Introduction

The requirement of a Janitor robot that can be deployed to perform janitorial tasks in washrooms and used as a
supplement to human employees, cleaning robots eliminating the need for workers to perform many simple and
repetitive tasks. Robots introduce a new level of efficiency that human cleaners are simply incapable of attaining due
to inherent limits of human productivity. Robots may indeed be the future of cleaning, eliminating the need for
service staff on the inside of restroom doors, and the need for human cleaning staff to perform time-consuming tasks
day after day.
Enclosed in this document is our project proposal that includes the concept to build a janitor robot that would
perform simple janitorial tasks in the washrooms. Conceptually this is achieved by making use of Stereo Camera to
determine depth & dimensions, SLAM (Simultaneous Localization and Mapping) techniques for mapping the
washroom, object detection techniques for detecting objects (i.e., door, trash, dustbin, countertop, washbasin etc…)
using deep learning, Robotic arms with 7 DOF for sanitizing, wiping and drying and pick and drop operations similar to
a human arm, Jetson Nano as the CPU using ROS as the brain of the robot.

Innovation Aspects:

1. Reliable Mapping and 3d image processing- This can be achieved by optimizing data points.
2. Slip Resistant Wheel- The robot will be equipped with a slip-resistant wheel which will not only improve the overall
balance of the robot dynamically but will also prevent the robot from losing control over wet surfaces. Since the robot
must be designed to handle wet surfaces, it must be able to move around the location with precision and accuracy.
This wheel increases the dynamic equilibrium, non-moving-state traction of the robot, and provides the necessary
stability while performing the clearing actions on the toilet.
3. IP4x Protection- The robot needs to be resistant to water and moisture in order to work for extended periods
without any parts wearing out. Since bathrooms relatively tend to have more moisture content, measures must be
taken to protect the electronic components, circuit boards, batteries, and other metallic joints. We will be making the
robot waterproof by making the body using plastic (instead of metal). Once the electronic components are completed,
we can coat them completely using Paraffin wax and that would keep our electronics dry and safe. At places where
openings are present, we can utilize the same technique. We shall also be utilizing Brushless Motors which would
assist us in the same.
4. SWD (Sanitizing, Wiping, Drying) Arm- The custom-designed arm that has been inspired specially to facilitate the
SWD process. The robot must be able to clean every corner of the bathroom without any issues, and clean under the
shelves, racks, etc., without disturbing the other items present in that place.
5. Autonomous charging - One of the main aspects that come into the picture when a robot is considered is its
longevity, duration of work, and up to what rate of automobility it will be able to provide. Every robot has a duration
after which it’s battery would be low, and it would need to be charged. We will be preparing a station that would
allow the robot to be charged. After a certain threshold in the battery percentage or after completing the set of tasks,
the robot will be able to map its position back to the charging station. The station will contain wireless charging pads
which will initiate the charging once the robot reaches there. The adapter built-in within the charger will also take care
in converting the AC to DC.
6. Advertising - One way by which we will be able to obtain revenue out of the robot is by utilizing an advertisement
model. The robot will contain one LED panel/small poster holder where advertisements of certain companies can be
posted. This will be a really good way to earn revenue.
Technical Details:

Block Diagram of Hardware Architecture:

Figure 1: Hardware architecture block diagram

Hardware Stack:

