypedevelopment and testing in this paper we will focus on theinterviews with the children

Thirty children were interviewed and observed over a twoweek period using various pieces of software for both de-veloping general programming skills and for controlling aMINDSTORMS robot which offer basic remote control overthe robots motors We now consider three of those childrenwho had typical responses from the group

Bruce (aged 9) performs near the top in his class in Englishand Maths and is quick to master computer interfaces He hasbeen given a Wordpress site which he sporadically updates itwith performances and puppet shows He plays the piano andshows aptitude in all subjects but he is especially keen oncreating music and writing stories He could be classified asa ldquodramatistrdquo

He quickly lost interest in the various remote controllerapplications available for MINDSTORMS He showed interestin writing stories for the robot to act out but when he wasshown the available tools he wasnrsquot sure how to use themand what they did When told that they were to help him tella story with the robot he couldnt say how he would achievethis When asked about how he would like to control the robothe mentioned that hed like to create an adventure story withthe robot

Leo (6) is beginning to write but struggles in school withMaths in particular His interests are street dancing art andfootball He is not condent using the computer and generallyplays games on CBBC and CBeebies and getting print outswhich he then colours in

He enjoyed using the basic applications and was skilled atmoving the robot around the room When asked what he wouldlike to get the robot to do he mentioned ldquochasing peoplerdquo andldquoyou tell it do something and it does itrdquo

Lilly (5) got on really well with the basic MINDSTORMSapplications Although her reading level wasnt at a level whereshe was condent to read the words in the ldquobasicrdquo interfaceWhen it was explained what she would have to do she saidshe would like to tell a story about a fairy robot who ldquodoesmagicrdquo but couldnt explain what she would have to do toachieve this with the interfaces of the standard applications

Requirements gathered from this stage therefore reinforcedthe research that storytelling could be an good way to engagethe ldquodramatistsrdquo and that it was important for the applicationto allow for complex interactions making use of both motorsand sensors Enough details had now been gathered to definethe final requirements and the following subsections discussthese

A User Requirements

Users should be able to write and run programs using com-plete English sentences via an NLP interface The interfaceshould be straightforward enough for a literate 6 year old touse and allow users to do the following

Intl Conf e-Learning e-Bus EIS and e-Gov | EEE18 | 11


bull write a program or ldquostoryrdquo in full English sentences onan Android device

bull adapt a ldquostoryrdquo written by another userbull save their ldquostoryrdquo for later usebull run and test their ldquostoryrdquo on a connected MINDSTORMS

robot

B Technical requirements

The main technical requirements werebull to implement an Android application that remotely con-

trols an MINDSTORMS robotbull to coordinate the robots actions with Android sounds and

musicbull to map this coordination to the sequence of instructions

dened by the userbull to translate the instructions from NLP sentences as dened

by dened RLIT ontology (see later)bull to upload sounds to the MINDSTORMS robotbull to read information from the following robot sensors

ultrasonic sound light switchbull to read the ldquotacho countrsquo (motor rotations) from each

robot motor portbull to direct the robot to move in a direction rotate play

sounds and beepbull to work on both handset and tablet size screens

C NLP specication mdash RLIT

Part of the specification stage was to define in detail theontological rules governing the NLP language RLIT

bull The language needed to provide a range of sentenceconstructions that would allow the user to control aMINDSTORMS robot via Bluetooth including settingmotor actions reading sensors and playing sounds andnotes

bull The language would also need to allow users to coordi-nate with this music tracks and spoken dialogue from theAndroid device

bull The language would not need to provide users with thefine grained control over the motors and sensors thatfor example the Lego software does It is sufficient tohave a sentence like ldquoROBOT moves forward a little andsteadilyrdquo

bull The language should allow for extensions to allow forcontrolling variables such as speed distance degrees ofturn and sensor thresholds

bull The language would need to include the ability for loop-ing statements and event handling This was essential toaddress the programming objectives in the Key Stage 1and 2 Computing curriculum

bull The language would need to support concurrency so thatmore than one event could happen at the same time forexample moving the robot around whilst playing musicfrom the Android device This would bring a freedomto the users programs and also allow for interestingsituations that would require the user to debug their

ldquoprogramsrdquo Again this is a stated objective in the KeyStage 2 curriculum

IV DESIGN AND IMPLEMENTATION

RL is designed as a multi-tier application based on An-droids MVC (Model View Controller) architecture using theKotlin programming language and its coroutines functionalityfor the concurrency aspects (Figure 2) Kotlin was preferredover Java as it has a more concise syntax with higherfunctionality and is a preferred language for the Androidplatform

