阅读说明：本技术 定位方法、定位系统及汽车 (Positioning method, positioning system and automobile ) 是由 宋聚宝 原诚寅 于 2019-10-22 设计创作，主要内容包括：本发明公开了一种定位方法、定位系统及汽车,方法包括：获取每个定位子系统在不同状态下的可信度并生成可信度数据表；根据每个定位子系统的实时定位数据从对应的可信度数据表中获取实时可信度；根据每个定位子系统的实时可信度求取每个定位子系统参与联邦卡尔曼滤波器融合运算的第一信息分配权重系数；通过主滤波器根据全局数据分别向每个子滤波器反馈每个定位子系统参与融合运算的第二信息分配权重系数；根据第一信息分配权重系数和第二信息分配权重系数确定每个定位子系统参与融合运算的最终信息分配权重系数。实现有效优化各个定位子系统的信息分配权重,进而提高系统定位精度并提高定位系统的鲁棒性。(The invention discloses a positioning method, a positioning system and an automobile, wherein the method comprises the following steps: obtaining the credibility of each positioning subsystem in different states and generating a credibility data table; acquiring real-time credibility from a corresponding credibility data table according to the real-time positioning data of each positioning subsystem; according to the real-time credibility of each positioning subsystem, a first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter is obtained; respectively feeding back second information of each positioning subsystem participating in fusion operation to each sub-filter through a main filter according to global data; and determining the final information distribution weight coefficient of each positioning subsystem participating in the fusion operation according to the first information distribution weight coefficient and the second information distribution weight coefficient. The information distribution weight of each positioning subsystem is effectively optimized, and the positioning precision of the system is improved and the robustness of the positioning system is improved.)

1. A method of positioning, comprising:

obtaining the credibility of each positioning subsystem in a plurality of different positioning subsystems in different states and generating a credibility data table of each positioning subsystem;

acquiring the real-time reliability of each positioning subsystem from the corresponding reliability data table according to the real-time positioning data of each positioning subsystem;

according to the real-time credibility of each positioning subsystem, a first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter is obtained;

the method comprises the steps that a main filter of the Federal Kalman filter is obtained, and second information distribution weight coefficients of each positioning subsystem participating in fusion operation are fed back to each sub-filter of the Federal Kalman filter according to global data;

determining a final information distribution weight coefficient of each positioning subsystem participating in fusion operation according to the first information distribution weight coefficient and the second information distribution weight coefficient;

and distributing a weight coefficient according to the final information of each subsystem through the Federal Kalman filter to perform the fusion operation and output a final positioning result.

2. The positioning method according to claim 1, wherein obtaining a first information distribution weight coefficient of each positioning subsystem participating in the federate kalman filter fusion operation according to the real-time reliability of each positioning subsystem comprises:

and calculating the percentage of the real-time credibility of each positioning subsystem and the sum of the real-time credibility of the plurality of subsystems, and taking each percentage as the first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter.

3. The method of claim 1, wherein determining a final information distribution weight coefficient for each positioning subsystem to participate in the fusion operation according to the first information distribution weight coefficient and the second information distribution weight coefficient comprises:

and adding the first information distribution weight coefficient of each positioning subsystem and the second information distribution weight coefficient of each subsystem, and taking an average value as the final information distribution weight coefficient of each positioning subsystem.

4. The method according to claim 1, further comprising, after said obtaining the trustworthiness of each of said positioning subsystems in different states:

respectively setting a reliability threshold value of each positioning subsystem, and when the real-time reliability of the subsystem is greater than or equal to the corresponding reliability threshold value, the subsystem participates in the fusion operation; when the real-time credibility of the subsystem is smaller than the corresponding credibility threshold, the subsystem does not participate in the fusion operation.

5. The method of claim 1, wherein the confidence level of the different states of each of the positioning subsystems is obtained by calculating a percentage of a best state to a worst state of an operating parameter in the historical statistical data of each of the positioning subsystems based on historical statistical data of the states of the operating parameters of each of the positioning subsystems.

