Four illustrated storybooks, one AR framework. Printed pages come alive as interactive 3D lessons in Sinhala and English, fully offline, on ordinary Android phones.
Measuring jars filling, seeds germinating, a circuit closing, a whole split into fractions: concepts that move can't be learned from a still picture. Each component has its own illustrated storybook, and every AR scene is entered through a printed story page.
Capacity, mass and time you can pour, weigh and read
Learners pour liquid into a measuring jar, place weights on a scale and read a clock face, turning instrument reading into a hands-on activity anchored to the storybook page.
A Measuring Day at Grandma's
ExploreLife cycles that grow and change in front of you
Learners trigger and order staged biological processes, such as the plant and butterfly life cycles, so a change that takes weeks plays out as an ordered sequence in 3D.
Wonders of the Living World
ExploreMaking the invisible visible: forces, circuits and simple machines
Learners apply force vectors, build simple series circuits and manipulate simple machines, with 3D simulations that respond in real time.
Dasun and the Moonstone
ExploreFractions, decimals, parity and place value built from 3D blocks
Learners group and recombine 3D unit objects to construct fractions, decimals, parity and place-value quantities.
Dinal and the Prism Temple
ExploreWhat the literature says, where it falls short, and how this project addresses it.
A review of IEEE Xplore, Scopus, ScienceDirect and Google Scholar covering 2012–2025.
AR in primary education
Augmented reality consistently improves engagement and conceptual understanding in primary STEM, especially for content that is spatial, dynamic or invisible, such as forces, life cycles and measurement.
Narrative learning
AR storybooks such as Metabook show that a continuous story sustains young learners' attention and gives each activity a purpose, but existing titles are generic and not curriculum-aligned.
Sri Lankan context
Sri Lanka has strong access to basic education but is at an emerging stage of classroom technology. Instruction relies on printed textbooks, and manipulatives are unevenly distributed.
| System | Domain(s) | Bilingual | Multi-domain | Offline |
|---|---|---|---|---|
| Wangid et al. (2020) | Mathematics | No | No | N/R |
| Metabook (2024) | Generic storybooks | No | No | No |
| Volioti et al. (2022) | Physics | No | No | N/R |
| Cai et al. (2019) | Mathematics | No | No | N/R |
| Cai et al. (2022) | Science (BCI) | No | No | N/R |
| This work | All four domains | Yes | Yes | Yes |
No reviewed system combines all of the properties a Sri Lankan primary classroom needs.
Monolingual
AR learning tools are overwhelmingly English-only. None support Sinhala-medium learners.
Single subject
Each tool covers one subject, so every subject brings a new app and a new interaction model.
Connectivity
Offline operation is rarely reported, yet many rural schools have unreliable internet.
Curriculum fit
Systems are built for curricula outside South Asia and do not map to the Sri Lankan Grade 4–5 syllabus.
The gap: a bilingual, offline, multi-domain AR framework, built around curriculum-aligned narrative storybooks and evaluated in a real Sri Lankan classroom.
How can a narrative-based, bilingual AR storybook framework help Sri Lankan Grade 4–5 students understand STEAM concepts that static textbook illustrations cannot show?
The Challenge
Grade 4–5 students struggle with topics that need visual and process-based reasoning, from measurement instruments and life cycles to circuits, fractions and place value.
The Context
These concepts unfold in three dimensions and over time. Reduced to a printed illustration, learners have no way to observe the process itself.
Invisible
processes
current flow, forces and growth can't be seen on a page
Abstract
symbols
fractions and place value have no physical form in print
Dynamic
quantities
capacity, mass and time change while you measure them
Main objective
Design, build and evaluate a bilingual, offline, narrative-based AR storybook framework for Sri Lankan Grade 4–5 STEAM education.
