Research On The Integrated Navigation And Terrain Following Of Quadrotor Based On Multi-sensor Fusion

Quadrotor is a kind of light weight,low cost unmanned aerial vehicle,which has been widely used in aerial photography,power patrol and inspection and crop protection.However,due to limited payload,quadrotor cannot carry high precision navigation equipment,which leads to the use of a single navigation device,such as GNSS receiver,vulnerable to signal occlusion or interference when it performs an investigation task in a complex terrain environment.In addition,when flying in the mountains or between trees,quadrotor is more likely to be affected by wind disturbance,which can especially cause altitude control delay and bad real-time.Therefore,improving the positioning accuracy and flying stability of quadrotor in unknown environment is of great significance and value for expanding its further applications.In this paper,an open source flight controller has been utilized as developing platform to perform research on the integrated navigation and terrain following based on multi-sensor fusion.The paper has been arranged as following order:(1)Dynamic model and simulation: according to the kinematic and dynamic characteristics of quadrotor,a twelve variables state equation is derived and the linearization of this mathematical model is carried out.Then the structure and dynamic parameters of a 450 mm wheelbase quadrotor are measured using the UAV mechanical testing devices.At last,a double loop control system of quadrotor is designed and simulated in Simulink.The simulation results show that the derived mathematical model corresponds to the characteristics of quadrotor.(2)Research on multi-sensor fusion: the output characteristics of multiple sensors are introduced and modeled.A 2-class complementary filtering structure is designed for the limited computation resource of flight controller.In the first class,the data of accelerometer and magnetometer are fused to compensate for gyroscope drift during attitude estimation process.Two kinds of complementary filters are introduced and compared in both static and dynamic situation with the data collected from IMU.In the second class,the velocity/position estimation is mainly constructed by complementary filter combining multiple sensors such as GNSS receiver,optical flow sensor and sonar rangefinder.Extensive experimental results demonstrate the effects of different fusion combinations.(3)Terrain following implementation and altitude control improvement: the path planning algorithm has been optimized to improve the accuracy of waypoint travelling.In order to solve the problem of control delay and instability under altitude hold flight mode,a cascade PID is proposed to control not only altitude but also vertical velocity.Simulation results acquired in Simulink demonstrates that the cascade PID can considerably improve the dynamic performance of altitude control.Further,the experiment performed on the quadrotor also shows that cascade PID has better real-time performance and stability than PID.