Design Of Indoor Environment Detecting Aircraft Based On Multi-sensor Fusion

Recently,due to the flexible 3d space mobility,unmanned aerial vehicles have attracted more and more attention.The micro quadrotor with advantages of simple structure,strong mobility and the performance of vertical take-off and landing,is widely used in field of investigation and rescue in indoor and outdoor environment.Given that the terrain of the unknown environment has greatly limited the mobile ability of ground robots,an indoor detecting aircraft have been designed,which fuses the information of accelerometer,gyroscope,electronic compass and camera.First of all,the dynamics model of the quadrotor is built by the Newton-Euler equation in this thesis.Furthermore,the cascade PID controller is designed for position and attitude control.And the visual position estimation system is composed of the embedded processing platform Jetson TX1 and the front view of camera.The reliability of the system is demonstrated through the actual flight test.The airborne inertial navigation system cannot provide accurate position information due to the long-term position estimation error.Aiming at this problem,a positioning and navigation strategy based on simultaneous localization and mapping(SLAM)and inertial navigation system is proposed.It mainly includes using respectively DSO(Direct Sparse Odometry)algorithm and complementary filtering algorithm to estimate the information of position and attitude.And then choosing extended kalman filter algorithm to fuse the position information and attitude information.Thus an indoor detecting aircraft is realized.To estimate the accurate attitude information,the data of the gyroscope is modified by the airborne accelerometer and the electronic compass through the complementary filtering algorithm.And then to realize the autonomous environment detection by the aircraft,using extended kalman filter algorithm to achieve the fusion of attitude information and visual position information.Finally,the experimental test shows that the indoor environment detecting aircraft can achieve autonomous flight and scene map construction in a specific environment.