Group SE-06
AUGMENTED REALITY TO
CREATE A HEALTHY
ENVIRONMENT
Group members:
● Filza Manzoor 19B-039-
SE
● Abdul Saboor 19B-040-
SE
Reference:
[Link]/MMWR7121
2
Centres for Disease Control & Preventi
on
PROBLEM STATEMENT
● Urban life is a blend of noise, pollution, chaos, haste and ailing health.
● Life is compromised in terms of health, leisure, greenery, and mental peace.
● Lack of space for the plantation in the houses. Less awareness about indoor
plantation.
● Unexplainable weather conditions in the city such as; Torrid, Heatstroke,
Downpour etc. destroy the greens.
Reference:
[Link]
rage-Weather-in-Karachi-Pakistan-Year-
Round
4
● Karachi's green areas have been decreased by 40% from 2005 to
2022. This has contributed to deaths with environmental diseases
upto a huge number.
1,544,131 infectious head
count
30,426 Deaths
Whoa! That’s a big number!
Reference: [Link]
5
What Should we do?
No adverse
reactions
Affordable
Naturally
Healing and
Stable & therapeutic
Beneficial
6
Our Solution
● Domestic Level
● Already existing
Tools
● Available on Tap
Greens & Plants
Catered by: Augmented Reality based Mobile Application
7
Which image reflects tranquility?
8
BACKGROUND AND
LITERATURE REVIEW
9
SIMILAR APPLICATIONS
IKEA PLACE
● This application uses AR technology to allow its users
to visualize how furniture will look in their own
homes. Not only would it take the hassle of furniture
shopping off, but it would also eliminate the burden of
returning any furniture that does not fit.
Reference: [Link]
apps/say-hej-to-ikea-place-pub1f8af050
10
SIMILAR APPLICATIONS
DULUX VISUALIZER
● The app allows you to either take a photo on your
phone or tablet of the wall you are considering
painting and then use the app settings to re-colour it on
screen.
Reference: [Link]
visualizer-app
11
ALGORITHMS
● Simultaneous Localization and Mapping (SLAM)
Simultaneous localization and mapping (SLAM) or Concurrent Mapping and Localization (CML) problem. SLAM is considered to be a
complex problem because to localize itself a object needs a consistent map and for acquiring the map the object requires a good estimate
of its location. The Complexity of all SLAM Algorithms is O(n2);
Reference: [Link]
12
● Monocular Visual - Simultaneous Localization and Mapping (VSLAM):
The Vision based / Visual SLAM uses cameras to perform the Mapping and Localization of the object/item. Visual SLAM uses
vision-based sensors: Monocular, Stereo and RGB-D Cameras.
Reference: [Link]
[Link]
13
● LiDAR SLAM
The technique known as light detection and ranging (lidar) primarily utilizes a laser sensor (or distance sensor). Laser sensors often
provide point clouds in 2D (x, y) or 3D (x, y, z) dimensions. For the purpose of creating maps using SLAM, the laser sensor point cloud
offers highly accurate distance measurements.
Reference: [Link]
14
DIAGRAMS
15
SYSTEM DIAGRAM
● The user would give a video input through the
camera integrated in the mobile.
● The video would then be send frame by frame in
the application server.
● The frames would then be processed in different
phases like Image processing, Image Tracking,
AR virtual environment
● After the process the user would be able to see
AR Environment Display and Augmented Images
through their mobile screen.
● The user can map plants and furniture items
virtually anywhere in the area.
● Image tracking is also done in the application
through AR components.
● The user would also be able to rotate and resize
augmented items virtually in the augmented
environment through the application.
16
ACTIVITY DIAGRAM
17
ACTOR USE CASE DIAGRAM
18
SOFTWARE ANALYSIS
19
FRONTEND
● Programming Language: Dart ● Programming Language: JavaScript ● Programming Language: C#
● Project Size: Flutter apps tend to ● Project Size: React Native applications ● Project Size: Unity apps also tend to
have larger file sizes. have a JavaScript runtime environment. have larger file sizes.
They usually have a bigger build size,
● Layout: Flutter uses a widget ● Layout: UI Toolkit comes with a
style for constructing the user ● Layout: For the construction of UI dedicated UI animation system that
interface. React Native uses JavaScript and JSX. simplifies the production of UI
Flutter widgets are pre-made.
transitions in your projects.
● Performance: React Native uses
● Performance: The Dart code that
JavaScript bridges to communicate with ● Performance: UI ToolKit (unity) takes
Flutter uses compiles directly to
C, which is about as close to the native code, and bridges can slow a larger time to compile and build so it
native as you can expect. Flutter app performance down. makes the performance slower.
code does not require any bridges
to communicate with native code.