6. A positioning system, comprising: the system comprises a plurality of positioning subsystems, a federal Kalman filter, a credibility database and a credibility evaluation module, wherein the federal Kalman filter comprises a main filter and a plurality of sub-filters;

the credibility database is used for storing credibility data tables of each positioning subsystem, wherein the credibility data tables store the credibility of each positioning subsystem in different states;

the credibility evaluation module is used for acquiring the real-time credibility of each positioning subsystem from the corresponding credibility data table according to the real-time positioning data of each positioning subsystem, and calculating a first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter according to the real-time credibility of each positioning subsystem;

the main filter is used for respectively feeding back second information of each positioning subsystem participating in fusion operation to each sub-filter according to global data and distributing weight coefficients;

the credibility evaluation module determines a final information distribution weight coefficient of each positioning subsystem participating in fusion operation according to the first information distribution weight coefficient and the second information distribution weight coefficient;

and the Federal Kalman filter distributes a weight coefficient according to the final information of each subsystem to perform fusion operation and output a final positioning result.

7. The location system of claim 6, wherein the confidence evaluation module assigns a weight coefficient by calculating a percentage of the real-time confidence of each location subsystem to a sum of the real-time reliabilities of the plurality of subsystems, and using each of the percentages as the first information for each location subsystem to participate in a Federal Kalman filter fusion operation.

8. The location system of claim 6, wherein the confidence evaluation module adds the first information distribution weight coefficient of each of the location subsystems and the second information distribution weight coefficient of each of the subsystems and then averages the added values to obtain the final information distribution weight coefficient of each of the location subsystems.

9. The positioning system according to claim 6, wherein the reliability evaluation module presets a reliability threshold of each positioning subsystem, and when the real-time reliability of the subsystem is greater than or equal to the corresponding reliability threshold, the subsystem participates in the fusion operation; when the real-time credibility of the subsystem is smaller than the corresponding credibility threshold, the subsystem does not participate in the fusion operation.

10. A vehicle, characterized in that it comprises a positioning system according to any one of claims 6 to 9.

Technical Field

The invention relates to the technical field of navigation positioning data processing, in particular to a positioning method, a positioning system and an automobile using the positioning system.

Background

The vehicle-mounted high-precision positioning technology is divided into absolute positioning and relative positioning. Common laser radars and cameras belong to the relative positioning category; the GNSS (global navigation satellite system) positioning is combined with the fusion positioning of the IMU inertial sensor, so that the latitude and longitude information of the vehicle and the current attitude information can be obtained, the absolute position of the vehicle in a terrestrial coordinate system is reflected, and the vehicle belongs to absolute positioning.

However, each type of position sensor has its own advantages and significant disadvantages.

For example, the laser radar has very high measurement accuracy, detection distance and final result, but has high cost, easy wear of parts, easy shielding in use, no ability to identify color, character and pattern, and poor mass production capability.

The visual positioning scheme is mainly realized through a camera, and is a more mainstream positioning scheme in the current automatic driving technology. The main advantages of the scheme are low cost, and the character color and pattern can be recognized. However, the visual positioning effect is affected by ambient light, and the positioning effect is also poor in the case of rain, snow, and fog.

GNSS satellite positioning and inertial navigation are used in vehicle navigation. Regarding GNSS positioning, the main advantages are low cost and high accuracy. Currently, GNSS positioning can reach dynamic centimeter level based on the differential services provided by thousand seek positions. Inertial navigation and GNSS are often used in the industry for fusion, because after inertial navigation is combined, a vehicle can still be accurately positioned under the condition that GNSS signals are lost.

However, since inertial navigation is realized by calculation, the accumulated error will be obvious after a long time, and it needs to obtain a more accurate initial position for calculation, and it generally does not exist as a separate navigation sensor, and often needs to be used with other sensors.

The sensors have advantages and defects, an optimal scheme which is compatible with all aspects is difficult to find in the practical application process, the advantages of the sensors cannot be fully utilized, and the system cannot acquire stable and reliable high-precision positioning results in certain special scenes, such as under an overhead, beside a high building, a tunnel and an underground garage and under a scene with a large number of tree shadows.

Therefore, a positioning method capable of obtaining a stable and reliable high-precision positioning result by fully utilizing the advantages of each sensor is needed.

Disclosure of Invention

The invention aims to provide a positioning method, a positioning system and an automobile, which can obtain a stable and reliable high-precision positioning result by fully utilizing the advantages of various sensors.

