Research On Multi-sensor Data Fusion Based Indoor Navigation And Control For Rotor UAVs

As the aerial robot,unmanned aerial vehicle(UAV)is a complex interdisciplinary system which involves machinery,electrical theory,materials science,aerodynamics,navigation,control and many other subjects.The same with other mobile robots,navigation and control are the core issues in UAVs research area.This paper carried out the related research on these two issues.In the aspect of the navigation,intellectualization and miniaturization in UAV development require UAVs to realize accurate,reliable low latitude and indoor navigation with as many detection means as possible.However,single sensor is hard to have a full range of perception.Utilizing date fusion technology to integrate multi-sensor data can make it easier,meanwhile,improve accuracy and robustness of the UAV navigation and lower the performance requirement of single sensor.This paper utilized Kalman filter(KF)and extended Kalman filter(EKF)as the navigating data fusion algorithm,gyroscope,accelerometer,optical flow sensor and other multi-sensor data as the information sources,to establish the indoor navigation system function,which indicated the roles of each sensor data in the system.This is the critical design step from the theory to an actual filter.Based on that,the whole indoor integrated navigation system was implemented.During implementation,timestamp technique,asynchronous sampling sequential fusion and other methods were applied to fix the space and time matching problems occuring in multi-sensor data sources situation.After verified by Simulink platform,the indoor integrated navigation system was applied to a quadrotor whose core combines with ARM and FPGA,to achieve the low altitude,indoor navigation.In the aspect of the control,along with the rapid development of rotor UAVs in recent years,many related control algorithms have been proposed.However,there is little research on control mode which is the expression of attitude,speed,and position used to realize the control algorithms.Since the rotor UAVs don't have the limitations coming from human,the maneuver which manned crafts cannot try will be finished well by rotor UAVs.The traditional rotor UAV control modes perform the manned crafts control style,such as keeping the UAVs nose to constraint,etc.After control mode optimization,the constraint can be released and the maneuverability advantage of the rotor UAVs can be fully utilized.This paper proposed a kind of desired thrust vectoring fast approach method based on the equivalent rotation vector of quaternion,which can be used for the rotor UAVs to optimize the control mode when executing trajectory tracking task.Based on the quadrotor kinematics and dynamics model established by Newtom-Euler function,the paper took a common rotor UAV attitude-position controller based on backstepping for example,analyzed its disadvantages in control mode aspect and proposed the fast approach method above.The minimum rotation vector and the minimum attitude change quaternion from current thrust vectoring tothe desired can be obtained by this method,then the desired attitude quaternion can be acquired which is provided for the attitude control loop to realize the desired thrust vectoring fast approaching.The paper introduced the theory and the brief geometric proof of the fast approach method.Then,we used this method to design a entire trajectory tracking control strategy for the quadrotor.Analysis has been done in Simulink,verified the effectiveness of optimization in control mode by the fast approach method.At last,the paper concluded the work about UAVs navigation and control,and analyzed the research tendency in this area.