Fig 2 Overall architecture of ROBOLITERATE

A UI and UI Controller Layers

The UI layer is composed solely of XML and resourcefiles which are accessed and inflated by the Activity andFragment classes in the UI Controller layer The controllerlayer also provides access to the RLIT model layer whichcontains classes implementing the ontology of the RLITnatural programming language The controller layer connectswith a storage data layer that is composed of an SQL databaseaccessed via the EXPOSED API [15] allowing users to saveand load stories and sounds

After the controller layer connects via Bluetooth with therobot device all subsequent communication with the robot ismanaged by a background service which the controller layerbinds to The controller sends via Messenger objects allthe information required for the service to communicate withthe robot including configuration choices and RLIT programinstructions Robot and other program events are then relayedback to the UI controller layer in a similar manner

This separation of concerns provides maximum flexibilityfor expansion and adaptation in the future By separating outthe UI layer from the controller layer any number of differentpermutations of design can be used without affecting the restof the application Similarly the RLIT model can be usedindependently in a different application environment

B RLIT Layer

Like other NLPs RLITs foundation unit is the ldquosentencerdquoEach sentence represents a distinct instruction for controllingeither the robot or Android device The phrases within eachsentence modulate how and when these instructions are per-formed In turn a set of sentences make up a ldquostoryrdquo Theldquostoryrdquo defines how the sentences are executed relative to each



other Once the story is written an interpreter analyses it andconverts it into a ldquoprogramrdquo that can be subsequently executedFigure 3 illustrates the mapping between RLIT componentsand Robot program components Unlike a fully developed

Fig 3 Mapping of RLIT elements and program instructions

NLP the user is guided through writing an RLIT sentencestep by step This ensures that the process is as simple aspossible especially for Key Stage 1 pupils

Phrases are organised into hierarchical structures definingtheir exact ordering in sentences Figure 4 shows a fragmentof this structure

Fig 4 Fragment of RLITs hierarchical design

Two possible routes for RLIT phrase combinations are shownin the figure one for combinations that instruct the robot tomove and the other to instruct the robot to wait for a targetreading from the ultrasonic sensor Each group of phrases

has an ID and their ldquochildrenrdquo are represented by a directedgraph For example if the phrases First then ROBOT areselected all phrases that have the parent ID of ROBOT (ID 10)are displayed to the user If the phrase moves forward isselected then all phrases with parent ID 105 are shown Thisprocess continues until all possible phrases are exhausted

RLIT was developed according to an ontological frameworkwhich provides sufficient power for users to program the robotin ways that can only otherwise be achieved using Legos nativeAPI However as indicated earlier fine-grained control wasdeliberately omitted to provide a soft storytelling environmentfor users rather than a mathematically exact one However thisfunctionality can be added at a later date (see Section VII)

The role of the interpreter in ROBOLITERATE is to takethe stories constructed in RLIT and convert them to sets ofinstructions that Android device and robot can execute Theinterpreter ascertains which instruction type to create based onthe InstructionID provided by the verb phrase in eachsentence Each instruction is represented by a unique ID andseveral RLIT verb phrases can share the same instruction IDif they describe variations on the same instructions

For example the RLIT verb phrases move backmove forward and stop moving all share the instruc-tion ID INSTRUCTION_ROBOT_MOVE The verb phrasesturn left and turn right share the instruction IDINSTRUCTION_ROBOT_TURN On the other hand theresonly one way to beep so the verb phrase BEEP isunique in being the only verb phrase to have the IDINSTRUCTION_ROBOT_BEEP

Once the instruction type is extracted it is instantiated bythe interpreter with parameters that are dened by an analysisof the eld values of all of the phrases in the sentence Anyparticular phrase eld value means something unique dependingon the instruction to which it is attached Once the instructionsare built they are saved as a ldquoProgramrdquo mdash a simple list ofinstructions that in turn contain all the data necessary for therobot control classes to use during program execution

The design had to allow for the later substitution of newmodels of MINDSTORMS robots in place of the current EV3device In addition the design needed to allow for futureversions where multiple devices can be controlled or wheremultiple Android devices could work together For this reasonAndroids Bound Service architecture was chosen since itallows for multiple applications to connect to the same service[] Using this architecture also allows for the loose couplingrequired to allow new robot models be substituted withoutaffecting the other components of the architecture