In order to achieve the above object, the present invention provides a positioning method, including:

obtaining the credibility of each positioning subsystem in a plurality of different positioning subsystems in different states and generating a credibility data table of each positioning subsystem;

acquiring the real-time reliability of each positioning subsystem from the corresponding reliability data table according to the real-time positioning data of each positioning subsystem;

according to the real-time credibility of each positioning subsystem, a first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter is obtained;

the method comprises the steps that a main filter of the Federal Kalman filter is obtained, and second information distribution weight coefficients of each positioning subsystem participating in fusion operation are fed back to each sub-filter of the Federal Kalman filter according to global data;

determining a final information distribution weight coefficient of each positioning subsystem participating in fusion operation according to the first information distribution weight coefficient and the second information distribution weight coefficient;

and distributing a weight coefficient according to the final information of each subsystem through the Federal Kalman filter to perform the fusion operation and output a final positioning result.

Optionally, obtaining a first information distribution weight coefficient of each positioning subsystem participating in the federate kalman filter fusion operation according to the real-time reliability of each positioning subsystem includes:

and calculating the percentage of the real-time credibility of each positioning subsystem and the sum of the real-time credibility of the plurality of subsystems, and taking each percentage as the first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter.

Optionally, determining a final information distribution weight coefficient of each positioning subsystem participating in the fusion operation according to the first information distribution weight coefficient and the second information distribution weight coefficient includes:

and adding the first information distribution weight coefficient of each positioning subsystem and the second information distribution weight coefficient of each subsystem, and taking an average value as the final information distribution weight coefficient of each positioning subsystem.

Optionally, after the obtaining the reliability of each of the positioning subsystems in different states, the method further includes:

respectively setting a reliability threshold value of each positioning subsystem, and when the real-time reliability of the subsystem is greater than or equal to the corresponding reliability threshold value, the subsystem participates in the fusion operation; when the real-time credibility of the subsystem is smaller than the corresponding credibility threshold, the subsystem does not participate in the fusion operation.

Optionally, based on historical statistics of the operating parameter status of each positioning subsystem, the reliability of the different status of each positioning subsystem is obtained by calculating the percentage of the best status to the worst status of the operating parameter in the historical statistics of each positioning subsystem.

Optionally, the obtaining the credibility of each positioning subsystem in different states includes: obtaining the credibility of the plurality of positioning subsystems in different states through at least one of a high-precision map mode, an internal parameter mode and an external parameter mode; the multiple positioning subsystems comprise a combined navigation positioning subsystem, a laser point cloud positioning subsystem and a camera vision positioning subsystem, and each sub-filter is used for filtering positioning data output by one positioning subsystem.

Optionally, the high-precision map manner includes:

setting the credibility of the combined navigation positioning subsystem corresponding to different places according to the building shielding degree of the high-precision map in the external environment information of different places, wherein the credibility of the combined navigation positioning subsystem is negatively correlated with the building shielding degree;

according to the high-precision map, acquiring the feature significance degree of surrounding objects in external environment information of different places, setting the credibility of the laser point cloud positioning subsystem corresponding to the different places, wherein the credibility of the laser point cloud positioning subsystem is positively correlated with the feature significance degree of the surrounding objects;

and setting the credibility of the laser point cloud positioning subsystem corresponding to different places according to the light change degree of the high-precision map in the external environment information of different places, wherein the credibility of the camera vision positioning subsystem is positively correlated with the light change degree.

Optionally, the internal parameter manner includes:

setting the reliability of the combined navigation positioning subsystem corresponding to different GNSS signal stability degrees and different connecting satellite numbers according to the GNSS signal stability degree and the connecting satellite numbers in the combined navigation positioning subsystem, wherein the reliability of the combined navigation positioning subsystem is positively correlated with the GNSS signal stability degree and the connecting satellite numbers;

setting the reliability of the laser point cloud positioning subsystem corresponding to different point cloud matching degrees according to the point cloud matching degree in the laser point cloud positioning subsystem; the reliability of the laser point cloud positioning subsystem is positively correlated with the point cloud matching degree;

and setting the credibility of the camera visual positioning subsystem corresponding to different feature matching degrees according to the feature matching degree in the camera visual positioning subsystem, wherein the credibility of the camera visual positioning subsystem is positively correlated with the feature matching degree.