Specific objectives
RQ1 · Learning outcome
What learning gains follow narrative-based AR storybook instruction, and is post-instruction performance higher than after textbook instruction?
RQ2 · Usability & engagement
How do learners and teachers experience usability, engagement and feedback quality, and what do the interaction logs show about on-task behaviour?
Four architecturally independent but structurally consistent apps: the same offline Vuforia + Unity pipeline and the same bilingual services layer.
AR content pipeline
Printed Bilingual Storybook
Each page is both a narrative panel and a Vuforia ImageTarget, authored in Adobe Illustrator.
Vuforia Marker Recognition
The phone camera recognises the page and reports its identifier.
Unity AR Content Engine
Anchors the matching Blender-modelled 3D scene to the plane of the page.
Four Curriculum Components
Measurement · Life Science · Physical Science · Number Concepts.
Shared services layer
Bilingual Content Layer
Sinhala and English labels, instructions and audio narration, toggleable by learner or teacher.
Interaction Manager
One consistent tap-based model reused across all four components.
Learning Feedback System
Animated responses, hints and audio confirmation after every interaction.
Learning Evaluation Module
Logs AR time, quiz time, on-task time, mistakes and retries per session.
A quasi-experimental, non-equivalent control-group study at a private tuition institute in Sri Lanka in August 2026. Students completed a 12-item, 55-point curriculum-aligned test written in Sinhala and reviewed by teachers. Engagement was measured through instrumented AR sessions, a student survey and a teacher panel.
Experimental · n = 32
Pre-test → AR session → Post-test
Control · n = 30
Textbook instruction → Test
Teacher panel · n = 5
Five-point Likert review
Session logs
41 logged · 35 analysable
Institute approval and guardian consent were obtained. Logs are keyed to anonymous session and device IDs only.
31 of 32 students improved after a single AR session, and none declined.
Mean gain (AR group)
+10.22
points of 55 · 95% CI 8.67–11.77
Above textbook control
+10.57
points · Hedges' g = 1.34
On-task time
72%
of session time across 35 sessions
Enjoyed the AR activities
96.9%
31 of 32 students
Knowledge test score (of 55)
Group means; whiskers show 95% confidence intervals.
AR pre-test
AR post-test
Control (textbook)
AR pre-test
AR post-test
Control (textbook)
Paired t(31) = 13.42, p < .001, dz = 2.37. Between groups, Welch t(58.7) = 5.39, p < .001. Wilcoxon and Mann–Whitney tests agree.
Test performance table
Same 55-point instrument for every group.
| Group | n | Mean (%) | SD | 95% CI |
|---|---|---|---|---|
| AR pre-test | 32 | 34.25 (62.3) | 7.13 | [31.7, 36.8] |
| AR post-test | 32 | 44.47 (80.9) | 8.52 | [41.4, 47.5] |
| Control (textbook) | 30 | 33.90 (61.6) | 6.87 | [31.3, 36.5] |
| AR gain | 32 | 10.22 (18.6) | 4.31 | [8.67, 11.77] |
Reading these results
Assignment was by intact class and the control group sat no pre-test, so the study shows an association between AR instruction and higher attainment rather than firmly attributed causation. The single-session, immediate post-test design means the effect is an upper bound.
| Component | Sessions | AR time (s) | Quiz time (s) | Mistakes | Retries |
|---|---|---|---|---|---|
| Basic Measurement | 11 | 77.4 ± 33.1 | 47.9 ± 20.3 | 1.00 ± 1.00 | 1.00 ± 1.00 |
| Basic Life Science | 11 | 66.3 ± 18.7 | 54.1 ± 25.1 | 1.00 ± 0.77 | 0.18 ± 0.40 |
| Basic Physical Science | 6 | 56.7 ± 12.6 | 40.9 ± 12.3 | 0.33 ± 0.52 | 0.00 ± 0.00 |
| Basic Number Concepts | 7 | 57.4 ± 29.2 | 30.5 ± 15.3 | 0.29 ± 0.49 | 0.00 ± 0.00 |
| Overall | 35 | 66.4 ± 25.9 | 45.2 ± 21.1 | 0.74 ± 0.82 | 0.37 ± 0.73 |
Student experience survey
Share of AR-group students agreeing, n = 32.
Enjoyed seeing the 3D objects appear
100.0%Enjoyed the AR activities
96.9%Enjoyed the story
93.8%Activity helped them understand the lesson
87.5%Would definitely use an AR storybook again