Five different families of robot commands are provided1) direct commands that control movement and sound 2)system commands to upload files 3) ldquoreplyrdquo commands thatobtain readings from the motor ports 4) commands that obtainreadings from passive sensors (eg switch) and active sensors(eg sound) and 5) specialist commands that send and receiveinformation from digital sensors (eg the ultrasonic sensor)

The software was implemented and tested in four stages



bull Phase 1 Implementation of the UI Layer and RLITmodel

bull Phase 2 Implementation of the robot service layerbull Phase 3 Integration of the UI Layer RLIT model and

Service layerbull Phase 4 Refinement of the UI layer and RLIT model

based on feedback from user testingThe software was tested iteratively throughout its develop-ment starting at the prototype stage and progressing throughtest case scenarios to integration testing system testing andfinally user evaluation

V RESULTS AND ANALYSIS

Thirty children aged between 6 and 10 were selected by theSchool Principals from three local schools on the basis of theirrange of skills The children were also identified according totheir interest in English and art

Five thirty minute sessions were conducted each withseveral students and a teacher observer The teacher observersincluded the Principal of the school an IT coordinator andan Art teacher There were two testers present at each sessionone who ran the session and the other recording the eventswith video and still cameras

The structure of the sessions were as follows1) The students were shown a short demonstration story

with the robot The story was about a lonely robot whoturns around and moves away when asked by Androidto play Then Android offers to play some happy musicto cheer the robot up When the music starts the robotstarts spinning around and beeping

2) The children were then told that it was their opportunityto play with the robot and Android They were intro-duced to the application ROBOLITERATE however theywere given no instructions on how to use it Insteadthey were told they could ask for help if they got stuckThe children worked in pairs using a variety of Androidbased tablets and handsets

3) The children were observed using the application forten minutes When they were ready to test what theyhad written they were guided through the steps ofconnecting to the robot

4) They were then observed as the program executed andasked questions such as ldquoWhat is happening nowrdquoldquoIs it what you expectedrdquo ldquoWhy is this happeningrdquoldquoWhat went wrongrdquo and ldquoHow would you change itrdquo

5) Each pair was then given another chance to go backto their story make changes and run the program onceagain If time permitted then they carried out a thirdcycle

6) As a plenary the children discussed what they liked anddidnrsquot like about the application and how they wouldlike to change it

7) Observations were also recorded from the Principal ITcoordinator and Art teacher

It was interesting to note that it was a younger group ofchildren (the 8 year olds) who showed the most fearlessness

Fig 5 Running the robot program

and joy in using the application Their shouts of delight alsonoted by the Principal showed that they were experiencingreal excitement at getting their robot to perform

The older children were more hesitant maybe because theywere more wary of getting things wrong however thoughtheir approach was more tentative by the end of the halfhour some were beginning to think more about the storiesthey wanted to write There was some disappointment thatthe robot couldnt do more things and the younger childrencouldnt quite understand why it couldnt do things like jumpor do exactly what they told it however this stimulated aninteresting plenary discussion about what it means to programa robot

The use of audio added a very important enriching layerto the experience enabling the children to personalise theirstory It was unfortunate that the EV3s speaker was so quietmeaning that unless there was complete quiet no one couldhear it

From these tests there are promising signs that this ap-proach could provide a method for introducing computationalthinking into the classroom in an entertaining and engagingway for children most interested in performance and tellingstories Evidence of learning was found by the testers andreinforced by the teacher observers that matches many of thestated objectives of the Computing programme of study forKey Stage 1 and 2

It had already been found during earlier user testing thatthe application can be used to teach Key Stage 2 objectivesin that they were engaged in

ldquodesigning writing and debugging programs to ac-complish specific goals with physical systemsrdquo

(KS2 objective)

However the critical test was to see if computational thinkingcould happen in a classroom environment where children arenot being led ldquorecipe-stylerdquo through scripted learning pointsbut through exploration based on their own imaginations andcreativity



Initial results based on this user evaluation show that thisapplications approach is a method that can be used to facilitatethis serving as a basic introduction to creative computationalthinking There was no explicit goal given to the evaluationgroups at the beginning of the session after they were shownan example program they were simply encouraged to have funand explore the application themselves It was emphasised tothe children that there was going to be no right or wronganswer By the end of the thirty minutes many children werealready engaged in aspects of computational thinking and self-setting goals for the next story they wanted to write