Optionally, the external parameter manner includes:

setting the credibility of the laser point cloud positioning subsystem corresponding to different PM2.5 values according to the PM2.5 value obtained by an external PM2.5 measuring sensor, wherein the credibility of the laser point cloud positioning subsystem is in negative correlation with the PM2.5 value;

and setting the credibility of the camera visual positioning subsystem corresponding to different light brightness values according to the light brightness values obtained by the external light sensor, wherein the credibility of the camera visual positioning subsystem is positively correlated with the light brightness values.

The invention also provides a positioning system, comprising: the system comprises a plurality of positioning subsystems, a federal Kalman filter, a credibility database and a credibility evaluation module, wherein the federal Kalman filter comprises a main filter and a plurality of sub-filters;

the credibility database is used for storing credibility data tables of each positioning subsystem, wherein the credibility data tables store the credibility of each positioning subsystem in different states;

the credibility evaluation module is used for acquiring the real-time credibility of each positioning subsystem from the corresponding credibility data table according to the real-time positioning data of each positioning subsystem, and calculating a first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter according to the real-time credibility of each positioning subsystem;

the main filter is used for respectively feeding back second information of each positioning subsystem participating in fusion operation to each sub-filter according to global data and distributing weight coefficients;

the credibility evaluation module determines a final information distribution weight coefficient of each positioning subsystem participating in fusion operation according to the first information distribution weight coefficient and the second information distribution weight coefficient;

and the Federal Kalman filter distributes a weight coefficient according to the final information of each subsystem to perform fusion operation and output a final positioning result.

Optionally, the reliability evaluation module assigns a weight coefficient by calculating a percentage of a sum of the real-time reliability of each positioning subsystem and the real-time reliabilities of the plurality of subsystems, and using each of the percentages as the first information of each positioning subsystem participating in the federate kalman filter fusion operation.

Optionally, the reliability evaluation module adds the first information distribution weight coefficient of each positioning subsystem and the second information distribution weight coefficient of each subsystem and then takes an average value as the final information distribution weight coefficient of each positioning subsystem;

optionally, the reliability evaluation module is preset with a reliability threshold of each positioning subsystem, and when the real-time reliability of the subsystem is greater than or equal to the corresponding reliability threshold, the subsystem participates in the fusion operation; when the real-time credibility of the subsystem is smaller than the corresponding credibility threshold, the subsystem does not participate in the fusion operation.

Optionally, a high-precision map and an inertial sensor are further included, the high-precision map and the inertial sensor being used to provide absolute position information;

the multiple subsystems comprise a combined navigation positioning subsystem, a laser point cloud positioning subsystem and a camera vision positioning subsystem, and each sub-filter is used for filtering positioning data output by one positioning subsystem.

The invention also provides an automobile comprising the positioning system.

The invention has the beneficial effects that:

the method comprises the steps of carrying out multi-source fusion operation on positioning data output by various different positioning subsystems through a Federal Kalman filter, obtaining a first information distribution weight coefficient according to the real-time credibility of the different positioning subsystems, adjusting the information distribution weight coefficient of each positioning subsystem finally participating in the fusion operation by combining the first information distribution weight coefficient and a second information distribution weight coefficient fed back by a main filter, effectively optimizing the information distribution weight of each positioning subsystem, and further improving the positioning precision of the system and the robustness of the positioning system.

The automobile using the positioning method and the positioning system can adapt to various application scenes of different environments and accurate positioning all the day.

The apparatus of the present invention has other features and advantages which will be apparent from or are set forth in detail in the accompanying drawings and the following detailed description, which are incorporated herein, and which together serve to explain certain principles of the invention.

Drawings

The above and other objects, features and advantages of the present invention will become more apparent by describing in more detail exemplary embodiments thereof with reference to the attached drawings, in which like reference numerals generally represent like parts.

FIG. 1 shows a Federal Kalman filter model architecture diagram.

Fig. 2 shows a schematic view of the federal kalman filter model in a positioning method according to an embodiment of the present invention.

Fig. 3 shows a schematic structural diagram of a positioning system according to an embodiment of the invention.

