Implementation of integrated navigation system on four rotor UAV

In recent years, with the rapid development of intelligent industry, unmanned aerial vehicle industry has gradually come into view. The structure of the four rotor UAV is simple and reliable, capable of vertical take-off and landing, hovering, stable speed flight and flight in a small space, so it has attracted wide attention and has been widely used in many fields.

Navigation system is one of the important components in the development of the four rotor UAV system. It takes on the task of providing the status data of the aircraft's position, velocity and attitude. At present, the common navigation methods are inertial navigation, satellite navigation, visual navigation, and their integrated navigation. Due to the single navigation system can not meet the development requirements of unmanned aerial vehicles, thus resulting in the different navigation systems of two or more than two kinds of combination in an appropriate way in the integrated navigation system together, has higher performance than using any single navigation system at. In this paper, the extended Calman filter is used to accomplish the navigation task by combining the inertial navigation system with the GPS navigation system. The system model is built and implemented on the four rotor unmanned aerial vehicle.

1 inertial navigation system

1.1 error analysis of inertial sensors

One of the main factors affecting the precision of navigation system is the inertial sensor errors, in order to reduce the need for sensor error, sensor calibration and error compensation to ensure the navigation system with high accuracy and high performance.

1.2 attitude matrix calculation of inertial navigation system

In the solution of navigation system calibration, the first is to get rid of the coordinate system, usually the inertial coordinate system (often referred to as I), the geographical coordinates (usually expressed as g), navigation coordinate (usually expressed as n), carrier coordinates (often referred to as b) etc.. Based on the analysis of the problem is that the navigation system navigation coordinate set for geographical coordinates, the geographic coordinate system using X to the East, to the north, Z Y refers to the day to form, can be used to figure 1 that the inertial guidance system principle equation.

Fig. 1 module diagram of navigation system equation

In the navigation system, the most important calculation is the attitude matrix. The common attitude matrix algorithms include four methods, Euler angle method and directional cosine method, 3 kinds. In the Euler angle method, the degeneration of the equation occurs, and the direction cosine method is usually very large. Therefore, the four element method is often used to solve the attitude matrix. However, there is no interchange error in the four element number method. In order to reduce this error, the equivalent rotation vector algorithm proposed by Bortz in 1971 is adopted in this paper.

The relationship between 1.2.1 rotation vector and four variables of attitude

Let Q (t+h) and Q (T) be the four variables of the attitude of the aircraft carrier at the t+h moment and the t moment, respectively:

The solution of 1.2.3 rotation vector

In general, the Taylor series expansion method is used to solve the rotation vector. If the linear fitting angular velocity, two sample rotation vector algorithm for:

2 integrated navigation system

Although the inertial navigation system can work continuously and efficiently provide attitude information, location information and velocity information, but because of inertial sensor error accumulation, inertial navigation system's accuracy will decline over time. While GPS can provide long time error for the high-precision location of meters output, and user equipment costs are low, but because the GPS signal will be blocked or interference, it can not rely solely on the GPS to provide continuous navigation parameters.

In view of the advantages and disadvantages of the INS and GPS systems, the two are combined to integrate the advantages of the two system and provide effective, long time, high accuracy and complete navigation parameters. The general structure of integrated navigation is shown in figure 2.

Figure 2 general structure of INS/GNSS integrated navigation

2.1 Calman filter

Kalman filter is an estimation algorithm, most state estimation algorithm based navigation system, smoothing, inertial navigation system such as satellite navigation results of alignment and calibration of inertial navigation system and satellite navigation sensors or other combination of navigation, and has become a key technology to obtain the optimal estimation results from measurement system navigation.

In practical engineering problems, most systems are nonlinear, so the extended Calman filter (EKF) is adopted. Extended Calman filtering (EKF) is the nonlinear form of Calman filtering.

The system dynamics model and observation model of EKF are respectively:

The function f (and) and H () cannot be applied directly to the covariance, and instead, the Jacobi matrix can be computed, which essentially linearize the nonlinear function at the current estimate.

