Read integrity monitoring techniques for vision navigation systems

1 Abstract

In aviation applications , Navigation integrity is crucial .GPS The integrity of the system is based on established standards . Visual navigation system has been considered as a suitable substitute for global positioning system , When unavailable , But it is unlikely to be used without measures of system integrity . We have done some work to detect the impact of a single measurement with deviation on ; However , The measurement geometry varies greatly with the environment . The environment can be tracked by sparse visual features , Or the environment can be a rich feature . With more features , There is a greater possibility of multiple damaged measurements . In order to ensure the integrity and reliability of the system , It is essential to detect and eliminate multiple damage measurements . This paper mainly studies the existing integrity monitoring methods , It is applied to multi fault detection in vision based navigation system . Current technology has been found to be capable of detecting single or multiple faults , But these technologies can't eliminate these faults . This paper develops an algorithm , Have the ability to eliminate multiple fault measurements , It increases the ability of existing integrity monitoring, fault detection and troubleshooting technology .

2 Introduce

Modern aerospace operations require accurate navigation systems . These operations include civil aircraft approach 、 Basic requirements for departure and route navigation . Other military operations that require navigation include precision bombing 、 Aerial refueling 、 Formation flying and unmanned operation .GPS The emergence and development of has brought excellent precision and accuracy to civil and military navigation users . With more global navigation satellite systems （GNSS） Installation , The performance level may continue to improve . The accuracy that can be achieved by this system or any other system is based on the ability of the system to estimate the real position given the current operating conditions and the measurement quality of the sensor . Regardless of the system , Modeling error 、 Environmental factors and equipment constraints hinder the perfect estimation of the real location .

Inertial navigation system （INS） Use specific force and angular velocity measurements to estimate the position of the vehicle 、 Speed and direction . The inertial navigation system is measured by the inertial measurement unit （IMU） produce ,IMU It consists of accelerometer and gyroscope . This type of navigation has the advantage of being independent of external signals , Therefore, it is impossible to be disturbed or deceived . therefore , It is a very useful system , Military applications and key civilian systems . Unfortunately , The system is also very dependent on the quality of the sensor , And subject to sensor misalignment 、 Effects of drift and deviations due to limitations of sensor material and Design . Besides , The specific force measured by the accelerometer is related to gravity . Errors in the gravity model used to convert a specific force into acceleration can also lead to systematic errors .

The working principle of inertial navigation system is to integrate the measured value of acceleration with time , To get the location . This integration process leads to the accumulation and secondary growth of errors in measurement with the passage of time , The system accuracy decreases with time . Because of this limitation , independent INS Mainly used for short-term navigation . For long-term navigation , One inertial navigation system and another navigation system, such as GPS Combination . By combining the two systems , Inertial navigation system can operate under the condition of error envelope , The result of fusion is better than that of single operation . This still leaves the navigation system vulnerable , Because it relies on external signals to operate GPS. These external RF signals may be interfered or deceived , Navigation solution that causes system failure or failure .

It is this vulnerability related to external signals that has led to the research of other excellent navigation systems . In order to eliminate the dependence on GNSS , An alternative being explored is to combine inertial navigation system with vision system . Such a system does not rely on external signals , It's totally passive , No RF emission . Not transmitting signals is a benefit for stealth applications , And ensure that if such a system is used on a large scale , There will be no problem in spectrum allocation . Such systems can also be very small 、 cheap , be relative to GPS System , The power consumption required for operation is very low .

The cited research shows the feasibility of vision assisted navigation system in many different configurations . These proposed systems have proven to provide reasonable accuracy at the level required for aviation applications . The accuracy available depends on the camera system used 、 Above INS Measurement error 、 Feature tracking algorithm 、 Vehicle trajectory and image scene .

With the rapid development of vision assisted inertial navigation system, it has become a feasible substitute for global positioning system , Research on visual navigation system integrity monitoring has been carried out . In the cited studies , An important first step has been taken ; Yes GPS The integrity monitoring method is redefined , And apply it to vision system . However , In these methods , Suppose there is only one bad measurement . about GPS Come on , This is a good assumption , But in the visual system , As the number of features tracked increases , The probability of multiple failures will also increase .

This paper mainly studies the existing integrity monitoring methods , It is applied to multi fault detection in vision based navigation system .

2.1 Mathematical symbols

For the sake of clarity , The following is a description of the mathematical symbols to be used in this proposal ：
Scalar ： Scalars are in italics , Such as $a$ or $b$.
vector ： Vectors are represented in lowercase bold letters , Such as $a\text{a}a$ or $b\text{b}b$, Usually in the form of columns .
matrix ： The matrix is represented by bold letters , Such as $A\text{A}A$ or $B\text{B}B$, The scalar value of the matrix can be quoted as $A_{ij}$, Including $i$ Xing He $j$ Column elements .
Transposition ： The transpose of a vector or matrix is superscript $T$ To express , Such as ${a\text{a}a}^T$.
Estimated variables ： Estimates of random variables are obtained by adding a “ hat ” A sign to indicate , for example $\hat{a}$.
Calculate and measure variables ： Variables that contain errors due to being measured are distinguished by wave symbols , Such as $\tilde{a}$.
Reference system ： The navigation vector is defined relative to the reference system , Superscript letters are used to indicate the current reference system , Such as ${x\text{x}x}^a$.
Direction cosine matrix ： The directional cosine matrix is a matrix that rotates a vector from one frame of reference to another , Such as $C_a^b$, When you multiply a vector to the left , Represents the vector from a Coordinate system conversion to b Coordinate system .
Unit matrix ： The identity matrix uses capital letters I Express , Such as $I$.
Relative position or movement ： When a vector represents relative position or motion , Subscripts grouped together , Such as ${p\text{p}p}_{ab}^c$, Express a The coordinate system is relative to b The position of the coordinate system , The position vector is represented in c In coordinate system .

2.2 Paper layout

The first 3 This section introduces the relevant background of this work . In the 3.5 In the festival , Outlines the current situation in GPS Methods used in integrity monitoring . The first 3.7 section , The current literature on integrity monitoring of visual navigation system is reviewed . The first 4 This section introduces a method to detect and eliminate multiple faults , The results of this method are shown in 5 section . The first 6 Section is the conclusion of this paper .

3 background

For details, please see 3 background ！

4 Multiple faults in vision system

For details, please see 4 Multiple faults in vision system ！

5 result

For details, please see 5 result ！