A Novel Integrated Navigation Method based on the Dynamic Model for Quadrotors
Pin Lyu, Jizhou Lai, Jianye Liu, Ling Zhang, Shichao Liu, Nanjing University of Aeronautics and Astronautics, China
Quadrotors have been widely used in some monitoring and operation tasks, such as aerial photography, traffic control, mineral exploitation, power line inspection, pesticide spraying and so on, thanks to their simplicity of operation and hovering and vertical take-off and landing (VTOL) capability.
A commonly used navigation system for quadrotors consists of inertial measurement unit (IMU), GPS, magnetic sensor and barometer. The above sensors are fused together through certain algorithm and can output accurate attitude, velocity and position in a normal case. However, due to the size, weight, power and cost limitation, those sensors are not reliable in some special situations. For example, IMU performance can be affected by the temperature or the quadrotor vibration, GPS can’t be used in indoor environment and its accuracy decreases when the quadrotor flies besides the building, magnetic sensor accuracy can be easily affected by the magnetic interference, barometer accuracy is affected by the wind. In the above conditions, the integrated navigation system may output wrong navigation information, resulting in unstable flight even crash. Therefore, fault tolerance navigation method is needed to provide reliable navigation information.
Nowadays, many fault tolerance navigation methods have been proposed to improve the flight safety. Those methods mainly concern about the faults of GPS, magnetic sensor, and barometer. Some alternate sensors have been studied, such as the optical flow sensor, visual sensor, motion capture system, sonar, laser radar and so on. Those sensors can output velocity, attitude and position and be treated as the redundancy information of GPS, magnetic sensor, and barometer. However, few works consider the IMU fault. One of the main reasons is that there lack the alternate sensors providing redundancy information of IMU, which are rate angular and acceleration.
In this paper, a novel navigation scheme for quadrotors is designed. In the scheme, a dynamic model/ GPS/ magnetic sensor/ barometer integrated navigation system is constructed. Compared with the traditional scheme, the IMU is replaced by the dynamic model in the proposed scheme. The dynamic model of the quadrotor is used as a virtual sensor, outputting the angular acceleration and velocity acceleration. In this paper, we want to investigate whether the dynamic model can perform as an alternate sensor of the IMU. It is a preceding work of the IMU fault tolerance navigation for quadrotors.
The dynamic model describes the relations between the forces and moments of the vehicle and its control inputs. As for a quadrotor, the main forces are the thrust and drag, the main moment is the torque, and the control inputs are the angular velocities of the four blades. Based on the dynamic model of the quadrotor, the thrust, drag and torque can be calculated from the angular velocities of the four blades, which can be measured through the electronic speed controllers (ESCs). According to moment of momentum theorem and Newton second law, the angular acceleration and velocity acceleration can be calculated from the forces and moments. The calculated angular acceleration and velocity acceleration can be treated as the alternate information of the IMU.
The dynamic model is fused with GPS, magnetic sensor, and barometer through Kalman filter. The state equation is established based on the kinematical equation of the quadrotor. The angular rate, attitude, position, velocity are chosen as the state variables, the angular velocities of the four blades are the system inputs. The heading, height, velocity and position are chosen as the measurement variables.
The proposed method is verified through simulation. The quadrotor dynamic model and the onboard sensors are simulated. The proposed dynamic model/ GPS/ magnetic sensor/ barometer integrated navigation scheme is compared with the traditional scheme. It is shown that the navigation accuracies of the two schemes are close, proving that the dynamic model can perform as an alternate sensor of IMU in the integrated navigation system.