Some of the observed learning includedbull learning that programs execute by following precise and

unambiguous instructionsbull practice creating and debugging their own programsbull using logical reasoning to predict the behaviour of their

programs andbull investigating why their program was not behaving as

expectedThese are stated objectives of the new Key Stage 1 curriculum

VI CRITICAL COMPARISON

Having conducted the user evaluation we wanted to com-pare these findings with results from testing other applicationsthat teach computational thinking to Key Stage 1 and 2children In particular we wanted to focus on engagementin learning and whether ROBOLITERATE can stimulate thedramatists to learn more than other applications can Giventhe lack of space we will only discuss the reaction of oneof the children although we examined the whole cohort Weconsider ldquoEngagementldquo and the extent to which the applicationpromotes ldquoComputational Thinkingrdquo

We chose two of the most visible applications for the iPadmdash ldquoHopscotchrdquo [16] and ldquoMove the Turtlerdquo [4] which areboth aimed at children aged 8-11 years old Hopscotch isheavily inspired by MITs Scratch allowing children to buildprogram routines that control cartoon characters by draggingblocks onto the screen ldquoMove the Turtlerdquo is similarly aimed at8-11 year olds and draws its inspiration from Logo enablingchildren to program a turtle to move and draw patterns on thescreen Screenshots of these applications are shown in Figure6

Fig 6 Two current iPad applications for teaching programming

A Engagement

Bruce would probably classify himself as a dramatist asnoted before in that he loves music and acting However he

is also very quick on the computer and iPad and was eagerto test the programming apps we downloaded for him

Both Hopscotch and ldquoMove the Turtlerdquo engaged Brucersquosinterest for a while and he returned to them a few times overa two week period It appeared that both applications haddifferent incentives for him to return however neither werestrong enough for him to want to keep on returning to for longcompared to say Garage Band [17] which he was constantlyrevisiting making and tweaking his musical compositions

ldquoMove the Turtlerdquo has a very strong tutoring system whichleads children step by step from programming the turtle tomove to navigating a maze to drawing fractals Bruce enjoyedthe reward system and was proud that he had already finishedthe first eight mini-levels and had reached Level 2 It appearedthat achieving a score was the main motivating factor for thisapplication and he wasnt so interested in the movements ofthe turtle because having got to Level 2 he didnt continue

ldquoHopscotchrdquo was more interesting for him mdash after a weekhe had managed to move three ldquocharactersrsquo across the screenpainting wavy trails behind them and was keen to show uswhat he had achieved By the next week however he hadnot picked up the application again With ldquoHopscotchrdquo hewas pleased with the results of his artwork achieved throughprogramming but there was a ceiling of interest that wasreached quite quickly before he moved on The issue withldquoMove the Turtlerdquo was that the rewards were insufficient tohold his interest This is an issue with many applications onthe market which focus on narrow skills-based programmingtasks but do not do enough to stimulate the minds of users andempower them to use programming for their own purposes

Bruce found that ROBOLITERATE engaged his interest forlonger It gives much more free reign to be creative to programmovements sounds and dialogues and to do so in a tangiblephysical space that gives the feel of a real performanceClearly more work will have to be done to see whetheran approach like ROBOLITERATE can in general engagecreative minds longer than the other applications but the earlyindications are positive

B Promoting computational thinking

In additional to testing engagement the applications wereanalysed together with ROBOLITERATE to ascertain the cov-erage of the key concepts and practices for computationalthinking [18] Table I summarises the concepts covered byeach application and displays an even coverage between theapplications

It is hard for an application alone to encourage com-putational thinking practices These practices are 1) beingincremental and iterative 2) testing and debugging 3) reusingand remixing and 4) abstracting and modularising It is theenvironment that is built around a tool and the context withinwhich it is used that are at least as important in influencingthe development of these practices One of Scratchs greatstrengths is its online community that encourages the reusingand remixing of other members projects



TABLE I Coverage of computational thinking concepts

The products in this study dont approach Scratch in termsof being able to encourage these practices however they all gosome way to achieving it A great strength of ROBOLITERATEis how its design encourages debugging which was witnessedduring the user evaluation It explicitly shows and updates thestatus of a running program in its console window similar toa debugging window in an IDE Many children used this todiscuss what was happening with their program and why itwasnt behaving as expected