Jetson Xavier NX: The NVIDIA Jetson Xavier NX Developer Kit brings supercomputer performance to the edge. It
includes a Jetson Xavier NX module for developing multi-modal AI applications with the NVIDIA software stack in as
little as 10 W. we can also take advantage of cloud-native support to more easily develop and deploy AI software to
edge devices. The developer kit is supported by the entire NVIDIA software stack, including accelerated SDKs and the
latest NVIDIA tools for application development and optimization. A Jetson Xavier NX module and reference carrier
board are included, along with an AC power supply.
Invento mg 495(Servo Motor): A servo is a precise and powerful way of converting rotational motion into linear
motion. Servo motors are used for robotic applications that require precision positioning. They are small, powerful,
easily programmable, and accurate. Most importantly, they allow for near-perfect repeatability of motion.
Adafruit 16-Channel 12-bit PWM/Servo Driver - I2C interface - PCA9685: Servo motors are often driven using the
PWM outputs available on most embedded MCUs. The Adafruit 16-Channel 12-bit PWM/Servo Driver-I2C interface
PCA-9685 can be used to communicate with a PWM driver and also using this breakout, we can easily drive up to 16
servo motors using Python libraries.
Dc Motor For Locomotion/Wiping/Drying: In robotics, permanent magnet DC motors are used. These motors are
known for their high torque to inertia ratio. DC motors are the most simple motors to use. They can reach a high
rotational speed that is dependent on the input voltage.
Dc Motor Driver Such as L298N: This L298N Motor Driver Module is a high-power motor driver module for driving DC
and Stepper Motors. This module consists of an L298 motor driver IC and a 78M05 5V regulator. L298N Module can
control up to 4 DC motors, or 2 DC motors with directional and speed control.
Li-ion Battery: Li-ion batteries can be recharged hundreds of times and are more stable. They tend to have a higher
energy density, voltage capacity, and lower self-discharge rate than other rechargeable batteries. This makes for
better power efficiency as a single cell has longer charge retention than other battery types
RealSense camera D455: The D455 is the fourth in the D400 series, designed from feedback and knowledge gained
from over 10 years of stereo camera development. The D455 extends the distance between the depth sensors to 95
mm which improves the depth error to less than 2% at 4 m. To improve the RGB image and the correspondence
between the depth and RGB images, the RGB sensor includes a global shutter and is matched to the depth FOV with
both having an 86° field of view. The camera integrates an IMU to allow your application to refine its depth awareness
in any situation where the camera moves. This allows improved environmental awareness for robotics and drones.
Mecanum wheels: The Mecanum wheel is based on a tireless wheel, with a series of rubberized external rollers
obliquely attached to the whole circumference of its rim. These rollers typically each have an axis of rotation at 45° to
the wheel plane and 45° to the axle line. Each Mecanum wheel is an independent non-steering drive wheel with its
powertrain, and when spinning generates a propelling force perpendicular to the roller axle, which can be vectored
into a longitudinal and a transverse component in relation to the vehicle.
Submersible water pump sanitizing Liquid: Submersible pumps are so named because they are designed to be
submerged in fluid. Sometimes, submersible pumps are called “electric submersible pumps” or “ESP’s.” Because they
have to be submerged, they are not subject to pump cavitation, which occurs when there is a high disparity in
elevation between the fluid surface and the pump. Submersible pumps are more efficient and require less power than
jet pumps because they push water to the surface instead of pulling it. Submersible pumps have a motor that is
hermetically sealed in the casing of the pump. This ensures that water never touches an electrical component.
Spray nozzle: A spray nozzle is a precision device that facilitates the dispersion of liquid into a spray. Nozzles are used
for three purposes: to distribute a liquid over an area, to increase liquid surface area, and create impact force on a
solid surface. A wide variety of spray nozzle applications use several spray characteristics to describe the spray.

Figure 2: Software architecture block diagram

Software Stack:

Operating System: Ubuntu 18.04

Software: ROS Melodic
Mapping:
- RTabMap is a VSlam package that will be used to understand and build a map of the environment. RTabMap is
available in the ROS package.
- RGB Depth data will be taken from the RealSense and can be passed to the RTabMap.
- IMU data will be taken from the RealSense camera and will be passed through the IMU filter Madgwick.
- Then additional Kalman filter EKF will be used to combine the above two types of data.
- This way, camera transform is going to be updated frequently from the IMU data and sometimes will be corrected
by the RTabMap data.
- Inbuilt loop closure feature will help to autocorrect the map after the robot will come to the initial position.
Object Detection:
- YOLO or “You Only Look Once” is a neural network based realtime object detection algorithm.
- The YOLOv3 model will be used for detecting the door, trash, bin, washbasin etc.
- The YOLOv3 model will be trained on the images of different types of objects.
- This model will help to predict the class probabilities using Keras library.
- An object detection node will be created in ROS. This node listens to the camera input and performs object
detection.
Technical problems to be addressed
1. Less camera field of view can cause missing some data while cornering the robot for building the 3d map as it
may rely only on the IMU data. (technical problem: 3D mapping)
2. YOLOv3 is faster but less accurate than Mask R-CNN. Mask R-CNN is more accurate, but because of its slow
speed, we cannot use it for real-time applications.
3. Rusting of Metal Joints and motors (on long term use)
4. Deciding Capacity of battery. We will need to work on the optimization of the circuit to maximize its lifespan.
(Technical Problem: battery capacity, life, and optimization)
5. Training and developing the CNN model to be accurate enough to detect objects and the path with minimal
errors.
6. Spraying mechanism of the sanitizer (Technical Problem: design of reliable spraying, Pick and Drop, sanitizing
mechanisms)
7. Reducing time in Pick and Drop of trash (Technical Problem: design of reliable spraying, Pick and Drop,
sanitizing mechanism)
8. Pipeline design for the sanitizer (Technical Problem: design of reliable spraying, Pick and Drop, sanitizing
mechanisms)

