Design Of Autonomous Investigation UAV System For Tunnel

At present,some tunnels in our country have entered the aging period,which requires regular inspection and maintenance.Lack of proper maintenance may lead to serious consequences.Aiming at the problems of large workload and high danger coefficient in the traditional way of tunnel inspection and considering the strong maneuverability of micro-rotor UAV,a tunnel autonomous investigation UAV for tunnel environments is designed,which integrates the information from the inertial measurement unit(IMU),visual sensors,LIDAR and other sensors.Firstly,the UAV system which can independently complete the tunnel investigation task is constructed.According to the characteristics of tunnel environment and the load capacity of UAV,camera,lidar and embedded processing platform ODROID-XU4 are selected as navigation unit,the tunnel environment investigation unit is designed with FLIR TAU2-336 infrared camera,temperature and humidity sensor and gas sensor,the software architecture and workflow of UAV system for tunnel investigation are designed.Secondly,aiming at the problem that conventional SLAM based laser is easy to lose stability along the tunnel axis,a location method based on the fusion of Unscented Kalman Filter(UKF)SLAM based laser and visual odometer is designed.In the case of unsatisfactory illumination in tunnel,airborne array LED is used to compensate illumination.the uneven illumination and lens glare caused by compensating illumination make the image produce abnormal intensity gradient,which has adverse effects on image feature extraction.Aiming at the adverse effects of abnormal intensity gradient on image feature extraction,a set of low-order image filters are designed to amplify the weak texture to ensure the accuracy of visual odometer.Thirdly,during the investigation process,the UAV is required to fly autonomously,therefore,two kinds of navigation control modes are designed,which includes the semi-autonomous navigation mode and fully autonomous navigation mode.In the semi-autonomous navigation mode,the UAV autonomously estimates the positions through the navigation unit and the speed and direction of the UAV are provided by the operator with the help of a hand-held remote controller or a ground station.In the mode of fully autonomous navigation,after the UAV acquires the start command,the flight path is planned on the two-dimensional grid map through the navigation unit and the UAV realize autonomous reconnaissance along the planned flight path.Finally,in the simulated tunnel environment and the real tunnel environment,the autonomous positioning method,navigation strategy and tunnel investigation method of the tunnel autonomous investigation UAV system are tested experimentally,the experimental results show that the system has certain autonomous inspection capability.