Description of reference numerals:

1. high-precision maps; 2. a combined navigation device; 3. a laser radar device; 4. a vision sensor device; 5. a federal kalman filter; 6. a main filter; 7. a sub-filter; 8. a combined navigation positioning and attitude determination unit; 9. a point cloud feature identification unit; 10. a laser point cloud positioning unit; 11. an image feature recognition unit; 12. a camera vision positioning unit; 13. a feature matching unit; 14. a location database; 15. a path planning unit; 16. and a credibility evaluation module.

Detailed Description

The invention will be described in more detail below with reference to the accompanying drawings. While the preferred embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

The Federal Kalman Filter (FKF) is the prior art, and the main principle is to perform dispersion processing first and then perform global fusion to obtain a global optimal or suboptimal estimation based on all observations. The general structure of FKF is shown in fig. 1, and the FKF algorithm is mainly divided into two parts: local estimation and global estimation.

The output value of the local filter in fig. 1 is a local estimation value, and the local estimation value at the same time is transmitted to the main filter again and is fused with the output value of the reference system to output a global optimal estimation. The main filter is divided into two steps: time update and optimal estimation. The time updating is mainly based on the estimation result of the last moment, and the optimal estimation is mainly to fuse the time updating value of the main filter and the transfer value of each sub-filter. The global optimal estimation and the estimation error variance matrix of the main filter are respectively fed back to each sub-filter in a certain coefficient proportion mode. In addition, the reference system needs to receive the global optimal estimation of the main filter and the feedback of the estimation error variance matrix at the same time. It should be noted that it is not necessary that the output values of the reference system are passed to the main filter.

Assume that the state vector is transferred from time k-1 to time k according to the following state equation

X_k＝F_k，k-1X_k-1+W_k-1

The observation equation of the ith subsystem is Z_ik＝H_ikX_ik+V_ik

Wherein X_kIs the state at time k, Z_ikIs an observed value of the ith subsystem, W_k-1Noise at time k-1, V_ikIs the observed noise of the ith subsystem. From the above formula we can find that the two equations of FKF are similar to linear KF, and that the noise statistics of FKF equation are also consistent with KF, both gaussian noise.

Suppose that the local estimates of the N sub-filters are

The variance matrix of the corresponding estimation error is P₁、P₂……P_NThe covariance matrix of the sum system noise is Q₁、Q₂……Q_N. Also, the time update value of the main filter is the state estimateVariance matrix P of estimation errors_mSystem noise covariance matrix Q_mThe global optimum estimate may be calculated according to a formula, and the positioning data of the plurality of positioning subsystems is the sum of the subsystems. Wherein P is_g ^-1、Q_g ^-1Is a state estimation value, P_g、Q_gThe sub-filter has an important function of correcting the state estimation value according to the observation value to obtain the optimal estimation value.

Q_g ^-1＝Q₁ ^-1+Q₂ ^-1+……+Q_N ^-1+Q_m ^-1

P_g ^-1＝P₁ ^-1+P₂ ^-1+……+P_m ^-1

The value fed back to the local filter distributes the whole information according to the following rule, and the state estimation value and covariance matrix of the sub-filtering system are updated by using data fed back by the global system, that is, the correction and feedback of the global system to the sub-filtering system can be realized, and the interference of the global system to the sub-filtering system is adjusted according to the information distribution strategy.

Q_i ^-1＝β_iQ_g ^-1

P_i ^-1＝β_iP_g ^-1

Wherein

For a global optimum estimate, P_i ^-1Estimating the covariance matrix, beta, for global optimality_iAssigning a coefficient (beta) to information_iA weight coefficient is assigned corresponding to the second information in the present invention).

A positioning method according to the present invention comprises:

obtaining the credibility of each positioning subsystem in a plurality of different positioning subsystems in different states and generating a credibility data table of each positioning subsystem;

acquiring the real-time reliability of each positioning subsystem from a corresponding reliability data table according to the real-time positioning data of each positioning subsystem;

according to the real-time credibility of each positioning subsystem, a first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter is obtained;

the method comprises the steps that a main filter of the Federal Kalman filter is obtained, and second information distribution weight coefficients of each positioning subsystem participating in fusion operation are fed back to each sub-filter of the Federal Kalman filter according to global data; determining a final information distribution weight coefficient of each positioning subsystem participating in fusion operation according to the first information distribution weight coefficient and the second information distribution weight coefficient;

and distributing a weight coefficient according to the final information of each positioning subsystem through a Federal Kalman filter to perform fusion operation and output a final positioning result.

