Integrated UWB-vision Approach For Autonomous Landing Of UAVs

With the current increasing demand for surveillance and search and rescue missions in complex environments affected by different types of hazards,UAVs are receiving more and more attention as their aerial perspective is very useful for autonomous navigation or human remote operations.However,as UAVs are constrained by endurance issues,they need to be returned to recharge or change batteries frequently,so UAVs（unmanned aerial vehicles）in partnership with unmanned ground vehicles（UGVs）can provide valuable insight into complex environments.In order to successfully execute their missions,multi-robot systems,for example,require UAVs to take off autonomously from a moving UGV,execute the mission,and return to the landing platform once the mission has been executed.To address the mission requirements in this context,this thesis focuses on the design of a distance-based long-range docking algorithm for position-unknown targets and a relative position estimation algorithm based on multi-sensor fusion and a landing control algorithm based on an altitude adaptive PID controller.The thesis covers the following three main areas.（1）To address the problem of long-range docking of unmanned aerial vehicles（UAVs）with unknown landing target positions,and to solve this problem,this paper proposes an integrated positioning navigation scheme using ultra-wideband（UWB）sensors and optical flow sensors as the primary sensors to simultaneously implement the relative positioning and navigation tasks of discrete-time integrators under bounded velocity control,including a non-linear adaptive estimation scheme for estimating the relative position of the target,and a specially designed velocity control scheme to ensure convergence of the estimation and to enable asymptotic docking of targets with unknown long-range positions.（2）A multi-sensor fusion estimation scheme based on ultra-wideband and a visual reference system is proposed for the relative position estimation problem of UAVs when landing on moving targets.Using the traditional visual detection method,the algorithm is susceptible to detection failure due to light,shadows and visual blur caused by the drastic motion of the UAV,and thus cannot complete a robust all-weather position estimation using visual detection alone.The estimation of the relative position of the landing target is improved by the generalized fusion of the range values of the ultra-wideband sensors and the target position change information transmitted via ultra-wideband technology.（3）A highly adaptive PID control scheme is introduced for the landing of a moving target platform.In the process of landing a moving target platform,the landing success rate is influenced by the speed of the target.In practice,the response speed required by the UAV varies at different altitudes and for approaching a landing platform the UAV needs to respond more quickly,which is not possible using traditional fixed gain PID control.In order to improve the success rate of landing,the combination of an altitude adaptive PID controller with a Kalman filter and the landing platform transmitting its own velocity and position information via ultrawideband technology is used as an input to the PID controller to improve the success rate of landing.Finally,through Matlab simulation and Gazebo simulation,the fully autonomous landing of the UAV on a moving target is achieved,which verifies the effectiveness of the algorithm developed in this thesis.The system proposed in this thesis can also be deployed on a real robot platform to complete the autonomous landing task.