Team Shinobi| Spot That Fire V3.0
Spot That Fire V3.0
SHINOBI-Semi-autonomous Fire Detection Robot by Deep learning and Terrai Adaptive suspension System.
Summary
Shinobi detects wildfire in its initial stage with a fire detection technique that spots fire using image processing. It gives a real-time view that the bot observes via IP webcam and transmits it to the controller operating it who is viewing it through VR. Once verified, it sends an alert by voice command to the nearest help. The PID based terrain adaptive suspension system helps the bot to stabilize itself on uneven surfaces of the forest. The bot has a temperature sensor that stops it to go further if the temperature exceeds the set limit and in the worst situation, the bot has the functionality of protecting itself with help of a cannon system that covers the bot with a layer of CO2.
How We Addressed This Challenge
We have developed “Shinobi” – a semi-autonomous bot that tackles and solves the problem of forest fire by detecting it at a very early stage and thus sends an alert to the nearest help who further can take necessary actions to prevent the worst disaster.
Wildfire has affected hectares of forest land, risked the lives of hundreds of species living there which has an adverse effect on our eco-system. The only way to prevent it is by sniffing the first smoke.
Shinobi exactly does that.
It detects fire using deep learning image processing method and sends a voice alert in case of fire detection using an AI speech recognition technique after verification AI boat send an email to the nearest help using python packages once verified by the controller. It has a PID based terrain adaptive suspension system which provides extra stability for the uneven surface of forests while tracing for fire. The self-protection system helps it to stop if the temperature exceeds the limit of the bot and to save itself it uses a cannon which throws CO2 balls that forms a layer of CO2 around it so that it suffers from minimum damage.
Our solution successfully solves the challenge of spotting that fire.
How We Developed This Project
The problem of wildfire is the worst disaster our eco-system is facing and the impact is also huge. Last year 30000 forest fires were reported in India alone which doesn’t really get global attention. Lives of animals are lost along with their habitat and 95% cause of this is anthropogenic.
All this could be prevented by detecting and predicting forest fire. We were struck by the idea of creating Shinobi - a ninja warrior bot and protect our eco-system from wildfires.
Our approach was to list down the problems and try to solve it using technologies at our disposal.
To develop the project we used:
Hardware: For the bot, we have used the following-
1. 4 Johnson geared motors (100 rpm)
2. 4 wheels (dia- 100mm, width- 2 inches)
3. 4 servomotors
4. Arduino UNO
5. Receiver & transmitter flysky i6 ( 10 channel)
6. Lipo battery 2200 mah (25c)
7. VR (handmade using cardboard, 100mm focal length lenses)
8. Temperature sensor and GPS module
Software: Used Python programming language. Packages are mentioned below.
1. Deep learning packages using Open CV i.e. an open-source computer vision and ML software library along with NumPy library.
2. Speech recognition package for making audio signal meaningful using sampling, artificial neural networks & ML. Also used pyaudio & pyttsx3 package to get our final result of sending voice alert.
3. Sending of mail is achieved using packages like smtlib (sending email between servers), os (the modules which helps to interact with the operating system), webbrowser (display web-based documents to the user)
4. The datetime module gives date and time.
We initially faced the problem of low efficiency of tracing the fire which we solved using more efficient image processing models. Our next major problem was to inculcate immense stability for our bot to work effectively on the uneven terrain of forests that lead us to build PID based Terrain Adaptive Suspension System. We initially used emails as an alert which we improvised to speech recognition. Another factor of using a laptop for controlling was later changed to VR for providing a real-time and real- view to the controller.
Web presence: https://smiritisharma17smi.wixsite.com/mysite-1
Instagram: https://www.instagram.com/team_shinobi_nasa/?hl=en
How We Used Space Agency Data in This Project
We can use nasa data and his outcomes with our real time data to check the accuracy of our prediction. But for the Time being we are using real time data collected through camera in real time.
Data & Resources
All the real time data has been used