Specifically, multi-source fusion is carried out on positioning data output by multiple different positioning subsystems through a Federal Kalman filter, a first information distribution weight coefficient is obtained according to real-time credibility of the different positioning subsystems, and the information distribution weight coefficient finally participating in fusion operation of each positioning subsystem is adjusted by combining the first information distribution weight coefficient and a second information distribution weight coefficient fed back by a main filterThe method can effectively carry out real-time positioning information distribution weight optimization on the positioning data output by each positioning subsystem participating in fusion operation, further improve the positioning precision of the system and the robustness of the positioning system, and can adapt to various application scenes of different environments and precise positioning all day long. In this embodiment, the multiple positioning subsystems include an integrated navigation positioning subsystem, a laser point cloud positioning subsystem, and a camera vision positioning subsystem, and each sub-filter is configured to filter positioning data output by one positioning subsystem. The filtering algorithm of the present invention is shown in FIG. 2, where X_iIs a state estimation value, P_iIs a covariance matrix, beta_iAnd assigning a weight coefficient to the information fed back by the main filter, namely assigning a weight coefficient to the second information.

In this embodiment, obtaining a first information distribution weight coefficient of each positioning subsystem participating in the federate kalman filter fusion operation according to the real-time reliability of each positioning subsystem includes:

and calculating the percentage of the real-time reliability of each positioning subsystem and the sum of the real-time reliabilities of the subsystems, and taking each percentage as a first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter.

Determining the final information distribution weight coefficient of each positioning subsystem participating in the fusion operation according to the first information distribution weight coefficient and the second information distribution weight coefficient comprises the following steps:

and adding the first information distribution weight coefficient of each positioning subsystem and the second information distribution weight coefficient of each positioning subsystem, and taking the average value as the final information distribution weight coefficient of each positioning subsystem.

Specifically, when the positioning system is in operation, the real-time reliability obtained from the state of the positioning device, such as a high-precision map, an external sensor, etc., is not directly used for the sub-filters, and needs to be adjusted according to the covariance matrix data fed back by the main filter, and the real-time reliability of each positioning subsystem is divided by the sum of the real-time reliabilities of all the positioning subsystems to obtain a weight distribution coefficient β_i' (i.e., the first information is assigned a weight)Coefficient) and then beta fed back with the main filter_i(i.e., the second information distribution weighting factors) are added and averaged to obtain an average value, and the average value is used as the true information distribution coefficient of each positioning subsystem (i.e., the final information distribution weighting coefficient of each positioning subsystem). The weight distribution coefficient of the scheme is calculated and obtained according to the input real-time reliability and the algorithm optimization of the filter, and the information distribution weight coefficient can be optimized in real time. In other embodiments of the present invention, a non-average calculation method may also be used to calculate the true information distribution coefficient of each positioning subsystem, for example, the first information distribution weight coefficient accounts for 30%, the second information distribution weight coefficient accounts for 70%, or others, and a person skilled in the art may select a specific calculation method according to actual situations, which is not described herein again.

In this embodiment, after obtaining the reliability of each positioning subsystem in different states, the method further includes:

respectively setting a reliability threshold value of each positioning subsystem, and when the real-time reliability of the subsystems is greater than or equal to the corresponding reliability threshold value, the subsystems participate in fusion operation; and when the real-time reliability of the subsystem is smaller than the corresponding reliability threshold value, the subsystem does not participate in the fusion operation.

Specifically, the information feedback strategy of the scheme uses a hybrid feedback model (that is, a feedback model and a non-feedback model which combine with FKF), and sets a reliability threshold, so that if the reliability of one subsystem is less than the corresponding reliability threshold, the current subsystem is subjected to feedback correction through the main filter, and if the reliability of each subsystem is greater than the reliability threshold, the main filter does not perform feedback correction on the positioning subsystem. For example, the system uses three different positioning schemes of combined navigation positioning, laser point cloud positioning and camera visual positioning, the main filter does not feed back each positioning subsystem under normal operation, when the combined navigation positioning is influenced by signal shielding, the output positioning reliability is less than a threshold value, the main filter starts to feed back and correct the combined navigation positioning, so that the combined navigation positioning does not participate in the multi-source fusion positioning algorithm, and only the positioning data output by the laser point cloud positioning and the camera visual positioning is used for multi-source fusion positioning operation and corresponding weight proportion is distributed, so that the accuracy of the system positioning can be increased, and the robustness of the system is further improved.

