A big step in the direction of Industrial safety: Preventing gas leakage using machine learning.

Summary

This video features a discussion with Vinit and D, winners of the Smart India Hackathon 2023. They developed a machine learning-based solution to prevent industrial disasters caused by gas leaks. Their system detects gas leakage in real-time and predicts affected areas, helping prevent accidents like the Bhopal gas tragedy. The solution uses existing sensors and digitizes them for real-time monitoring. The team aims to expand the project by incorporating expert knowledge to enhance their dataset and improve accuracy, ultimately safeguarding lives and infrastructure in industrial settings.

Takeaways

• 🏭 The industrial sector is crucial for national progress but faces challenges, especially with safety risks such as gas leaks.
• 🎉 Vinit and D, winners of the Smart India Hackathon 2023, developed an innovative solution to address gas leaks in industrial settings.
• ⚠️ Gas leaks in factories pose serious dangers, often going unnoticed until it's too late, leading to fatalities like the Bhopal gas tragedy.
• 🚨 Their solution provides real-time notifications and alerts to factory managers and workers, preventing potential disasters.
• 📱 The app they developed digitizes existing industrial sensors, sending critical data about gas leaks directly to users.
• 🌬️ The solution includes a machine learning model that predicts gas leakage spread based on environmental factors like temperature, wind, and terrain.
• 🔧 Industry personnel and emergency services can access the app to monitor sensor data, check gas pressure, and control ventilation in real-time.
• 🌍 The app will be available on Play Store and iOS, allowing widespread access to workers and managers for preventive safety.
• 💡 Their project is currently in the testing phase, with a focus on building datasets to refine the machine learning model for accurate predictions.
• 🔄 They aim to prevent future industrial disasters by providing rapid alerts and information to save lives and safeguard infrastructure.

Q & A

What is the main issue that the winning team at Smart India Hackathon 2023 addressed?

The team addressed the critical issue of gas leakage in industrial settings, which can cause significant harm and even fatalities if not detected and managed in time.

How did the team plan to tackle the problem of gas leakage in industries?

They developed a software solution to digitize existing sensors in factories and provide real-time data to industry owners. Additionally, they created a machine learning model to predict the direction and extent of gas leakage.

What was the team's main motto when developing their solution?

The team's main motto was to save lives by preventing gas leakages before they occur, rather than just fighting the consequences of such incidents.

How does the machine learning model predict the danger zone of a gas leak?

The model takes into account various factors such as temperature, humidity, and landscape to determine how far a gas leak might spread and in which direction, thus identifying the danger zone.

What happens when a gas leakage is detected by the system?

Upon detecting a gas leak, the system triggers an API to send data to a backend server, which then uses the machine learning model to predict the danger zone and notifies relevant stakeholders through the app and local network providers.

Is there a need for subscription to access the team's gas leakage detection application?

No, the application is open source and will be available on the Play Store for anyone to download. It also collaborates with local network providers to notify people in the affected area without the need for an app.

How does the application help in managing different levels of gas potency?

The application is currently being updated to manage different levels of gas potency by improving the machine learning model to consider the weight, density, and other characteristics of individual gases.

What additional features does the application offer to industrial managers?

The application allows managers to check real-time sensor status, control sensors, and contact emergency services directly from the app. It also sends notifications to healthcare services and authorities in case of hazardous situations.

Are the team's sensors proprietary, or do they work with existing industrial sensors?

The team's solution works with existing sensors in industrial setups, digitizing their data and making it accessible through their application for real-time monitoring and control.

What is the current status of the application, and when can it be expected to be implemented?

The application is mostly ready, but the team is still working on improving the machine learning model with a more accurate dataset to enhance the Hazard Zone prediction. The timeline for implementation is not specified but is dependent on the completion of this dataset.

What impact does the team expect their application to have in real-world scenarios?

The team expects their application to significantly help in preventing gas leakage incidents, thereby saving lives and protecting industrial infrastructure, similar to the Bhopal gas tragedy, from occurring.

