Rotor Unmanned Aerial Vehicle(UAVs) Indoor Intelligent Search And Rescue System Based On Multi-Sensor Data Fusion

Earthquakes,fires,mine disasters,and other accidents have brought great threats to people's lives and property.The accident scenes caused by such disasters often take place in narrow,closed indoor environments,such as tunnels,high-rises,shopping malls,underground,etc.It is difficult to carry out rescue work effectively because of the unclear indoor environment and inability to locate the victims.With the development of science and technology,unmanned aerial vehicle(UAV)technology has been a representative technology in the new era with its applications in military,agricultural,production,inspection,and other fields.The UAVs have the characteristics of small size,high degree of freedom,and vertical taking off and landing.This thesis designs an indoor search and rescue system based on quad-rotor UAVs(QRUAVs),including control navigation system,target recognition system,and information transmission system.The navigation in the indoor environment of the QRUAVs and missions of search and rescue are achieved via the combination of the UAV indoor navigation technology and the image processing technology.As for the control-navigation system,the QRUAV flight platform is built,whose mathematical model parameters are also identified by CIFER software.Then,based on the identified model,a backstepping controller is designed for the flight control.Furthermore,an indoor intelligent navigation algorithm is realized based on an RBF neural network prediction compensation scheme.The multi-source data fusion is conducted with the two-dimensional laser scanners,optical flow sensors,and inertial navigation sensors.Taking the advantages of the multi-agent-based orbit feature extraction technology and the YOLOv3 image target recognition detection algorithm,the target detection system is then formulated,which enables the QRUAVs to navigate autonomously in the indoor environment and identify and detect the victims,and complete the search and rescue missions.At the end of this thesis,fixed-point and trajectory experiments in indoor environment are launched with the QRUAVs.Meanwhile,the target classification and detection experiments are also completed by using the testing-field photo data.The experimental results show that the search and rescue system proposed in this thesis has certain feasibility and can complete the expected tasks.