Firefighting robot with deep learning and machine vision

While extinguishing the fire, firefighters find it difficult to reach certain areas due to narrow spaces or debris blocking the way. In urban cities and industrial areas, there is a constant need to have firefighters ready in case of emergencies. This can lead to a shortage of manpower. Thus, the firefighting robot can act as assisting support for firefighters and will also lower down the risk of their life. Even though many firefighter robots have been developed currently to overcome this problem, these robots are expensive and difficult to maintain. We propose an intelligent robot that uses deep learning to not only detect and classify fire but also extinguish the detected fire based on its class. The proposed firefighter robot is cheaper, autonomous, and easier to maintain. We have used a combination of AlexNet to detect fire and ImageNet for detecting the type of fire. We achieved a classification accuracy of fire detection up to 98.25%, and the classification accuracy of fire-type classification was around 92%. The firefighter robot can be deployed in places that are hard to reach for the firefighters and thereby reduce the burden on firefighters.

Automated summary

The traditional firefighting work is hindered by space limitations and obstacles, requiring the use of firefighting robots for assistance. Existing firefighting robots are expensive and difficult to maintain, so we propose an intelligent robot that uses deep learning, which is cheaper, autonomous, and easier to maintain.