Outlines

00:00

🏭 Introduction to Industrial Safety and Innovation

The introduction highlights the critical role of the industrial sector in national progress, emphasizing the responsibility of safeguarding infrastructure. The video introduces the guests, Vinit and D, who are part of the winning team from the Smart India Hackathon 2023. Vinit, a computer science student, introduces himself as a founder of a startup, while D shares that he is also a part-time developer for a US-based organization. Their team’s solution addresses industrial safety, particularly in the context of gas leakage prevention.

05:01

🔥 Addressing the Problem of Industrial Gas Leakage

This section dives into the core problem the team tackled: gas leakage in industrial environments, which can be deadly if undetected. The team explains that past tragedies, such as in Bihar and elsewhere, could have been prevented with timely information. Their solution aims to alert factory owners and workers before such accidents occur, leveraging software to detect and prevent issues. They also developed a machine learning model to predict the spread and danger zone of leaked gas based on factors like temperature, humidity, and the surrounding landscape.

10:01

📊 How the Solution Works: Detection and Notifications

The technical functionality of the solution is detailed here. Sensors installed in gas cylinders constantly monitor pressure and gas type. When a leak is detected, the system sends data to a machine learning server that predicts the danger zone and alerts users. Notifications are sent to industry workers, managers, and local network providers, ensuring that even the general public can be warned. The solution is open-source and available to everyone through mobile apps, without a subscription, making it accessible to a wide audience.

📱 App Features and Community Involvement

The discussion moves on to how users interact with the system. Individuals can download the app, sign up, and receive alerts based on their area. The app has features like displaying affected zones and offering real-time updates on the situation. Users can also help during emergencies by contributing skills, and the app connects directly with emergency services like hospitals and fire stations. The focus remains on ensuring that information is shared rapidly, helping prevent or mitigate industrial disasters.

🔧 Enhancing the Solution and Future Plans

This section discusses the team's efforts to further improve the application. They are working with chemists to better understand gases and their behaviors, which will refine their machine learning model. The team is manually testing different gases and environmental conditions to create a robust dataset. Once they gain more knowledge about gases, they plan to enhance the application’s accuracy and bring it into real-world industrial scenarios.

📈 Real-World Implementation and Impact

Here, the team talks about the final stages of the application’s development, focusing on real-world implementation. The app enables factory managers to monitor sensors, control equipment, and contact emergency services remotely. Existing sensors in factories are digitized through the app, allowing managers to act quickly in case of gas leaks or other hazards. The team is optimistic about the impact their solution will have, believing it can prevent tragedies like the Bhopal gas disaster by providing early warnings and helping save lives.

🎯 The Final Vision: Preventing Industrial Tragedies

The conclusion ties the project’s significance to industrial safety and its potential to prevent disasters. The team reiterates their commitment to delivering a solution that can help avoid incidents like the Bhopal gas tragedy, ensuring that people are informed and can take action in time. They express their hope that the application will soon be fully implemented and recognized for its contributions to industrial safety and public health.

Keywords

💡Industrial Sector

The industrial sector refers to the part of the economy involved in manufacturing and production. In the video, the industrial sector is described as the 'backbone' of the nation's progress, implying its critical role in economic development. The team's project focuses on enhancing safety within industrial environments, specifically by addressing risks such as gas leaks.

💡Smart India Hackathon 2023

The Smart India Hackathon 2023 is a competition where teams of students create technological solutions to real-world problems. The team featured in the video participated in this event and developed a project aimed at preventing industrial accidents, such as gas leaks. Their victory in the hackathon signifies the innovation and importance of their solution.

💡Gas Leakage

Gas leakage is the accidental release of gas, which can pose serious health and safety risks. In the video, gas leakage in industrial settings is identified as a 'deadly threat' that has caused numerous tragedies. The team's solution aims to detect and prevent such leaks to safeguard workers and nearby residents.

💡Machine Learning Model

A machine learning model is a system that uses algorithms to analyze data, learn from it, and make predictions. The team in the video developed a machine learning model to predict the spread of gas leaks based on environmental factors like temperature and humidity. This model helps in identifying potential danger zones and warning people in real time.

