Fires threaten lives, property, and the environment. Fighting fire and rescuing those trapped by a fire is risky but essential. New research in the International Journal of Advanced Mechatronic Systems, looks at how autonomous wheeled rescue robots might be used in fire rescue. The work involves the development of a novel path-planning algorithm with an advanced fire-recognition system for indoor fires. The system offers significant benefits over conventional map-based approaches and finds the shortest and safest escape route for those trapped by the fire.
Shaun Q.Y. Tan, V.J. Karthik, A. Govind, and P.M. Rajasree of the Electronics and Instrumentation Department at RV College of Engineering in Bengaluru, India, tackled the challenge of fire recognition and localization using convolutional neural networks (CNNs) along with various image processing techniques. To optimize their CNN model effectively, they employed the particle swarm optimization (PSO) approach.
The team reports that their CNN model outperformed the MobileNet architecture, demonstrating exceptional recognition and localization accuracy on simulated indoor fire scenarios. Future work will lower the computing demands of the system as well as boosting accuracy. Indeed, a more refined model will be vital to ensure the practicality and efficacy of unmanned rescue robots in real-world emergencies. The system also be adapted to other kinds of emergencies.
The system has a limitation in that the algorithm cannot take into account potential obstacles in the path of the robot. Future improvements would incorporate local optimization techniques, ensuring seamless navigation even in complex indoor environments. With increased accuracy, adaptability, and practicality, the technology promises to become a powerful asset in safeguarding lives and property in the face of unpredictable fire emergencies.
Tan, S.Q.Y., Karthik, V.J., Govind, A. and Rajasree, P.M. (2023) ‘An approach into navigation and vision for autonomous fire fighting robots‘, Int. J. Advanced Mechatronic Systems, Vol. 10, No. 3, pp.156-164.