Fire Fighting Bot To Allow A User To Control A Fire Fighter Robot
Fire Fighting Bot To Allow A User To Control A Fire Fighter Robot
has some limitations and designed to sense fire with the smoke, limited areas.
To reduce limitations and to optimize with new technology, the project is proposed.
We propose cost-effective fire detection MCNN architecture for fire detects videos. . The entire
code is written in pure python language using the open CV library for image processing.
The raspberry pi hardware with vision camera is using for the real-time implementation with
the feature of SMS and fire extinguisher with Manual control model and object detection
sensor.
This can greatly minimize the chances of big disasters and can control an abnormal event on time with
comparatively minimum possible loss. Among such abnormal events, fire is one of the commonly happening
events, whose detection at early stages during surveillance can avoid home fires and fire disasters.
Fire fighter robot with deep learning 2021/ The principle used in the design, such that whenever fire There is algorithm is implemented to detect the
and machine vision Springer occur in the particular area immediately the camera will Fire in real-time .
Author: The process is a time consuming process to detect
detect the fire and tries to control the fire at the initial
the Fire
Amit Dhiman, stage for that we are implementing Artificial Intelligence
Neel Shah and Neural Network algorithm. With the help of smart
phone we can monitor and control the robot. With the
help of ultrasonic sensor we can avoid the collision for
the robot. In this project, we control the fire at the initial
stage so we can avoid destruction.
DESIGN AND IMPLEMENTATION 2021 / The principle used in the design, such that whenever fire There is algorithm is implemented to detect the
OF FIRE DETECTION ROBOT A Journal occur in the particular area immediately the camera will Fire in real-time .
Of The process is a time consuming process to detect
USING RASPBERRY PI Compositi detect the fire and tries to control the fire at the initial
the Fire
TECHNOLOGY on Theory stage for that we are implementing Artificial Intelligence
Author: and Neural Network algorithm. With the help of smart
Y.Naveen Kumar, phone we can monitor and control the robot. With the
M.Ramya Sri, help of ultrasonic sensor we can avoid the collision for
the robot. In this project, we control the fire at the initial
stage so we can avoid destruction.
presentation title
JOURNAL NAME AND AUTHOR Year ALGORITHM AND OBSERVATION Adv/Dis
Convolutional Neural Networks 2018/ The recent advances in embedded processing have NO hardware is processed.
based Fire Detection in Surveillance IEEE Access enabled the vision based systems to detect fire during Only the simulation part is done
Videos surveillance using convolutional neural networks
The algorithm simulation part also not achieved
AUTHORS (CNNs). However, such methods generally need more the efficiency as required.
KHAN MUHAMMAD computational time and memory, restricting its
JAMIL AHMAD
implementation in surveillance networks. In this
research article, we propose a cost-effective fire
detection CNN architecture for surveillance videos.
The model is inspired from GoogleNet architecture,
considering its reasonable computational complexity
and suitability for the intended problem compared to
other computationally expensive networks such as
“AlexNet”.
FIRE FIGHTING ROBOT 2018
JCAS
There are many possibilities a fire can The hardware with sensors and
WITH ARTIFICIAL (Journal of start in an industry or in any remote area. image processing done without
INTELLIGENCE applied
science and
It’s a worst-case scenario, causing heavy any AI algorithms.
AUTHORS
computation) losses not only financially but also
Dr. T.K.S Ratish Babu destroying areas surrounding it. Robotics Proposed systems not included
Dr.V.Balaji is the emerging solution to protect human the combination results with
Dr.V.Anantha Krishna lives and their wealth and surroundings. hardware and software.
The aim here is to design a fire detector
and extinguisher robot using
RASPBERRY PI 3 MODEL B
Microcontroller. A robot capable of
20XX presentation title 6
fighting a simulated household fire will be
Existing System
• In existing system, sensors are used to detect the fire. But the main
disadvantage of the sensors is, they detect the fire when it reaches a
particular level of the temperature. Before it reaches a particular
temperature there may occur a large destruction like human loss or
property loss.
• Majority of the research since the last decade is focused on traditional
features extraction methods for flame detection.
• The major issues with such methods is their time consuming process of
features engineering and their low performance for flame detection.
• Such methods also generate high number of false alarms especially in
surveillance with shadows, varying lightings, and fire-colored objects.
20XX presentation title 7
Proposed system
• we extensively studied and explored deep learning architectures for early flame
detection. Motivated by the recent improvements in embedded processing
capabilities and potential of deep features, we investigated numerous CNNs to
improve the flame detection accuracy and minimize the false warnings rate.
• Then proposed with modified CNN for the suitable with both detection and
efficiency for the fire.
• Raspberry Pi also offers many other advantages like the coupling of additional
hardware, gives many opportunities to use software and networking, which helps
this project to use in the future with other projects as well.
20XX presentation title 8
PROPOSED SYSTEM:
BLOCK DIAGRAM:
GSM MODEM
U S B
H
D UART
CAMERA M
I
Raspberry Pi
Processor
W
I Motor
F Driver
Ultrasonic And Relay
I
Sensor
SD CARD
Robot
Water Model
Relay
Pump
• The MCNN gives the efficient fast detection for fore detection
with variation from CNN as iteration increases.
• The error in the first methods is being applied to all the stages
to overcome and achieve the fast detection.
CNN, the over fitting problem, and combines dropout and the Sequential optimizer with the CNN.
• An improved CNN algorithm based on dropout and the sequential optimizer is proposed, and parallelization is
• The use of Optimizer leads the MCNN-DS to work faster because it does not use the whole training instances in each
integration and can dynamically adjust the estimation of the first- and second-order matrices of the gradient of each
parameter according to the loss function, which can reduce the amount of calculations and improve the computing
speed.
Phase 2
• The hardware based robot to be designed with raspberry pi
control and checking with real-time environment as prototype.
necessitates the importance of developing early fire detection systems. Therefore, in this research
article, we propose a cost-effective fire detection MCNN architecture for surveillance videos. it is
proved that the proposed architecture dominates the existing hand-crafted features based fire detection
methods.
• Through this project, we can detect the fire by using camera and extinguish it with the help of water
pump. Ultrasonic sensor helps us to avoid collision between robot and obstacle. This fire detection
robot using raspberry pi technology is effective enough to fight against fire on a small scale. It can
sense fire better at darker places and flat surface areas. The prototype to be done in the phase II.