VII CONCLUSIONS AND FUTURE WORK

ROBOLITERATE introduces a different way to approachprogramming and computational thinking for Key Stage 1and 2 children It has the potential to open up programmingto children and teachers who feel more comfortable withwords and stories than they do with patterns and symbolsIt could provide a means for children to explore theme-basedprogramming that is cross-curricular and not tied to a specificskills-led agenda At the same time it has been shown in testingto be rigorous enough that in a short thirty minute sessionmany aspects of computational thinking are touched uponcovering several aspects of the English Computing curriculum

In general ROBOLITERATE has met our original objectivesbull Promoting computational thinking ROBOLITERATE

gives students experience in sequencing looping eventsand other concepts and introduces them to the practiceof creating and debugging simple programs and sharingwork with others

bull Possess a fast learning curve in testing all children couldmaster the interface quickly and some of these childrenhad English as a second language However more workwill need to be done to introduce children to the rangeand depth of what they can create and not just how

bull Appeal to the dramatists from the evaluation therewas much excitement and interest with children alreadyplanning what stories they wanted to tell More workwill need to be done to help dramatists tell their storieshowever with further examples for them to work withand richer functionality and behaviours for the robot

bull Be future ready the aim of the system design was toensure that the application could be adapted easily forlater iterations of MINDSTORMS Only time will tell ifthis has been the case

Further work has been identified which will improve ROBO-LITERATErsquos stability educational aspects user-friendlinessand enhance the rewards for the dramatists Crucial improve-ments include the following

bull Re-introducing ldquoproceduresrdquo but designed to t with theRLIT interface This will allow for complex behavioursto be used right from the beginning and later adaptedand rened by users

bull Extend the language and interface to allow for controlover variables and data

bull Add adaptive feedback layers that not only help userswith the interface but also give guidance on how theexplore the full range of actions available

The approach gained lots of interest from students and teachersalike with one Principal sufficiently impressed to want to startinvesting in MINDSTORMS for his school

REFERENCES

[1] Sherrard Ann Rhodes Amy (October 2014) ldquoComparison of the LEGOMindstorms NXT and EV3 Robotics Education Platformsrdquo Journal ofExtension 52 (5)

[2] Department for Education (2013) ldquoCurriculum in England ComputingProgrammes of Study Retrieved from httpsbitly2GEA8Y0

[3] Terrapin Software (2013) Bee-Bot and Bee-Bot Support MaterialsRetrieved from httpwwwterrapinlogocombee-botphp

[4] Move the Turtle (2013) ldquoMove the Turtlerdquo Retrieved from httpmovetheturtlecom

[5] Alimisis D amp Moro M (2007) ldquoRobotics amp constructivism in educa-tion the TERECoP projectrdquo

[6] McNerney Timothy S (2004) ldquoFrom turtles to Tangible ProgrammingBricks explorations in physical language designrdquo Personal and Ubiq-uitous Computing 8 326-337

[7] Papert S (1980) ldquoMindstorms children computers and powerfulideasrdquo NY Basic Books

[8] Eguchi A Hughes N Stocker M Shen J Chikuma N (2012)ldquoRoboCupJunior mdash A Decade Laterrdquo In T R e al (Ed) RoboCup2011 (pp 63-77) LNCS 7416

[9] Marder M (2016) ldquoIs STEM Education in Permanent Crisis PerpetualSTEM teacher shortages hold our country backrdquo Education Week httpsbitly2E6Mwuy

[10] Rusk N Resnick M Berg R amp Pezalla-Granlund M (2008) ldquoNewPathways into Robotics Strategies for Broadening Participationrdquo Jour-nal of Science Education and Technology(17) 59-69

[11] httpwwwmindcraftnet[12] Scratch Retrieved from httpscratchmitedu[13] httpswwwcodecluborgukprojects[14] Scratch - Explore up-to-date statistics about the Scratch Online Com-

munity Retrieved from httpsscratchmitedustatistics[15] httpsgithubcomJetBrainsExposed[16] Hopscotch (2013) Hopscotch Technologies httpswwwgethopscotch

com[17] httpswwwapplecomukmacgarageband[18] Brennan K amp Resnick M (2012) ldquoNew frameworks for studying and

assessing the development of computational thinkingrdquo Paper presentedat the American Educational Research Association (AERA) annualconference

[19] Klassner F amp Anderson S D (2003) ldquoLEGO MindStorms Not Justfor K-12 Anymorerdquo IEEE Robotics amp Automation Magazine

