FireCop
FireCop was born as a joke when we saw the ads of the Amazing Robot Competition held in Helsinki in 2022. Although we were late for the registration, we got admitted to the competition and we started implementing FireCop. Despite building this robot in two weeks, my team managed to perform well and we got second place.
​
FireCop is an autonomous firefighting robot that uses Machine Vision, multiple sensor fusion techniques, and Artificial Intelligence to recognize potential hazards and quickly react in the event of a fire. While in Patrolling mode, it continues to seek for any potential risks until it discovers any and then changes to Alert mode, where it checks for any signs of fire. In the event of any fire, it switches into extinguishing mode alerting everyone about it and extinguishing the fire.The robot is developed with the aim of automatically detecting the fire and swinging into action without any human intervention thus not risking the lives of firemen.
​
Technical implementation
It uses the Intel Realsense RGBD camera to capture the surroundings along with the depth and uses Computer vision techniques to identify possible hazards. Also, using the data from the infrared sensors and with IMU and wheel odometry, it navigates towards the potential hazard and check for any sign of fire. If the fire is detected, we extinguish it by blowing air with the help of a DC motor and a pump.
Fully ROS based, the brain of this robot is a Nvidia Jetson Nano, running an Ubuntu20 image. Since Nvidia does not provide maintened images of Ubuntu 20 on old boards, we had to find many workarounds for running all our ROS nodes, and this was a great learning experience.