Overview of the proposed solution

Figure 3: Proposed janitor robot design and associated challenges

Bill of materials:

Component name/Specification Qty Price

Controller Unit

Jetson Xavier NX, 8GB RAM, Metal case,64GB SD card 1 45,000

Camera Unit

RealSense camera d455 1 36,000

Motors and Drivers

Invento mg 495(Servo motor) 16 9,600

Adafruit 16-channel 12-bit PWM/Servo Driver- 12C interface-PCA9685 1 400

DC motor 6 12,000

DC motor driver such as L298N 3 600

Others

Li-po battery 1 15,000

Battery charger 1 5,000

Mecanum wheels 4 14,000

A set of 4 127mm Aluminium Mecanum Bearing type roller wheels(2 left
and 2 right)

Submersible water pump sanitizing liquid 1 500

Spray Nozzle 1 1,000

3D printing

Extra material(acrylic,screws,nuts,cables,rods,connecting wires etc) 1 10,000

Tools 10,000

Total 1,59,100
APPENDIX

References

1. https://github.com/IntelRealSense/realsense-ros/wiki/SLAM-with-D435i
2. https://www.intelrealsense.com/depth-camera-d455/
3. http://wiki.ros.org/rtabmap
4. https://arxiv.org/abs/1506.02640
5. http://wiki.ros.org/robot_pose_ekf
6. https://www.youtube.com/watch?v=dEPvZZ4luV4
7. https://www.nvidia.com/en-in/autonomous-machines/embedded-systems/jetson-xavier-nx/
8. https://patents.google.com/patent/US20150158601A1/en?oq=us20150158601
9. https://youtu.be/dEPvZZ4luV4
10. https://youtu.be/9gVupqHqJws
11. https://patents.google.com/patent/US20190246858A1/en?q=janitor+robot&oq=janitor+robot
12. https://patents.google.com/patent/US8382906B2/en?q=autonomous+wet+and+dry+floor+cleaning
Team Description

Team member name Department Other Responsibilities Skills & Interests

Amareshwar Karatagi Pursuing BTech ECE, - Chair at IEEE Education Artificial Intelligence, Natural
REVA University, Society Student Branch Language Processing, Deep
Bangalore Chapter Learning, AR/VR and Web-App
- ECIM Chair of IEEE VTS | Developer.
Student Standing
Committee Chair of IEEE
Student Branch REVA
University
- Researcher in R&D in IoT
on AI, NLP, ML RU
|Honorary Award Wins in
IEEE VTS Global
Innovation Challenge
Contest.

Akshay Dilip Kumar Pursuing BTech EEE, Python,C++,C,Java,JavaScript,CS

REVA University, S, HTML, VHDL, Robotics, Ml
Bangalore and AI, IoT, Computer vision,
Power systems, Embedded
systems

Siddhanta Mandal Pursuing BTech CSE, C/C++,Python,Problem

REVA University, Solving,Computer Vision,Deep
Bangalore Learning,AI,Robotics &
Automation

Niran N Pursuing BTech in - Google Developer - Proficient in the fields of

ECE, REVA Students Club Lead IoT, Deep Learning, Native
University, - Cybersecurity Teaching and Web app development.
Bangalore Assistant, REVA Academy - Passionate about research.
for Corporate Excellence - I play piano and learn
(RACE) Japanese when free.
- Vice-Chair, IEEE SB REVA
University | Core,
Hackathon Club | Vice-
President, Coders' Club

Kapilesh Simha Pursuing B.Tech Python, HTML, 3d Modeling,

CSE(AI & ML), Web Scraping, Data Science
REVA University,
Bangalore
Description of the facilities at the team's disposal

- Desktop computers with parallel computing abilities

- PCB Manufacturing and prototyping facilities.
- Licensed simulation software Autocad, Fusion 360, Blender for simulation and animation of the proposed design.
- 3D Printing facilities for rapid prototyping.
- IIT-Bombay (Eyantra) sponsored robotics and automation lab- The lab provides facilities for Research, Innovation,
and training. Artificial Intelligence, Swarm Robotics, Robot structural design, kinematics, and Nano Technology for
Robots are the thrust areas of research. e-Yantra Robots (IIT Bombay), Nex Robotics Robots are the major types of
equipment used in the research center. The lab has also been recognized for projects such as the Sewage cleaning
robot, Thermal scanner, Jeeva Sethu(Ventilator), UAV projects, Alexa based humanoid robot.