💡Danger Zone

The danger zone refers to the area around an industrial site that is most at risk during a gas leak. The video explains how the team's machine learning model predicts these zones by analyzing factors such as gas type, wind speed, and surrounding landscape. This enables timely evacuation and minimizes the impact of the leak.

💡Real-Time Data

Real-time data refers to information that is delivered immediately after collection, without any delay. In the context of the video, the team's solution digitizes sensor data from factories, allowing managers to monitor gas leaks and other hazards as they happen. This enables quicker responses to prevent accidents.

💡Sensors

Sensors are devices that detect changes in the environment and send this information to a system for analysis. In the video, the team mentions that many factories already have sensors installed to detect gas leaks, but these sensors are not digitized. Their project integrates the data from these sensors into a mobile app to improve monitoring and control.

💡Open Source Application

An open source application is software that is made freely available to the public, allowing anyone to use or modify it. The team's solution, as described in the video, is an open source mobile app available on platforms like the Play Store. It provides widespread access to gas leak detection and safety information.

💡Emergency Contact List

An emergency contact list is a feature in the team's application that allows factory managers to quickly reach emergency services in case of a hazard. This functionality is highlighted in the video as a critical tool for minimizing response time during incidents like gas leaks, ensuring that help arrives promptly.

💡Bhopal Gas Tragedy

The Bhopal Gas Tragedy was a catastrophic gas leak in 1984 in Bhopal, India, that resulted in thousands of deaths. It is mentioned in the video as a historical example of a preventable disaster. The team’s project aims to create a solution that could prevent such tragedies by quickly alerting people to hazardous gas leaks.

Highlights

The industrial sector is crucial for the nation's progress, and safeguarding its infrastructure is vital.

Vinit and D's team won the Smart India Hackathon 2023 with a project focused on enhancing industrial safety.

Their solution addresses the significant problem of gas leaks in industrial settings, which can lead to deadly consequences.

Previous industrial gas leak disasters, such as those in Bihar and Bhopal, inspired the team's project.

The team developed a system to digitize existing sensors in industries to detect gas leaks in real time.

Their machine learning model predicts the spread of leaked gas by considering factors like temperature, humidity, and the landscape.

The model maps out danger zones to warn people and factory owners of potential hazards before they occur.

The application will notify users via a mobile app and through SMS in collaboration with local network providers.

The system is open-source, accessible through both the Play Store and the iOS App Store, with no subscription required.

Users can enter their area and industry ID for personalized notifications in case of gas leaks.

The system will contact local emergency services, including hospitals, fire stations, and police, as soon as a hazard is detected.

Factory managers can control sensors remotely and check real-time data on gas pressures and sensor statuses through the app.

The team is working on improving the accuracy of the machine learning model by collaborating with chemists to enhance their gas leak predictions.

Their solution is designed to prevent incidents like the Bhopal gas tragedy from happening again by providing timely alerts and notifications.

The project demonstrates how technology and machine learning can significantly improve safety in industrial environments.

Transcripts

[Music]

[Music]