In this embodiment, based on the historical statistical data of the operating parameter state of each positioning subsystem, the reliability of the different states of each positioning subsystem is obtained by calculating the percentage of the best state and the worst state of the operating parameter in the historical statistical data of each positioning subsystem.

Specifically, the credibility of each positioning subsystem is calculated according to different states of the operating parameters of each positioning subsystem, the algorithm obtains different credibility of the positioning subsystem in different states according to the counted percentage between the best state and the worst state of the operating parameters, and the credibility is used as a part of the sensor parameters, counted in advance and set in a corresponding self-credibility data table.

In this embodiment, obtaining the reliability of each positioning subsystem in different states includes: and obtaining the credibility of the plurality of positioning subsystems in different states through at least one of a high-precision map mode, an internal parameter mode and an external parameter mode.

Wherein, high-accuracy map mode includes:

setting the credibility of the combined navigation positioning subsystem corresponding to different places according to the building shielding degree in the external environment information of different places acquired by the high-precision map, wherein the credibility of the combined navigation positioning subsystem is negatively correlated with the building shielding degree;

setting the credibility of the laser point cloud positioning subsystem corresponding to different places according to the obvious degree of the peripheral object features in the external environment information of different places acquired by the high-precision map, wherein the credibility of the laser point cloud positioning subsystem is positively correlated with the obvious degree of the peripheral object features;

and setting the credibility of the laser point cloud positioning subsystem corresponding to different places according to the light change degree of the external environment information of different places acquired by the high-precision map, wherein the credibility of the camera vision positioning subsystem is positively correlated with the light change degree.

Specifically, the high-precision map can preset the credibility of each positioning subsystem according to different environments and a pre-measuring and calculating result, for example, the combined navigation between high buildings is greatly influenced by shielding, and the credibility of the combined navigation positioning subsystem is lower at the moment; surrounding features on an open field such as a highway and an expressway are not obvious, and the credibility of the laser point cloud positioning subsystem is low; in dark or light and shade alternating areas such as tunnels and shade streets, the credibility of the camera visual positioning subsystem is low.

The internal parameter mode comprises the following steps:

setting the reliability of the combined navigation positioning subsystem corresponding to different GNSS signal stability degrees and different connecting satellite numbers according to the GNSS signal stability degree and the connecting satellite numbers in the combined navigation positioning subsystem, wherein the reliability of the combined navigation positioning subsystem is positively correlated with the GNSS signal stability degree and the connecting satellite numbers;

setting the reliability of the laser point cloud positioning subsystem corresponding to different point cloud matching degrees according to the point cloud matching degree in the laser point cloud positioning subsystem; the credibility of the laser point cloud positioning subsystem is positively correlated with the point cloud matching degree;

and setting the credibility of the camera visual positioning subsystem corresponding to different feature matching degrees according to the feature matching degree in the camera visual positioning subsystem, wherein the credibility of the camera visual positioning subsystem is positively correlated with the feature matching degree.

Specifically, the reliability of each positioning subsystem can be obtained according to the characteristics of each system, the reliability of the combined navigation positioning subsystem can be obtained according to the number of connected GNSS signals and the number of connected satellites, and the reliability is higher if the signals are stable and the number of connected satellites is more; the credibility of the laser point cloud positioning subsystem can be obtained by taking the point cloud matching degree as a basic parameter, and the higher the point cloud matching degree is, the higher the credibility is; the credibility of the camera visual positioning subsystem can be obtained by taking the feature matching degree as a basic parameter, and the higher the feature matching degree is, the higher the credibility is.

The external parameter mode comprises the following steps:

setting the credibility of the laser point cloud positioning subsystem corresponding to different PM2.5 values according to the PM2.5 value obtained by an external PM2.5 measuring sensor, wherein the credibility of the laser point cloud positioning subsystem is in negative correlation with the PM2.5 value;

and setting the credibility of the camera visual positioning subsystem corresponding to different light brightness values according to the light brightness value obtained by the external light sensor, wherein the credibility of the camera visual positioning subsystem is positively correlated with the light brightness value.

