Every year, an Inter-hall Hardware modelling competition is held as a part of the General Championship at IIT Kharagpur. The competition has an open problem statement, so any hardware with usability can be prototyped and presented. It was the first time that our hall MMM was participating in this competition, and I was given the responsibility to lead the team of ten students. We found out some astonishing figures regarding the casualty due to fire accidents. Considering the problem, we came up with a product with great potential to replace a human fire-fighter. It was nicknamed as SAFFV(Semi-Autonomous Fire Fighting Vehicle). Following are the major functionalities of the robot-
- Capable of all-terrain traversal with the help of skid steer drive consisting of caterpillar mechanism similar to tanks. Angled front and back extension of the belt provides the ability to traverse on debris and rubble, giving the bot an all-terrain traversal ability.
- Live thermal imaging and vision feed, which can be accessed with the help of Wifi. The thermal camera feed can be easily used to find humans, even in the presence of thick smoke.
- Audio feed and audio communication through 2.4GHz Radiofrequency
- Full Wireless control of the robot over 2.4Ghz Radio wave frequency
- Provision of continuous water supply of 10000 cc/min with the ability to increase it as per requirement. Secondary CO2 based fire extinguisher (2kg) to extinguish small fires at a rapid rate.
- Safety box containing gas mask, helmet, and other fire safety equipment for the survivors.
I worked on all aspects of the robot including the mechanical design and manufacturing, electronic architecture and the software interface. The whole robot works on the Robot Operating System (ROS) platform. The Arduino Mega and the Raspberry PI are the significant components. The input from various sensors like gas sensors, encoder and PIR sensor is read from the Arduino and sent to the Raspberry PI via a topic. This data is further hosted on a local server, which can be accessed by a client device.
The joystick is used to send commands for various motors and servos as well as for triggering the fire extinguishing mechanism. The data received from the joystick by the Raspberry PI is processed and sent back to the Arduino as commands. The process can be understood in detail from the figure given below-
The FLIR Lepton 3.5 is the most important part of the whole robot. It is capable of producing thermal images of size 160 x 120 at an impressive rate of more than 8 fps. This colourful thermal image can be easily used to point out the fire and also spot humans in the presence of thick smoke. The thermal camera feed is processed and hosted on a local web server which can be streamed from a client device. This camera is mounted on the rotating platform which allows human detection in all directions. Due to some constraints, the FLIR Lepton 3.5 thermal imaging camera has not been used. Instead, an IR camera with night vision has been used for the prototype model.
A Logitech C-270 webcam is mounted on the front side of the robot. Its feed can also be streamed simultaneously along with the sensor data and the thermal camera on a client device. Its major use is to get a vision in the direction of motion of the robot so that the bot can be manoeuvred without hitting obstacles.
Speakers and USB microphone are attached to the Raspberry Pi. The microphone records live audio and stream it over the local server. Similarly, on the other side, the person controlling the robot can send voice commands to the robot which are amplified using speakers. They can assist the survivor in accessing the safety equipment box for a gas mask and other equipment.
Though we didn't finish on the podium, it was a great experience to lead the bunch of talented team members. The full overview of the robot can be found in the design report attached below. In addition to that, the presentation has also been uploaded. Some of the software packages used can be found at my Github page.