the industrial sector is the backbone of

our nation's progress and with the

progress comes the responsibility to

safeguard the infrastructure today we

are going to meet a winning team whose

Triumph in the smart India haathon 2023

is not just a victory for themselves but

a significant milestone in advancing

safety measures within our industrial

Landscapes let's meet vinit and th

welcome to the show vinit and th please

introduce yourself to us first thank you

sir thank you

sir hello sir I am Vin Jan currently

pursuing my bachelor in computer science

engineering at chandigar University I

the found and a programmer science

startup profit

and now D will talk about himself yeah

myself is D and I'm also puring my best

degree from CH University currently I'm

start I'm in like second year and I'm

also working as a part-time developer

for an US based organization

T great so you two participated along

with your team in the smart India

haathon 2023 and came up as winners so

wonderful congratulations for that now

when you came up with that winning

formula or the winning uh solution uh

let's start talking about that so before

we go on to the solution let's talk

about the problem so what is the problem

area that you you know dealt with so the

currently the problem in the industry is

that uh you know in industrial settings

there's a deadly threat that is gas

leakage and you know Industries uses

hazardous substances uh that chemicals

that are when released in air can cause

significant harm to people there were

many cases in past like Bihar and

uh in other countries as well where the

gas leaked overnight and people couldn't

even know and by the morning that caused

a lot of deaths so the industry owners

Factory people the managers they even

didn't didn't know about the gas leak

that happened and people and media got

to know way after it was too late for

the you know ambulances and

Healthcare uh facilities to be informed

so we have created a solution

that targets on fixing this and

spreading the information very fast so

that before something happens like gas

leakage or fires we can notify people

and the factory owners that has happened

indeed that is certainly a very very

critical area and who would have

forgotten Bal gas strategy back then

that tragedy was like you know that

shook the nation and I wish we had

something like the solution that you

have come up with and uh it certainly is

going to be a pathbreaking solution I'm

sure so uh please tell us about you know

how did you start working on this

particular solution and what impact do

you think is going to make we had the PS

155 which targets or mean specifies on

aial Hazard we were thinking on how we

can provide the solution which can save

life our main motto was to save life you

know it's better to uh prevent fire than

fighting fire so our main motto was to

prevent the gas leakage from happening

most of the factories already have

sensors and all installed but they are

not digitized so we have provided a

software solution to digitize those

application and provide the data to the

industry owners real time so they can

know even a slightly change in data like

gas leakage they will be notified in the

app and they can instantly go and fix it

and stop it so that is the main part

that we have we are preventing before it

is occuring but we have also created a

machine learning model which predicts

the gas leakage you know when a gas

leaks in the air it spreads in the

entire area we need a proper machine

learning model to predict for in which

direction and up to how far that gas

leak will occur so we have made a list

of factors that determines how much the

gas will leak like temperature humidity

the landscape of the area it is a fact

you know open land or it is in between

multiple buildings so we have counted

all those factors and created a machine

learning model that predicts the Zone

Danger Zone in which the gas will be

leak so people can get notified of that

area and stay away from that area okay

great that sounds really wonderful and

uh in the era of you know Bal gas

tragedy this would have certainly made a

lot of difference so now tell me how

does it really work so when a gas is

leaked the sensor triggers a uh you know

the sensor in in each suppose a gas

cylinder is there it has sensors

installed that continuously checks the

pressure in the cylinder and also have

the data which kind of gas is this and

its density and all so whenever a gas is

leaked it triggers a API to the out to

our back end and sends the data about

how much gas is leaked at what rate the

gas is leaking and the all information

about the gas and in the back end we

have a machine learning server which

python server which predicts from the

data that what gas is leaked and

according to that gas leak we predict

the uh danger zone which is around that

area Okay once that leakage is detected

and your it comes to your system then

what what happens next like do you you

know put this up this information

somewhere or a person needs to access

your platform to get this information

how does it work after that basically

once we get the notification that the

gas is leaked in our database or in our

system then first step we will do we

will predict the area and we will map

that area to the user application which

is to all the like industry workers

industry employeer manager and all also

apart from this whenever it gets sleak

in certain amount of area so we will be

also collaborating with the local

network provider like ATL PS and all so

what we will do we will notify them by

means of like SMS or some kind of

messaging app so that the common people

also can get to know that this happens

here okay so does one need to subscribe

to this kind of a thing or how is it

available to Common masses or the

industrial no there is no kind of

subscription and all and it is open

source our application will be available

on Play Store everyone can like download

that and apart from the application also

if you don't have if you do not have

access to our application you can also

get benefit of this because we will be

collaborating as I already mentioned we

will be collaborating with the local

area provider local network provider so

if some kind of trasy some kind of has

that happens to your area location so

you will be notified so here is the

application that is available to

everyone from Play store or iOS store

they can download it so once they

download it they have a page for sign up

