My presentation slide for final year project in 2022/2023 session. International Islamic University Malaysia (IIUM). Mechatronics Department More info: https://github.com/KOE-Wayfind

1. Development of interactive
turn-by-turn navigation in
Kulliyyah of Engineering
FYP1 Presentation
Muhammad Fareez Iqmal
1914577
Mechatronics Eng.
Supervisor: Dr Ahmad Imran

2. PROBLEM
STATEMENT
To address the difficulty in navigating the Kulliyyah of Engineering building. The
building is large and complex, with many rooms and corridors, which can be
confusing for first-time visitors. This can lead to frustration and wasted time for
trying to find their way around. Additionally, the lack of clear landmarks in the
building can make it difficult for users to orient themselves and determine their
location.

3. To develop a turn-by-turn navigation
system for mobile application for the
KoE building using Augmented Reality
(AR)
To improve the user experience and
accessibility of the building for
students, staff, and visitors
To provide a more efficient, cost-
effective, and scalable solution for
indoor navigation than existing
technologies.
To reduce the frustration and wasted
time associated with navigating the
building

4. Literature Review
AR
Location
based
GPS
Vision
based
Marker-
based
QRCode
ARTag
Markerless
SLAM
Template
matching
Different types of AR navigation

5. Literature Review
MARKER BASED MARKER-LESS
Marker are easy to identify and track
Need good algorithm to accurately
recognizes the environment
Not require powerful processing power
Utilizes GPU for processing the data
fast
Costly and reduces the building’s visual
aesthetics
No addition needed to be done on the
building

6. Literature Review
• Intuitive UI is necessary to ensure that users can easily
and effectively navigate through the space, while also
considering safety concern
• UI need to be designed so that is doesn’t distract the
user from their surrounding
• Arrows is good visual aids
• Text based auditory instruction helps improving the
overall user experience
USER EXPERIENCE OF USING AR-BASED INDOOR
NAVIGATION SYSTEM

7. Methodology DATA COLLECTION
TEMPLATE MATCHING ALGORITHM
AR IMPLEMENTATION
NAVIGATION SYSTEM DEVELOPMENT

8. Data Collection
Starting with KOE E1 block
LEVEL 2
• Conference Room A
• Conference Room B
• Engineering Student
Society Office
• One Stop Centre
• Academic Staff Office
Images were taken during
the day to ensure the best
lighting condition. So, all of
the image features can
clearly see.
Image were captured from
multiple angle

9. Template matching development
SuperGlue is a graph neural network that
simultaneously performs context aggregation,
matching and filtering of local features for wide-
baseline pose estimation. It is fast, interpretable,
and extremely robust indoors and outdoors.
SuperGlue
Learning Feature Matching with Graph
Neural Networks

10. Template matching development
• Extreme wide-baseline
image pairs in real-time on
GPU
• State-of-the-art
indoor+outdoor matching
with SIFT & SuperPoint

11. AR implementation
ARFoundation is chosen as AR framework
• Cross Platform – Supports Android, iOS, etc.
• First-party support for AR technologies - ARFoundation
• Large community – Easy to find support
• Established 3D engine – Suitable for creating AR
applications in 3D space.

13. Navigation system
• Build 3D model of a building from 2D image close to
actual scale
• 3D model can be exported to Unity

14. Navigation System
• Pathfinding using Unity’s Navigation Mesh
(NavMesh)
• Ground as walkable area, walls as obstacle.
• NavMesh will calculate the optimum path
from the origin to the destination.
• Allow for dynamic updates as the user
moves.
• Hence, NavMesh play crucial role in the
development of turn-by-turn navigation
development.

15. UI Prototype
• Blue line indicates the path that the user
need to follow
• Heads up direction telling the user what to
do
• The destination is shown at the bottom of
the screen

16. Project
planning
FYP1 TIMELINE
Week
No. Project Activities 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 Selection of Project Title
2 Planning and Research
Define project scope
Define project outline
Literature review research
3 Analysis
Collecting data
Processing data
Pre-alpha development
4 FYP1 report submission

17. Project
planning
FYP2 TIMELINE
Week
No. Project Activities 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1
Template matching
algorithm
Data collection
Algorithm development
2AR Development
Navigation system
Localization
UI Development
3Minimal Viable Product
4User testing
Collecting feedback and
improvement
4Presentation and showcase

18. Results
Blue dots representing the
features discovered from the
inference image

19. Results
A different image (matcher)
that are not in training is
tested against the inference
image
Found 366/461 inlier
correspondences.
(79.39% match)

20. Results
The results of the template
matching method using the
SuperGlue algorithm
were found to be good
The consistency across different images and
landmarks further confirms the robustness of
the algorithm. However, further adjustment
may need to be done to improve the accuracy
of the algorithm.

21. This projects aims
• To develop turn-by-turn navigation, mobile application for the KoE using AR
• Using markerless approach of localization - SuperGlue algorithm
• Use Unity to as a development platform
• Highlights the feasibility and effectiveness of using natural landmark for indoor
navigation
• The complete development will be done in FYP2
Conclusion

22. Thank you!