2.2 system model and state selection

In this paper, ins and GPS navigation systems are fused, and the system model is established in the local navigation coordinate system. If the Calman filter estimates the attitude and velocity errors relative to the earth and projected to the local navigation coordinate system, and the estimated position errors are represented by latitude, longitude and height, the state vector becomes:

In the formula, the superscript n is projected to the local navigation coordinate system.

In addition to the angular velocity of the earth and the gyro measurements, the attitude propagation equation also introduces a transfer rate term, and the attitude error in the local navigation coordinate system is:

The main noise source of an inertial navigation system is the velocity error random walk caused by the noise of the accelerometer, and the random walk of the attitude error caused by the angular velocity of the gyro. If the dynamic zero bias of the accelerometer and gyro is estimated separately, the variations of the accelerometer and gyro bias can be approximately white noise at runtime.

In the INS/GPS combination, the state vector is updated using the difference between the measured output of the GPS user device and the predicted value based on the inertial navigation parameter, and which measurements are dependent on the composite structure.

3 Simulation and experimental results

This experiment chooses Pixhawk as the main control board aircraft flight control unit, MATLAB as a simulation software of the inertial navigation system and integrated navigation simulation system, four rotor UAV built environment is described as follows:

The selection of materials for nylon fiber to build the aircraft frame with diagonal frame, wheelbase is 35 cm; select brushless motor model MT2312-960KV for multi rotor aircraft, power output; selection of battery capacity is 5000 mA, the maximum discharge current of 30 A. The vehicle remote control model for the Le Di AT9, the receiver model corresponding to the 2.4 G and 9 R9D channels; ground station software recommended by 3DR for PX4/PIXHAWK design of the new QGroundControl calibration and debugging of aircraft in the environment.

3.1 flight trajectory in ideal state

In order to better understand the performance of the navigation system, the flight order of the four rotor unmanned aerial vehicle (UAV) is set up as shown in figure 3.

Fig. 3 flight sequence of simulation trajectory

The specific process can be described as hovering over time was 15 s; in the process of acceleration, the acceleration time is 10 s, the size of 0.5g (g for the acceleration of gravity); climbing process, time is 25 s, the elevation velocity of 2 DEG /s; during the dive, time is 25 s, angle velocity of 2 DEG /s; reduction in the process, the time is 5 s, acceleration 1G.

3.2 experimental results

(1) the attitude error curve of pure INS navigation system is shown in figure 4.

Fig. 4 attitude error in pure inertial navigation system

(2) the attitude error curve of the integrated navigation system is shown in figure 5.

Fig. 5 attitude error of integrated navigation system

The experimental results show that the ideal state of the flight, the aircraft roll angle and yaw angle error are numerical value (close to 0), only the pitch angle changes, as shown in Figure 4; if the pure inertial navigation, aircraft attitude error increases with time, at the end of the flight, to the East (attitude error the pitch angle error) will reach about 1.1 degrees to the north, the attitude error (roll angle error) to about 1.6 degrees, to the attitude error (yaw angle error) to about 1.5 degrees, as shown in Figure 5; if the use of INS/GPS integrated navigation system, the aircraft to the East (pitch) and the North (roll). The attitude error is less than 0.5 degrees, and less than 0.2 degrees in a long time, and the days to the (yaw) attitude angle error while in a short period of up to 1 degrees, can be maintained at about 0.5 degrees in the other period of time but, as shown in figure 6. Therefore, under the same conditions, the performance of the integrated navigation system is better than that of the pure inertial navigation system, and the error is small, which can provide more accurate navigation information for the four rotor unmanned aerial vehicle.

4 Conclusion

This paper mainly analyzes the principle and algorithm of navigation system, after the calibration and compensation of sensor error, using the extended Calman filter fusion of the strapdown inertial navigation system and GPS navigation system, and successfully applied to the four rotor unmanned aircraft. The simulation results show that the feasibility of the system implementation in the four rotor UAV, and for a long time can provide more accurate navigation information, the error is small, to ensure that the unmanned aircraft flight.