so they can enter their details and

according to their area they will get a

industry ID so if they are living in

bopal in certain area the each and every

Industries will have their own IDs so

when they sign up they will enter the

industry ID here

so once the sign up is complete we also

take the skills of that people so in

case of suppose a fire has occurred and

healthcare service cannot reach on time

so we can contact these people and based

on their skills we can get help from

local people as

well so here is the application so it is

currently showing that a gas is leaked

in this area and this is the Red Zone

which is the danger zone and this orange

zone is the means it is kind of Hazard

Zone but outside that that is the safe

Zone basically our Landing model gives

us two part of areas one which is like

intensely affected like 100% affected

and second area is like it might be

affected or not might be affected it's

like

5050 I guess every gas leakage would

have a different level of potency so is

the application also able to manage that

a certain gas will be you know less

deadly than the other one so your hazard

areas takes that into

account

yeah that's what we are like working on

right now because we know is gas is like

it's on it's own weight it's on like

density and all so for that kind of

reason we are working on that we are

preparing our machine learning model to

work on like each and every individual

guas part so that uh in future we will

be implementing that with the

application you are trying to update a

certain feature in there so what else

are you like you know working on right

now to make this even

better yeah so right now we are like

Gathering a team for like some chemist

peoples so those peoples will help us to

understand understand more about the

guesses the factors which we are not

aware about the gases they will help us

to like know about the GES and they will

help us to prepare our data set so after

preparing the data set we will be

implementing that data set in our

machine learning model and improving it

basically right now we are like

constructing our own data set or why we

are testing different different gases we

are manually spreading them we are

manually putting the like temperature

manually putting the wind speed and all

and like testing different kind of

scenarios and preparing the data s so

once that part of knowledge comes to you

once you know you have a better

understanding of gases and how these

operate then you'll make this

application even more robust and at that

point probably you would want to you

know Implement that in the actual field

conditions am I right yes sir there are

several other features as well in the

app like if you are the manager of the

factory and you are not in the factory

so you can open the app and check your

emergency contact list and contact the

Emergency Services fast from the app on

directly also our app directly contacts

to the hospitals fire office and police

station so whenever Hazard situation

occurred app already sends the

notification to each and every you know

healthcare services and all also in the

app there the managers can see the

realtime sensor status like all the

sensors are working correctly or not and

the pressure of each gesses and

everything these data is updated

like they can control the sensors as

well from the app like if they want the

doors closed or ventilation open they

can directly do that from the app

so these sensors that we are talking

about they are the ones which are

already existing in the any industrial

setup right or are you putting planning

to put your own sensors there which are

compatible with your

application very common in the uh you

know in industry but they are all

working in offline mode they are not

digitized so we are just digitizing

those sensor and putting the data on the

application and controlling the sensors

from the

application so basically you'll be using

the existing sensors which are there in

the industrial setup already right yes

sir okay so if I may ask how much more

time do you need to you know bring this

application or bring this you know

solution to the real world scenario I

think uh hard to like give you like like

appro time Al because we are like quite

slow in the process because you need to

understand that dealing with the guesses

like it's little bit like more complex

task the most important part of the

application is the machine learning

model and that model needs a data set

and so creating that data set is the

time-taking part overall the application

is ready so to make the this Hazard Zone

more accurate we need the machine

learning model to be better okay so in

the end I would like to ask you what

impact do you expect your application to

make in the real world scenario when you

actually implement it on ground so when

the application is uh you know the live

in live it will help more mostly in

preventing any situation from occur our

main motto was that so that this never

happens such like bopal gas strategy it

should have never happened if the

application was existing people could

have got notified about the hazard that

has occurred and they could have run

away from the area so that is the main

part we will be notifying and spreading

the information very fast you know

through social media and uh

notifications and alerts yeah of course

uh rightly said I mean if you're able to

avoid those kind of tragedies there

can't be anything better which can be

given to the as a gift to the industrial

world so great guys it was really nice

talking to you and understanding the

solution that you provided for

industrialized setup of the country and

I'm sure it's going to immensely help

the infrastructure that we have built

over a period of time and which is

helping the economy of the country and

you're trying to protect that very

infrastructure wonderful I wish you all

the best and I wish that this particular

application this particular solution

sees light of the day very very soon if

not much later thank you so much for

joining us today thank and we wish you

all the best in an era where

technological advancements are shaping

Industries this team has demonstrated a

profound understanding of the critical

importance of fire safety in industrial

settings their winning project is a

testament to their Ingenuity and

commitment to addressing the unique

challenges associated with industrial

safety we will meet another ingenious

team in the coming Edition watch out for

this

space