Biographical sketches of the academic mentor and industry mentors

Academic Mentors
1. Dr.Manjula R.B
Area of research: Underwater Communication, Internet of Things, Robotics, Machine Learning.
Recent Publication: Journal Publication (Few Listed)
https://scholar.google.com/citations?user=JC0duXcAAAAJ&hl=en
1. Venkateswara Raju, Manjula R Bharamagoudra, “An Architecture for Enabling IoT interoperability between cross-
platforms”, International Journal of Internet Technology and Secured Transactions, (Accepted June 2020)
2. Venkateswara Raju Konduru, Manjula R Bharamagoudra. “An Analysis on Machine Learning Based Approaches in
Sensor IoT Applications”. International Journal of Advanced Science and Technology, Vol. 29, Issue 10, pp. 7751-
7761, Jun. 2020
3. Manjula R. Bharamagoudra, Sunilkumar S. Manvi “Agent Based Secured Routing for Underwater Acoustic Sensor
Networks”, International Journal of Communication Systems, Wiley Vol. 30, No. 30, 2017 e3281.
4. Manjula R. B., Sunilkumar S. Manvi, Bilal Gonen ``Event Driven Energy Depth and Channel Aware Routing for
UWASNs: Agent Oriented Clustering Based Approach", submitted to Elsevier International Journal of Computer
and Electrical Engineering, January 2017.

Patents (Recent):
1. Manjula R. B, SunilKumar S. Manvi, Venkateshwara Raju, R. C. Biradar ”A System & Method For Granting Access
To A User”, Patent Application No. 201841033354 , India , Published (2017; Under Examination – Last Hearing
Pending)
2. Manjula R. B., R.C. Biradar, Aryalakshmi, Karthik “Thermal Scanner”, Published (2021).
3. On-going projects:
a. Augmented Reality based learning system.
b. Humanoid Robot for Surveillance and Visitor Assistance.
c. Lake Monitoring Using Underwater Communication and Networking.
d. AI based Thermal Scanner and Authentication System (Funded through ATAL Innovation Contest a grant
of Rs. 0.5L and startup support).

2. Kartik Cholachgudda (Ph.D.)

M.Tech. in Communication Engineering

Area of research: Computer Vision, Image Processing, 3D Depth Imaging Technology, Data fusion Algorithms
Recent Publication: K. E. Cholachgudda, R. C. Biradar, and M. Lokanath, "Geometric Calibration of a Time-of-Flight
Depth Sensor and a Colour Camera Pair," 2019 11th International Conference on Advanced Computing (ICAC),
Chennai, India, 2019, pp. 316-323, DOI: 10.1109/ICoAC48765.2019.246859.
On-going projects:
1. Smart Contactless Authentication and Health Monitoring System
 2. High Throughput Phenotyping for Early Detection of Fungal Diseases in Grape Vineyard using Remote
Sensing Techniques
Contribution towards the Proposed Janitor Robot Project:
1.Guidance in developing 3D depth acquisition algorithm for SLAM,
2.Check suitability of existing open-source libraries for 3D mapping and localization
3.Support in selecting right hardware and software specifications

3.Kouame Yann Olivier Akansie

Assistant Researcher, School of ECE, REVA University

Area of research: Computer Vision, Image Processing, Embedded Systems, and IoT
Recent Publication: Kouame Yann Olivier Akansie, Manjula R. Bharamagoudra, Rajashekar C. Biradar. (2020). “Remote
Surveillance and Monitoring System using IoT”, IJAST, vol. 29, no.10s, pp.7741-7750, Jun. 2020
On-going projects:
1. Smart Contactless Authentication and Health Monitoring System
2. Design and Development of a Wheel-Legged Robot for Terrestrial Locomotion
Contribution towards the Proposed Janitor Robot Project:
1.Guidance in designing the robot
2.Guidance in hardware design
3.Support in selecting the right hardware and software specifications

Industry mentors
1.Mr. Suraj N
Founder and CEO of Drona Automations

Mr. Suraj is a first-generation entrepreneur whose passion lies in innovation and exploring new concepts and
contributing to the world through his work. He is the founder and CEO of the startup, “Drona Automations”. He
completed his B-Tech in Mechanical Engineering from REVA University, Bengaluru. He was a Research Intern at IISc
Bangalore for 1 year. His main technical expertise is in Product Design and Design for Manufacturing.
Patents:
6 Patents Filed
1 Patent Granted
Awards:
1. Startup Karnataka Elevate – Karnataka Level
2. JLL prop-tech – India Level
3. MisFits Award – South Asia