[20] Bers M (2008) ldquoEngineers and storytellers Using robotic manipulativeto develop technological fluency in early childhoodrdquo In O S B Spodek(Ed) Contemporary Perspectives on Science and Technology in EarlyChildhood Education Information Age Publishing



majority of users are of middle school age and above peakingat age thirteen [14] No programming environment has yetappeared for younger children that can inspire the creativityand dedication of Scratch (even considering the popularity ofMindCraft [11])

Fig 1 Age distribution of Scratch users

Young pupils have been found to engage easily with robotspossibly because they are comfortable with playing withLego bricks and doing so reminds them of toys rather thanassignments [19] They are also very powerful with complexbehaviours made available through coordinated use of themotors and sensors In general young children feel verycomfortable using mobile handsets with a huge variety ofgames and educational applications available on Google Play(and the Apple App Store) for four year olds and upwardsThe technology for communication between Android devicesand the Lego MINDSTORMS robot via Bluetooth is wellestablished and has been integrated into a diverse set ofproducts and platforms

There were four main objectives for this work1) Promote computational thinking mdash targeted at the cur-

rent UK KS1 and KS2 computing curriculum2) Possess a fast learning curve mdash the application should

provide an interface that young students (and teachers)can pick up and use instantly without the need for askills-based tutoring system

3) Appeal to the dramatists mdash [20] discusses how youngchildren start out by being ldquoboth little storytellers andlittle engineersrdquo but a division happens later on inschooling as a some children lose confidence in newabstractions that they cannot tie to concrete experiencethe aim was to engage both the dramatists and the so-called patterners

4) To provide an extensible architecture which can easilybe adapted to a changing Internet-of-Things world

The approach has remained closely aligned to using story-telling as a method for engaging children in programmingby integrating an Android based application with the LegoMINDSTORMS product utilising a novel natural languagestyle interface

III REQUIREMENTS GATHERING

Three methods were used to gather requirements for theproject mdash- Interviews Document Analysis and Prototypedevelopment and testing in this paper we will focus on theinterviews with the children

Thirty children were interviewed and observed over a twoweek period using various pieces of software for both de-veloping general programming skills and for controlling aMINDSTORMS robot which offer basic remote control overthe robots motors We now consider three of those childrenwho had typical responses from the group

Bruce (aged 9) performs near the top in his class in Englishand Maths and is quick to master computer interfaces He hasbeen given a Wordpress site which he sporadically updates itwith performances and puppet shows He plays the piano andshows aptitude in all subjects but he is especially keen oncreating music and writing stories He could be classified asa ldquodramatistrdquo

He quickly lost interest in the various remote controllerapplications available for MINDSTORMS He showed interestin writing stories for the robot to act out but when he wasshown the available tools he wasnrsquot sure how to use themand what they did When told that they were to help him tella story with the robot he couldnt say how he would achievethis When asked about how he would like to control the robothe mentioned that hed like to create an adventure story withthe robot

Leo (6) is beginning to write but struggles in school withMaths in particular His interests are street dancing art andfootball He is not condent using the computer and generallyplays games on CBBC and CBeebies and getting print outswhich he then colours in

He enjoyed using the basic applications and was skilled atmoving the robot around the room When asked what he wouldlike to get the robot to do he mentioned ldquochasing peoplerdquo andldquoyou tell it do something and it does itrdquo

Lilly (5) got on really well with the basic MINDSTORMSapplications Although her reading level wasnt at a level whereshe was condent to read the words in the ldquobasicrdquo interfaceWhen it was explained what she would have to do she saidshe would like to tell a story about a fairy robot who ldquodoesmagicrdquo but couldnt explain what she would have to do toachieve this with the interfaces of the standard applications

Requirements gathered from this stage therefore reinforcedthe research that storytelling could be an good way to engagethe ldquodramatistsrdquo and that it was important for the applicationto allow for complex interactions making use of both motorsand sensors Enough details had now been gathered to definethe final requirements and the following subsections discussthese

A User Requirements

Users should be able to write and run programs using com-plete English sentences via an NLP interface The interfaceshould be straightforward enough for a literate 6 year old touse and allow users to do the following





robot


























B RLIT Layer











































(KS2 objective)













A Engagement


























REFERENCES





















robot


























B RLIT Layer











































(KS2 objective)













A Engagement


























REFERENCES


























































(KS2 objective)













A Engagement


























REFERENCES




























A Engagement


























REFERENCES

































REFERENCES



















first steps in creative computational thinking with ... · 1) promote computational thinking —...

Documents