20
MIDDLE END
● ● EasyAR Sense is a ● Vuforia uses Image tracking
AR CORE supports in Android 7.0+
and ios 11.0+ standalone SDK, it provides features using image targets,
flexible dataflow-oriented cylinder targets, multi-
component-based API targets, and VuMarks.
● ARCore presents points, plane
detection, pose, light estimation,
● It can work with or without ● Vuforia uses object tracking
anchors, image tracking, face
tracking, object occlusion, and AR Foundation. to recognize objects by their
cloud anchors. shape. It does this by using a
pre-existing 3D model of the
● EasyAR Sense provides object.
● ARCore are environmental sensing abilities of the real
understanding, including catching world, supporting Image
Tracking, Object Tracking, ● Environment tracking
the size and location of surfaces,
and light estimation, including real- Surface Tracking, Motion features allow scanning the
life lighting conditions. Tracking and Sparse Spatial surroundings with the Vuforia
Map, Dense Spatial Map. Area Target Creator app.
21
22
BACKEND
Core Features Core Features Core Features
● Realtime Databases (NoSQL) ● Open-Source ● Authentication
● REST API ● RDBMS(SQL) structure ● GraphQL and REST APIs
● Scalable hosting ● Pre-Defined Schemas ● Datastore
● Analytics ● Authentication ● Analytics
● Test lab ● Data Security ● Notifications
Pricing Pricing Pricing
● Free tier available ● Free To Download ● Free tier available
● Pay as you go for paid plans ● Pay as you go for paid plans
23
PROTOTYPE
24
25
26
27
28
29
Prototype Link
30
Group SE-06
Milestone# 2
Group members:
● Filza Manzoor 19B-039-
SE
● Abdul Saboor 19B-040-
SE
PROOF OF
CONCEPT
● Plane detection
● Mapping of fixed models
● Mapping of variable models
● Model Mapping
(Relocation/Rotation/Resizing)
32
Objective 1
Plane detection
33
Objective 2
Model Mapping (Fixed)
34
Objective 3
Model Mapping (Variable)
35
Objective 4
Model Mapping (Relocating/Rotation/Resizing)
36
USE CASE DIAGRAM
(Revised)
37
USE CASE 1: Select
Environment
38
FRONT-END
CLASS DIAGRAM
CLASS (Backend)
39
USE CASE 2: Select
Item
40
USE CASE 3: Place
Item
41
USE CASE 4: Move Item
42
CLASS DIAGRAM
CLASS (Backend)
43
All Use-
Cases
44
COMPONENT
DIAGRAM
45
DESIGN
ARCHITECTURE
3-TIER
46
ALGORITH
M
1. SLAM & VSLAM
● A device may map its surroundings using Simultaneous Pseudo Code:
Localization and Mapping (SLAM) technology while also
keeping track of where it is in relation to the map. robot_pose = update_pose(robot_pose, sensor_data)
# Update the map
● Local Mapping: New 3-D map points are constructed utilising
the current frame if it is found to be a key frame. map = update_map(map, sensor_data, robot_pose)
# Check for convergence
● This type of algorithm allows the device to create a map of its
environment. if check_convergence(map, robot_pose):
Pose pose = hitResults[0].pose; //getting the
● VSLAM uses cameras to perform Mapping and Localization of intersection point
objects or items.
Instantiate([Link],
● This is a technique that uses vision based sensors. [Link], [Link]);
● This algorithm calculates the position and orientation of a
device with respect to its surroundings
47
ALGORITHMS
COMPARISON
ALGORITHMS TIME COMPLEXITY SPACE ERROR HANDLING SCALABILITY
COMPLEXITY
VSLAM O(n^2) O(n) ● Redundancy ● feature based approaches
● Robust Estimation tend to be more scalable
● Re-localization than direct methods.
● Scalability depends upon
the complexity of the
problem being solved.
LIDAR O(n^3) or O(n^2) O(n) ● Redundancy ● depends on specific
● Robust Estimation methods and the
● Re-localization resources available.
EKF O(n^3) O(n^2) ● Monitoring and ● Scalable for moderate
diagnosis sized systems.
● Adaptive tuning
● Usage of an accurate
model of the system
UKF O(n) O(n) ● Monitoring and ● More scalable than EKF
diagnosis algorithms.
● Adaptive tuning ● Scalability depends on
● Usage of an accurate the methods used and the
model of the system available resources.
48
References: (Reference to Slide 47 & 48)
● Mehmet Korkmaz, Nihat Yılmaz and Akif Durdu, Comparison of the SLAM algorithms: Hangar experiments, Selcuk University in Konya,
TURKEY
● Yanhao Zhang, Teng Zhang, Shoudong Huang, Comparison of EKF based SLAM and Optimization based SLAM Algorithms, Centre for
Autonomous Systems, University of Technology Sydney, Sydney, Australia
49
Thank
you!
Any questions ?
50