87.5%Found the instructions easy to understand
81.3%Teacher panel ratings
Mean on a 1–5 scale, n = 5 practising primary teachers.
AR activities hold students' attention
Story-based approach increases enjoyment
Would recommend to other teachers
Quizzes support the learning objectives
Overall rating of the AR storybook
Content appropriate for primary students
What worked
Teachers rated attention retention and narrative enjoyment highest (4.8/5). Favourite activities were spread evenly across all four domains, so no single component carried the appeal.
What to improve
6 students (18.8%) found instructions hard to follow, and 14 reported at least one difficulty. Audio clarity, 3D manipulation and curriculum-level calibration are the main refinement targets.
Next steps
A randomised multi-site trial with a pre-tested control and delayed post-test, device-level benchmarking, and item-level response capture for instrument validation.
Unity 3D
Cross-platform engine running the AR scenes, interactions and C# app logic.
Vuforia Engine
On-device ImageTarget recognition and tracking of the printed storybook pages.
Blender
Modelling and animation of the 3D objects, characters and processes.
Adobe Illustrator
Storybook illustration and layout, designed to double as AR markers.
Firebase
Optional sync of anonymised session logs when a connection is available.
Android
Target platform: ordinary Android 8.0+ phones, no headsets or depth sensors.
Fully offline
Tracking and rendering run on-device after install.
Low-cost hardware
No headsets, depth sensors or peripherals.
Privacy by design
Logs keyed to anonymous session and device IDs only.
B2B Institutional Licensing
Annual per-school licences for all four components with an admin dashboard.
Printed AR Storybooks
Physical sales of the four marker-enabled storybooks per student.
Freemium Entry
One chapter per subject free to lower adoption barriers.
Expansion to Grades 6–9 Science and Mathematics, a Tamil-medium release, and SAARC region localisation position AR STEAM as a regional category leader.
Select an assessment to see what it covers, when it takes place and the marks allocated.
Date: March 20, 2026
Marks: Good
Defines the research problem, gap, objectives and methodology for the group and for each of the four components, assessed through a proposal report and presentation.
Deliverables
Date: May 15, 2026
Marks: Very Good
Reviews the system design, the shared AR framework and the first working prototype of each component.
Deliverables
Date: August 31, 2026
Marks: Very Good
Demonstrates the near-complete storybooks and AR scenes and presents the classroom evaluation plan.
Deliverables
Date: October 21, 2026
Marks: Pending
Final demonstration of the complete framework with the classroom study results, together with the final reports, research paper and project website, followed by an individual oral examination (viva) on each member's component, contribution and understanding of the research.
Deliverables
Documents produced so far and those still pending. Pending items will be linked here once submitted.
Project Charter
Group scope, roles and schedule
Proposal Documents
Individual component proposals
Check List Documents
Progress check lists
Final Documents
Group report and individual final reports
Slides from past presentations for each component. Upcoming presentations will be added as they are delivered.
Proposal Presentation
Research problem, objectives and proposed methodology
Progress Presentation 1
Design, architecture and early prototype
Progress Presentation 2
Near-complete system and evaluation plan
Final Presentation
Completed system, classroom results and conclusions
R26-IM-003 · Final year research group, SLIIT 2026.

Mendis B.M.U.I
Group Leader

Bandara M.R.J.K
Member

Jayasundara R.K.M.J.Y
Member

Ilayperuma M.I
Member
Supervised by

Mr. Ishara Gamage
Supervisor

Mr. Nushkan Nismi
Co-Supervisor
Achievements
Questions about the research, the storybooks or a classroom trial? Send us a message.
Sri Lanka Institute of Information Technology
New Kandy Road, Malabe, Sri Lanka
Team phone numbers