Research On Multi-sensor Based UAV Obstacle Avoidance Technique

There has been witnessed the rapid development of UAV technology and the wide application of UAVs in daily work and life.Currently,the manipulation of UAVs is commonly depending on the expertises of the operational personnels,and there’s lacking of the environmental perception capability and autonomous flight ability of the UAVs.To further improve the environmental perception capability and expand the application areas,UAVs are equipped with multiple kinds of sensors.Therefore,the research on the multi-sensor-based UAV localization and obstacle avoidance technology is of great theoretical significance and practical importance.In this dissertation,we first summarize the development and the current status of UAV localization and obstacle avoidance technologies.In order to improve the environmental perception capability of UAV,a UAV platform with multi-sensor such as vision,laser,ultrasonic and inertia sensors is designed and implemented.We propose a multi-sensor data-fusion method for UAV localization and mapping.To overcome several shortcomings of optical sensors,an auxiliary ultrasonic-based obstacle avoidance subsystem is designed by which the flight safety in both indoor and outdoor environment can be improved.At the same time,to improve the effectiveness and efficiency of UAV tests,a multi-sensor-based simulation platform is designed which provides certain feasibility to simulate the UAV flight algorithms.The main contribution of the dissertaion is summarized as follows:1,By reviewing the development of UAV technology,the current status of UAV localization and obstacle avoidance technology both at home and abroad is provided.2,The existing results regarding multiple kinds of sensors,such as...are compared and analyzed.Also,a UAV platform is designed with stereo camera,optical flow sensor and lidar as its main sensor.3,Based on the analysis of visual SLAM algorithm(ORB-SLAM)and the laser SLAM algorithm(Cartographer),a modified laser SLAM algorithm is proposed.The algorithm realizes the data-fusion regarding the optical flow sensor and stereo camera sensor,and it improves the accuracy and robustness of localization and mapping.4,To correct the defects of optical sensor in obstacle detection scenario,an auxiliary ultrasonic-sensors-based system is implemented.The system consists of two modules.The forward ultrasonic module provides certain detection range,it also provides the stabilization ability of the pitch detection angle by fusion IMU data.The central signal processing module filters the ultrasonic sensor data and sends the barrier signal to the flight controller to achieve obstacle avoidance.5,A multi-sensor-based UAV simulation platform/testbed is implemented to improve the effectiveness and efficiency of the UAV test.Particularly,within the simulation platform,not only the IMU,camera,lidar and other sensors modules can be tested,the localization and obstacle avoidance algorithms can be tested through real tasks settings as well.