Rescue Bot

Natural disasters have been a major problem for humanity and cause of majority of deaths in the past few centuries. Just in the last 2 decades about 7348 disasters were recorded worldwide causing 1.23 million deaths. Governments across the globe have implemented various methods and strategies to solve this issue bus all of them are limited by one way or another i.e. being costly or having low efficiency or being slow to implement. Search and Rescue (SAR) operations have primarily been conducted using the combined efforts of police and relief squads, but with recent development in Artificial intelligence (AI) and deep learning, disaster management have become more efficient and fast in saving lives. Use of unmanned aerial vehicle (UAV) has helped the operatives to scan the terrain and locate disaster struck people. This paper presents a solution to help SAR operatives by combining both the deep leaning technology and bot technology such as buggy and UAVs. With the use of microprocessors such as raspberry pi or orange pi, the Machine Learning (ML) model is connected to the buggy via Wi-Fi. Using this technique a low cost solution has been achieved with a mean average precision (mAP) of 30.3 has been achieved. Combining this system with Thermal Imaging Camera (TIC), temperature readings are recorded allowing the operator to detect objects under rubble by looking at abnormal temperature readings