Specifically, the external sensor is mainly applied to a laser point cloud positioning subsystem and a camera vision positioning subsystem, the laser point cloud positioning subsystem can be externally connected with a PM2.5 sensor and is used for detecting the influence of dust in the air on a laser radar, and the higher the PM2.5 value is, the lower the reliability of the laser point cloud positioning subsystem is; the camera vision positioning subsystem can be externally connected with a light sensor and used for detecting the influence of light on the camera vision, and the higher the temperature and stability of the light is, the higher the reliability of the camera vision positioning subsystem is.

As shown in fig. 3, the present invention further provides a positioning system, including: the system comprises a plurality of positioning subsystems, a federal Kalman filter 5, a credibility database (an embedded database) and a credibility evaluation module 16, wherein the federal Kalman filter 5 comprises a main filter 6 and a plurality of sub-filters 7, and each sub-filter 7 is used for filtering the positioning data of one positioning subsystem;

the credibility database is used for storing credibility data tables of each positioning subsystem, wherein the credibility data tables store the credibility of each positioning subsystem in different states;

the credibility evaluation module 16 is used for acquiring the real-time credibility of each positioning subsystem from the corresponding credibility data table according to the real-time positioning data of each positioning subsystem, and solving a first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter according to the real-time credibility of each positioning subsystem; the main filter 6 is used for feeding back second information of each positioning subsystem participating in fusion operation to each sub-filter 7 according to global data;

the credibility evaluation module 16 determines a final information distribution weight coefficient of each positioning subsystem participating in fusion operation according to the first information distribution weight coefficient and the second information distribution weight coefficient;

and the Federal Kalman filter 5 distributes a weight coefficient according to the final information of each positioning subsystem to perform fusion operation and output a final positioning result.

In the embodiment, the credibility evaluation module calculates the percentage of the real-time credibility of each positioning subsystem and the sum of the real-time credibility of a plurality of subsystems, and uses each percentage as a first information distribution weight coefficient of each positioning subsystem participating in the fusion operation of the Federal Kalman filter 5;

in this embodiment, the reliability evaluation module 16 adds the first information distribution weight coefficient of each positioning subsystem and the second information distribution weight coefficient of each positioning subsystem, and then takes an average value as a final information distribution weight coefficient of each positioning subsystem;

the credibility evaluation module 16 is preset with a credibility threshold of each positioning subsystem, and when the real-time credibility of the subsystem is greater than or equal to the corresponding credibility threshold, the subsystem participates in fusion operation; and when the real-time reliability of the subsystem is smaller than the corresponding reliability threshold value, the subsystem does not participate in the fusion operation.

In the embodiment, the system further comprises a high-precision map 1 and an inertial sensor, wherein the high-precision map 1 and the inertial sensor are used for providing absolute position information;

the multiple subsystems comprise a combined navigation positioning subsystem, a laser point cloud positioning subsystem and a camera vision positioning subsystem, and each sub-filter 7 is used for filtering positioning data output by one positioning subsystem. The integrated navigation positioning subsystem comprises integrated navigation equipment 2 and an integrated navigation positioning and attitude determination unit 8; the laser point cloud positioning subsystem comprises at least one laser radar device 3, a point cloud feature identification unit 9 and a laser point cloud positioning unit 10; the machine vision positioning subsystem comprises at least one vision sensor device 4 (a variety of cameras), an image feature recognition unit 11 and a camera vision positioning unit 12.

The positioning system of the present embodiment further comprises a feature matching unit 13, a positioning database 14 and a path planning unit 15. The feature matching unit 13 is used for performing feature matching on feature values identified by the point cloud feature identification unit 9 and the semantic feature identification unit, and the positioning database 14 and the path planning unit 15 are used for providing position data information and navigation planning paths of the high-precision map 1.

An automobile comprises the positioning system of the above embodiment. The automobile adopting the positioning system can adapt to various application scenes in different environments and accurate positioning all day long.

Having described embodiments of the present invention, the foregoing description is intended to be exemplary, not exhaustive, and not limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments.