CN113916253A

CN113916253A - Method for determining covariance and related device

Info

Publication number: CN113916253A
Application number: CN202010657732.3A
Authority: CN
Inventors: 王建国; 陈默
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2020-07-09
Filing date: 2020-07-09
Publication date: 2022-01-11
Anticipated expiration: 2040-07-09
Also published as: CN113916253B; WO2022007465A1

Abstract

The embodiment of the application provides a method and a related device for determining covariance, in particular to a sensor, which can be used in the fields of auxiliary driving, automatic driving or unmanned driving. The method comprises the following steps: acquiring position measurement data from at least one sensor; and determining the covariance of the rectangular coordinate conversion measurement error of the position measurement data according to the position measurement data and the statistical characteristics of the rectangular coordinate conversion measurement error. The method and the device provided by the application can improve the covariance estimation precision of the rectangular coordinate conversion measurement error of the sensor position measurement data, and more accurately describe the statistical characteristics of the position measurement data, so that the performance of functions such as state estimation, data association, occupied grid estimation and the like of the position measurement data is greatly improved in application scenes such as tracking, navigation, environmental perception and the like, and the ADAS capability of an advanced driving assistance system is improved.

Description

Method for determining covariance and related device

Technical Field

The present application relates to the field of sensor data processing technologies, and in particular, to a method and a related apparatus for determining covariance.

Background

The sensor is a detection device which can sense the measured information and convert the sensed information into an electric signal or other information in a required form according to a certain rule to output so as to meet the requirements of information transmission, processing, storage, display, recording, control and the like. With the development of society, intelligent terminals such as intelligent transportation equipment, intelligent home equipment, and robots are gradually entering the daily lives of people. The sensor plays an important role in the intelligent terminal.

A sensor designed based on the time-of-flight (TOF) principle comprises: millimeter wave radar, ultrasonic radar, sonar, laser radar, or the like. The measurement data of such sensors are usually recorded in a polar coordinate system or a spherical coordinate system, and the content usually includes information describing the position of the object, such as distance, azimuth angle, pitch angle, and the like. However, in practical applications, the measurement data is more conveniently converted into a rectangular coordinate system. For example, the sensors are widely arranged in an Advanced Driver Assistance System (ADAS) or an Automated Driving (AD) system or an unmanned aerial vehicle system or an intelligent system such as a robot, and sense surrounding environment information through measurement data, so as to be used in a target tracking scene. In the above scenario, a rectangular coordinate system would be more useful for modeling the motion of the object, and therefore polar or spherical coordinate position measurement data is typically converted to rectangular coordinate use.

Although the individual measurement errors of the polar or spherical coordinate measurement data described above are usually statistically independent, the corresponding converted measurement errors after conversion to the rectangular coordinate system are statistically dependent. The statistical properties of the transformed measurement error may be measured by a covariance matrix, where the covariance between the two measurement error components is typically used to measure the statistical correlation of the two error components. Current sensor related products typically provide only the variance or standard deviation or Root Mean Square Error (RMSE) of the rectangular coordinate measurement error, or only the measurement error of the spherical coordinate or polar coordinate measurement data for a particular angle; this will result in the loss of covariance or low accuracy of the rectangular coordinate transformation measurement error, which leads to performance loss for functions such as state estimation, data correlation, and Occupancy Grid Map (OGM) estimation in application scenarios such as tracking, navigation, and environmental perception.

Disclosure of Invention

The embodiment of the application discloses a method and a related device for determining covariance, which can improve the covariance estimation precision of a rectangular coordinate conversion measurement error of sensor position measurement data and more accurately describe the statistical characteristics of the position measurement data, so that the performance of functions of state estimation, data association, occupied grid estimation and the like of the position measurement data is greatly improved in application scenes of tracking, navigation, environmental perception and the like.

In a first aspect, an embodiment of the present application discloses a method for determining covariance, including:

acquiring position measurement data from at least one sensor;

and determining the covariance of the rectangular coordinate conversion measurement error of the position measurement data according to the position measurement data and the statistical characteristics of the rectangular coordinate conversion measurement error.

In the embodiment of the present application, position measurement data from at least one sensor is first acquired, the position measurement data includes position information of an environment or an object, and the position measurement data of most sensors is in a spherical coordinate or polar coordinate form. Spherical coordinate position information, such as distance, azimuth angle, pitch angle and the like of an obstacle or a moving target in the environment relative to the sensor; polar coordinate position information, such as distance and azimuth of an obstacle or moving object in the environment relative to the sensor;

in actual application scenes such as tracking, navigation, environment perception and the like, specific functions such as state estimation, data association, occupied grid estimation and the like are achieved, and the polar coordinate or spherical coordinate position measurement data are converted into a rectangular coordinate form for use more conveniently. At this time, it is necessary to determine the statistical characteristics of the rectangular coordinate position measurement data and the corresponding converted measurement error, which may include variance or standard deviation or root mean square error. Although polar or spherical position measurement errors are generally statistically independent, the transformed measurement errors corresponding to rectangular position measurement data are generally statistically correlated, and thus, determining the covariance of the transformed measurement errors of the position measurement data may more accurately describe the statistical properties of the rectangular position measurement errors. By the embodiment, the covariance estimation precision of the rectangular coordinate conversion measurement error of the sensor position measurement data can be improved, and the statistical characteristics of the position measurement data can be described more accurately, so that the performance of functions such as state estimation, data association, occupancy grid estimation and the like of the position measurement data can be greatly improved in application scenes such as tracking, navigation, environmental perception and the like.

In one possible implementation manner of the first aspect, the determining the covariance of the cartesian transformation measurement error of the position measurement data includes:

and determining the covariance of the rectangular coordinate conversion measurement error of the position measurement data according to the diagonalization relation of the covariance matrix of the rectangular coordinate conversion measurement error, wherein the diagonalization relation is determined according to the position measurement data.

In the embodiment of the present application, the covariance of the cartesian coordinate conversion measurement error of the position measurement data is determined according to a diagonalization relation of a covariance matrix of the cartesian coordinate conversion measurement error based on the statistical properties of the position measurement data and the cartesian coordinate conversion measurement error, wherein the diagonalization relation may be an eigenvalue decomposition form of the covariance matrix of the cartesian coordinate conversion measurement error, such as R ═ UDU ═ u^TOr U^TRU ═ D, or a reasonable variation of the above diagonalized form, where R is the covariance matrix, D is the diagonal matrix, U is the orthogonal matrix determined from the position measurement data^TIs a transpose of the orthogonal matrix.

From the diagonalization relationship, the covariance of the transformed measurement error of the position measurement data can be determined from the diagonalization relationship of the covariance matrix based on the orthonormal matrix determinable from the position measurement data and the statistical properties of the transformed measurement error of the cartesian coordinates, such as the variance or standard deviation or root mean square error of the transformed measurement error of the cartesian coordinates.

The diagonal matrix D in the diagonalization equation is unknown, and in this embodiment, the solution of the diagonal matrix is an unnecessary step, and does not affect the determination of the covariance of the measurement error of the position measurement data conversion^TRU derives the diagonal matrix D, which can be used to determine the elliptical region of error, which can be applied accordingly to the occupancy grid estimation to improve the accuracy of the occupancy grid. The covariance of the rectangular coordinate conversion measurement error of the position measurement data is determined through the diagonalization relation of the covariance matrix, the covariance estimation precision of the rectangular coordinate conversion measurement error of the sensor position measurement data can be improved, and the statistical characteristics of the position measurement data can be described more accurately, so that the performance of functions such as state estimation, data association, occupied grid estimation and the like of the position measurement data is greatly improved under the application scenes such as tracking, navigation, environmental perception and the like.

In yet another possible implementation of the first aspect, the position measurement data is two-dimensional data, and the statistical characteristic includes a first variance or standard deviation or root mean square error of a first dimension component in the rectangular coordinate conversion measurement error of the position measurement data and a second variance or standard deviation or root mean square error of a second dimension component in the rectangular coordinate conversion measurement error of the position measurement data;

the determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the position measurement data and the statistical characteristics of the rectangular coordinate transformation measurement error comprises:

and determining a covariance matrix of rectangular coordinate conversion measurement errors of the position measurement data according to the position measurement data, the first variance or standard deviation or root mean square error and the second variance or standard deviation or root mean square error.

In the embodiment of the present application, the position measurement data is two-dimensional position measurement data, and the covariance of the rectangular coordinate transformation measurement error of the position measurement data is determined by using the two-dimensional position measurement data, a first variance or standard deviation or root-mean-square error, and a second variance or standard deviation or root-mean-square error, where the first variance or standard deviation or root-mean-square error is a statistical characteristic of the rectangular coordinate transformation measurement error, and taking the first variance and the second variance as an example, the first variance represents a variance of a first dimension component in the rectangular coordinate transformation measurement error of the position measurement data, and the second variance represents a variance of a second dimension component in the rectangular coordinate transformation measurement error of the position measurement data. The implementation mode for determining the covariance of the two-dimensional position measurement data conversion measurement errors can improve the covariance estimation precision of the two-dimensional position measurement data conversion measurement errors, so that the statistical characteristic description accuracy of the position measurement data is greatly improved.

In yet another possible implementation of the first aspect, the covariance matrix of the rectangular coordinate transformation measurement error of the position measurement data includes:

wherein R is a covariance matrix of the rectangular coordinate transformation measurement error, and R is₁₁Is the first variance, the R₂₂Is the second variance, the R₁₂The covariance between the first dimension component and the second dimension component of the rectangular coordinate transformation measurement error is derived from the position measurement data, the first variance, and the second variance to represent the covariance between the first dimension component and the second dimension component of the rectangular coordinate transformation measurement error.

In the embodiment of the present application, the position measurement data is two-dimensional position measurement data, and the covariance between the first-dimensional component and the second-dimensional component of the rectangular coordinate conversion measurement error is obtained from the statistical characteristics (the first variance and the second variance) of the position measurement data and each component of the rectangular coordinate conversion measurement error, so that the covariance matrix of the position measurement data conversion measurement error can be determined based on the above-mentioned diagonalized relational expression. The implementation mode for determining the covariance of the rectangular coordinate conversion measurement error of the two-dimensional position measurement data can improve the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data, so that the accuracy of the statistical characteristic description of the rectangular coordinate conversion measurement error of the position measurement data is greatly improved.

In yet another possible embodiment of the first aspect, the R₁₂The method comprises the following steps:

wherein, R is₁₂C is a covariance between the first dimension component and the second dimension component representing a rectangular coordinate transformation measurement error of the position measurement data_θRepresents cos θ, said s_θDenoted sin θ, said angle θ being the azimuth angle. The cos θ and sin θ may be calculated from the measured or predicted or filtered or smoothed values of the azimuth angle, or may be determined from the rectangular coordinate components x and y.

In the embodiment of the application, the position measurement data is two-dimensional position measurement data, and the covariance R of the rectangular coordinate conversion measurement error of the position measurement data is determined by using the position measurement data and the first variance and the second variance₁₂Above, determining R₁₂The formula implementation mode can improve the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data, so that the accuracy of the statistical characteristic description of the rectangular coordinate conversion measurement error of the position measurement data is greatly improved.

In yet another possible implementation manner of the first aspect, the position measurement data is three-dimensional data, and the statistical characteristics include a first variance or standard deviation or root mean square error of a first dimension component in the position measurement data rectangular coordinate conversion measurement errors, a second variance or standard deviation or root mean square error of a second dimension component in the position measurement data rectangular coordinate conversion measurement errors, and a third variance or standard deviation or root mean square error of a third dimension component in the position measurement data rectangular coordinate conversion measurement errors;

and determining the covariance of the rectangular coordinate conversion measurement error of the position measurement data according to the position measurement data, the first variance or standard deviation or root mean square error, the second variance or standard deviation or root mean square error and the third variance or standard deviation or root mean square error.

In the embodiment of the present application, the position measurement data is three-dimensional data, and the covariance of the position measurement data conversion measurement error is determined by using the three-dimensional position measurement data, a first variance or standard deviation or root-mean-square error, a second variance or standard deviation or root-mean-square error, and a third variance or standard deviation or root-mean-square error, where the first variance or standard deviation or root-mean-square error is a statistical characteristic of the position measurement error at the rectangular coordinate conversion position, and the first variance, the second variance, and the third variance are taken as examples, the first variance represents the variance of a first dimensional component of the position measurement data rectangular coordinate conversion measurement error, the second variance represents the variance of a second dimensional component of the position measurement data rectangular coordinate conversion measurement error, and the third variance represents the variance of a third dimensional component of the position measurement data rectangular coordinate conversion measurement error. The implementation mode for determining the covariance of the position measurement data of the three-dimensional data can improve the covariance precision of the rectangular coordinate conversion measurement error of the position measurement data, so that the description accuracy of the statistical characteristic of the rectangular coordinate conversion measurement error of the position measurement data is greatly improved.

In yet another possible implementation manner of the first aspect, the determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the position measurement data, the first variance or standard deviation or root mean square error, the second variance or standard deviation or root mean square error, and the third variance or standard deviation or root mean square error comprises:

determining a covariance matrix of the rectangular coordinate transformation measurement error of the position measurement data according to the first variance, the second variance, the third variance, a first target covariance, a second target covariance and a third target covariance;

wherein the first target covariance represents a covariance between the first dimension component and the second dimension component in the position measurement data rectangular coordinate transformation measurement error; the second target covariance represents a covariance between the first dimension component and the third dimension component in the position measurement data rectangular coordinate transformation measurement error; the third target covariance represents a covariance between the second-dimensional component and the third-dimensional component in the position measurement data rectangular coordinate conversion measurement error.

In this embodiment, the position measurement data is three-dimensional position measurement data, and the covariance between the components in the rectangular coordinate conversion measurement error of the position measurement data can be obtained according to the variance of each component in the rectangular coordinate conversion measurement error and the three-dimensional position measurement data. Specifically, a first target covariance is obtained according to the position measurement data, the first variance and the second variance, wherein the first target covariance represents a covariance between a first dimension component and a second dimension component in a rectangular coordinate transformation measurement error of the position measurement data; similarly, a second target covariance is obtained according to the position measurement data, the first square difference and the third square difference, wherein the second target covariance represents a covariance between the first dimensional component and the third dimensional component in the rectangular coordinate conversion measurement error of the position measurement data; and obtaining a third target covariance according to the position measurement data, the second variance and the third variance, wherein the third target covariance represents the covariance between the second-dimensional component and the third-dimensional component in the rectangular coordinate conversion measurement error of the position measurement data. A covariance matrix of a rectangular coordinate transformation measurement error of the position measurement data is determined based on the first variance, the second variance, the third variance, the first target covariance, the second target covariance, and the third target covariance. By the implementation mode for determining the covariance of the three-dimensional position measurement data, the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data can be improved, and the description accuracy of the rectangular coordinate conversion measurement error statistical property of the position measurement data is greatly improved.

wherein R is a covariance matrix of the rectangular coordinate transformation measurement error, and R is₁₁Is the first variance, the R₂₂Is the second variance, the R₃₃Is the third variance, the R₁₂Is the first target covariance, the R₁₃Is the second target covariance, the R₂₃Is the third target covariance.

In the embodiment of the application, a specific formula implementation of a covariance matrix of a rectangular coordinate conversion measurement error of three-dimensional position measurement data is provided, so that the covariance of the rectangular coordinate conversion measurement error of the position measurement data can be determined based on a diagonalized relation of the covariance matrix. By the implementation mode of determining the covariance matrix of the three-dimensional position measurement data, the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data can be improved, and the accuracy of the statistical characteristic description of the rectangular coordinate conversion measurement error of the position measurement data is greatly improved.

and determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the first variance, the second variance, the third variance and an angle matrix, wherein the angle matrix is determined by the trigonometric function value of the azimuth angle and/or the pitch angle.

In the embodiment of the present application, another implementation manner is provided in which the position measurement data is three-dimensional position measurement data, and the covariance of the rectangular coordinate transformation measurement error of the position measurement data is determined according to a first variance, a second variance, a third variance, and an angle matrix, where the first variance, the second variance, and the third variance represent statistical characteristics of the rectangular coordinate transformation measurement error of the position measurement data, and the angle matrix is determined according to the position measurement data, and specifically, may be determined according to a trigonometric function value of an azimuth angle and/or a pitch angle of a position, for example, a sine function value or a cosine function value of the azimuth angle and/or the pitch angle. By the implementation mode for determining the covariance of the rectangular coordinate conversion measurement error of the three-dimensional position measurement data, the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data can be improved, and the accuracy of the statistical characteristic description of the rectangular coordinate conversion measurement error of the position measurement data is greatly improved.

In yet another possible implementation of the first aspect, the angle matrix includes:

wherein A is the angle matrix, the

To represent

C is mentioned_θRepresents cos θ, said

To represent

S is_θDenotes sin θ, said

Is a pitch angle and theta is an azimuth angle.

In the embodiment of the present application, the position measurement data is three-dimensional position measurement data, and another implementation manner for determining the covariance of the rectangular coordinate transformation measurement error of the three-dimensional position measurement data is provided, that is, the position measurement data is three-dimensional position measurement data, according to a specific implementation manner of an angle matrix, wherein the angle matrix is determined according to a trigonometric function value of an azimuth angle and/or a pitch angle, and the trigonometric function value of the azimuth angle and/or the pitch angle can be obtained according to the position measurement data,

the pitch angle of the target relative to the sensor and the azimuth angle of the target relative to the sensor are theta, and the target can be the environment around the sensor, such as an obstacle in the environment and the like, and can also be a moving target; further:

wherein d is a spatial distance of the target relative to the sensor, r is a planar distance of the target relative to the sensor, and is a projection component of the spatial distance d on a plane formed by a first dimensional component and a second dimensional component of the rectangular coordinate system, x is a first dimensional component in the rectangular coordinate system of the position of the target relative to the sensor, y is a second dimensional component in the rectangular coordinate system of the position of the target relative to the sensor, and z is a third dimensional component in the rectangular coordinate system of the position of the target relative to the sensor. By the implementation mode for determining the covariance of the rectangular coordinate conversion measurement error of the three-dimensional position measurement data, the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data can be improved, and the accuracy of the description of the rectangular coordinate conversion measurement error statistical characteristics of the position measurement data is greatly improved.

In yet another possible implementation of the first aspect, the covariance of the cartesian transformation measurement error of the position measurement data comprises:

wherein A is the angle matrix and R is₁₁Is the first variance, the R₂₂Is the second variance, the R₃₃Is the third variance, the R₁₂Is a first target covariance, said R₁₃Is a second target covariance, said R₂₃A third target covariance, the first target covariance representing a covariance between the first dimension component and the second dimension component in the position measurement data rectangular coordinate transformation measurement error; the second target covariance represents a covariance between the first dimension component and the third dimension component in the position measurement data rectangular coordinate transformation measurement error; the third target covariance represents a covariance between the second-dimensional component and the third-dimensional component in the position measurement data rectangular coordinate conversion measurement error.

In the embodiment of the present application, the position measurement data is three-dimensional position measurement data, and a further implementation manner for determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data is further supplemented, that is, a formula relationship among a first variance, a second variance, a third variance, an angle matrix and the covariance of the rectangular coordinate transformation measurement error of the position measurement data is given, and it can be seen from the formula relationship that the covariance of the rectangular coordinate transformation measurement error of the position measurement data can be determined according to the first variance, the second variance, the third variance and the angle matrix, wherein the determined first target covariance, the second target covariance and the third target covariance respectively represent the covariance between two different components in the rectangular coordinate transformation measurement error of the position measurement data, and further can be further determined according to the first target covariance, And obtaining a covariance matrix of the rectangular coordinate conversion measurement error of the position measurement data by using the second target covariance, the third target covariance, the first variance, the second variance and the third variance and using the diagonalization relation of the covariance matrix. By the implementation mode for determining the covariance of the three-dimensional position measurement data, the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data can be improved, and the accuracy of the description of the rectangular coordinate conversion measurement error statistical characteristics of the position measurement data is greatly improved.

In yet another possible implementation of the first aspect, the method further comprises:

and determining a diagonal matrix according to the covariance matrix and an orthogonal matrix, wherein diagonal elements of the diagonal matrix are the variances of statistically independent errors, and the orthogonal matrix is determined by the position measurement data.

In the embodiment of the present application, an implementation manner of obtaining a diagonal matrix is provided, and specifically, the diagonal matrix may be further determined based on a covariance matrix and an orthogonal matrix of a rectangular coordinate transformation measurement error of position measurement data and a diagonalization relation. In the diagonalized relation of the covariance matrix, the diagonal matrix is unknown, and in the process of determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data, the solution of the diagonal matrix is an unnecessary step and does not affect the covariance of the rectangular coordinate transformation measurement error of the position measurement data, however, the diagonal matrix can be used for determining the elliptic region of the error, and accordingly, the diagonal matrix can be applied to the occupied grid estimation to improve the accuracy of the occupied grid.

In yet another possible implementation of the first aspect, the position measurement data is data in one of the following coordinate systems: rectangular coordinate system, polar coordinate system, spherical coordinate system.

In yet another possible embodiment of the first aspect, the position measurement data comprises a distance, an azimuth angle, and a pitch angle; alternatively, the position measurement data includes range and azimuth.

In the embodiment of the present application, it can be seen that, when the position measurement data is two-dimensional position measurement data, the two-dimensional position measurement data includes a distance between the measured object and the sensor and an azimuth angle of the measured object relative to the sensor, and when the position measurement data is three-dimensional position measurement data, the three-dimensional position measurement data includes a distance between the measured object and the sensor, an azimuth angle of the measured object relative to the sensor, and a pitch angle of the measured object relative to the sensor.

In a second aspect, an embodiment of the present application discloses an apparatus for determining covariance, including:

an acquisition unit for acquiring position measurement data from at least one sensor;

and the determining unit is used for determining the covariance of the rectangular coordinate conversion measurement error of the position measurement data according to the position measurement data and the statistical characteristics of the rectangular coordinate conversion measurement error.

In a possible embodiment of the second aspect, in determining the covariance of the cartesian coordinate conversion measurement error of the position measurement data, the determining unit is specifically configured to determine the covariance of the cartesian coordinate conversion measurement error of the position measurement data according to a diagonalized relationship of a covariance matrix of the cartesian coordinate conversion measurement errors, wherein the diagonalized relationship is determined according to the position measurement data.

The diagonal matrix D in the above diagonalized relation is unknown, in this caseIn an embodiment, the solving of the diagonal matrix is an unnecessary step, and does not affect the covariance for determining the measurement error of the position measurement data conversion, and optionally, the method may further include determining the covariance based on the covariance matrix R, the orthonormal matrix U, and the diagonalized relation D ═ U^TRU derives the diagonal matrix D, which can be used to determine the elliptical region of error, which can be applied accordingly to the occupancy grid estimation to improve the accuracy of the occupancy grid. The covariance of the rectangular coordinate conversion measurement error of the position measurement data is determined through the diagonalization relation of the covariance matrix, the covariance estimation precision of the rectangular coordinate conversion measurement error of the sensor position measurement data can be improved, and the statistical characteristics of the position measurement data can be described more accurately, so that the performance of functions such as state estimation, data association, occupied grid estimation and the like of the position measurement data is greatly improved under the application scenes such as tracking, navigation, environmental perception and the like.

In yet another possible embodiment of the second aspect, the position measurement data is two-dimensional data, and the statistical characteristic includes a first variance or standard deviation or root mean square error of a first dimension component in the rectangular coordinate conversion measurement error of the position measurement data and a second variance or standard deviation or root mean square error of a second dimension component in the rectangular coordinate conversion measurement error of the position measurement data;

in the aspect of determining the covariance of the cartesian coordinate conversion measurement error of the position measurement data according to the position measurement data and the statistical characteristics of the cartesian coordinate conversion measurement error, the determining unit is specifically further configured to determine the covariance matrix of the cartesian coordinate conversion measurement error of the position measurement data according to the position measurement data, the first variance, the standard deviation, or the root-mean-square error, and the second variance, the standard deviation, or the root-mean-square error.

In yet another possible embodiment of the second aspect, the covariance matrix of the rectangular coordinate transformation measurement errors of the position measurement data comprises:

In yet another possible embodiment of the second aspect, the R₁₂The method comprises the following steps:

wherein, R is₁₂C is a covariance between the first dimension component and the second dimension component representing a rectangular coordinate transformation measurement error of the position measurement data_θRepresents cos θ, said s_θDenoted sin θ, said angle θ being the azimuth angle. The cos θ and sin θ may be calculated from a measured value or predicted value or filtered value or smoothed value of the azimuth angle, or may be determined from the rectangular coordinate components x and y.

In yet another possible embodiment of the second aspect, the position measurement data is three-dimensional data, and the statistical characteristics include a first variance or standard deviation or root mean square error of a first dimension component in the position measurement data rectangular coordinate conversion measurement errors, and a second variance or standard deviation or root mean square error of a second dimension component in the position measurement data rectangular coordinate conversion measurement errors, and a third variance or standard deviation or root mean square error of a third dimension component in the position measurement data rectangular coordinate conversion measurement errors;

in the aspect of determining the covariance of the cartesian coordinate conversion measurement error of the position measurement data according to the statistical characteristics of the position measurement data and the cartesian coordinate conversion measurement error, the determining unit is specifically further configured to determine the covariance of the cartesian coordinate conversion measurement error of the position measurement data according to the position measurement data, the first variance, the first standard deviation, the second variance, the second standard deviation, the third variance, the third standard deviation, the third root mean square error.

In a further possible embodiment of the second aspect, in determining the covariance of the cartesian coordinate transformation measurement error of the position measurement data based on the position measurement data and the first variance or standard deviation or root mean square error, and the second variance or standard deviation or root mean square error, and the third variance or standard deviation or root mean square error, the determining unit is further configured to determine the covariance matrix of the cartesian coordinate transformation measurement error of the position measurement data based on the first variance, the second variance, the third variance, a first target covariance, a second target covariance, and a third target covariance;

In a further possible embodiment of the second aspect, in determining the covariance of the cartesian transformation measurement error of the position measurement data based on the position measurement data and the first variance or standard deviation or root mean square error, and the second variance or standard deviation or root mean square error, and the third variance or standard deviation or root mean square error, the determining unit is further configured to determine the covariance of the cartesian transformation measurement error of the position measurement data based on the first variance, the second variance, the third variance, and an angle matrix, the angle matrix being determined by trigonometric values of an azimuth angle and/or a pitch angle.

In yet another possible embodiment of the second aspect, the angle matrix includes:

wherein A is the angle matrix, the

To represent

C is mentioned_θRepresents cos θ, said

To represent

S is_θDenotes sin θ, said

Is a pitch angle and theta is an azimuth angle.

wherein d is a spatial distance of the target relative to the sensor, r is a planar distance of the target relative to the sensor, and is a projection component of the spatial distance d on a plane formed by a first dimensional component and a second dimensional component of the rectangular coordinate system, x is a first dimensional component of the position of the target relative to the sensor in the rectangular coordinate system, y is a second dimensional component of the position of the target relative to the sensor in the rectangular coordinate system, and z is a third dimensional component of the target relative to the sensor in the rectangular coordinate system. By the implementation mode for determining the covariance of the rectangular coordinate conversion measurement error of the three-dimensional position measurement data, the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data can be improved, and the accuracy of the description of the rectangular coordinate conversion measurement error statistical characteristics of the position measurement data is greatly improved.

In yet another possible implementation of the second aspect, the covariance of the position measurement data rectangular coordinate transformation measurement error comprises:

In a further possible implementation manner of the second aspect, the determining unit is further configured to determine a diagonal matrix according to the covariance matrix and an orthogonal matrix, wherein diagonal elements of the diagonal matrix are variances of statistically independent errors, and the orthogonal matrix is determined by the position measurement data.

In yet another possible embodiment of the second aspect, the position measurement data is data in one of the following coordinate systems: rectangular coordinate system, polar coordinate system, spherical coordinate system.

In yet another possible embodiment of the second aspect, the position measurement data comprises a distance, an azimuth angle, and a pitch angle; alternatively, the position measurement data includes range and azimuth.

In a third aspect, an embodiment of the present application discloses a sensor, including a sensor element, a conversion element, a memory, and a processor, where the sensor element is configured to obtain position measurement data from at least one sensor, the memory stores a computer program, and the processor invokes the computer program stored in the memory to perform the following operations:

acquiring position measurement data from at least one sensor;

In a fourth aspect, an embodiment of the present application discloses an electronic device for determining covariance, the electronic device comprising a memory and a processor, the memory storing a computer program, when the computer program runs on the processor, the electronic device performing the method according to the first aspect or any one of the possible implementation manners of the first aspect.

In a fifth aspect, this application discloses a computer-readable storage medium, in which a computer program is stored, which, when running on one or more processors, performs the method as set forth in the first aspect or any one of the possible implementations of the first aspect.

In a sixth aspect, the present application discloses a sensor system, which may include at least one sensor including the apparatus for determining covariance of the second aspect, or the sensor of the third aspect, or the electronic device of the fourth aspect, and is configured to implement the method shown in the first aspect or any one of the possible implementations of the first aspect.

In a seventh aspect, an embodiment of the present application discloses a chip system, which includes at least one processor and an interface circuit. Optionally, the interface circuit and the at least one processor are interconnected by a line, and the interface circuit is used for connecting an external device to the processor. The chip system may further comprise at least one memory storing a computer program, or the interface circuit may be adapted to provide the at least one processor with a computer program stored in an external memory; the computer program, when executed by the at least one processor, is adapted to carry out the method of the first aspect or any of its possible implementations.

According to the embodiment of the application, the covariance of the rectangular coordinate conversion measurement error of the position measurement data is determined according to the diagonalization relation of the covariance matrix by acquiring the statistical properties of the position measurement data and the rectangular coordinate conversion measurement error, the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data of the sensor is improved, and the statistical properties of the position measurement data are described more accurately, so that the performance of functions such as state estimation, data association and occupied grid estimation of the position measurement data is greatly improved in application scenes such as tracking, navigation and environmental perception.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments or the background art of the present application, the drawings used in the embodiments or the background art of the present application will be briefly described below.

Fig. 1 is a schematic flowchart of a method for determining covariance according to an embodiment of the present application;

FIG. 2 is a schematic coordinate diagram of position measurement data provided in an embodiment of the present application;

FIG. 3a is a schematic flow chart illustrating another method for determining covariance according to an embodiment of the present disclosure;

FIG. 3b is a block diagram illustrating a method for determining covariance according to an embodiment of the present application;

FIG. 4a is a schematic flowchart of another method for determining covariance according to an embodiment of the present disclosure;

FIG. 4b is a block diagram illustrating another method for determining covariance according to an embodiment of the present application;

FIG. 5a is a schematic flowchart of another method for determining covariance according to an embodiment of the present disclosure;

FIG. 5b is a block diagram illustrating another method for determining covariance according to an embodiment of the present application;

fig. 6 is a schematic view of a sensor application scenario provided in an embodiment of the present application;

fig. 7 is a schematic view of an imaging scene of a sensor according to an embodiment of the present disclosure;

fig. 8 is a schematic structural diagram of an apparatus for determining covariance according to an embodiment of the present application;

fig. 9 is a schematic structural diagram of an apparatus for determining covariance according to an embodiment of the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more clear, the present application will be further described with reference to the accompanying drawings.

The terms "first" and "second," and the like in the description, claims, and drawings of the present application are used solely to distinguish between different objects and not to describe a particular order. Furthermore, the terms "comprising" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions. Such as a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those steps or elements listed, but may alternatively include other steps or elements not listed, or inherent to such process, method, article, or apparatus.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those skilled in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.

In this application, "at least one" means one or more, "a plurality" means two or more, "at least two" means two or three and three or more, "and/or" for describing an association relationship of associated objects, which means that there may be three relationships, for example, "a and/or B" may mean: only A, only B and both A and B are present, wherein A and B may be singular or plural. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship. "at least one item(s) below" or similar expressions refer to any combination of these items. For example, at least one (one) of a, b, or c, may represent: a, b, c, "a and b," a and c, "" b and c, "or" a and b and c.

In order to describe the scheme of the present application more clearly, some knowledge related to covariance is introduced below.

Covariance: covariance is used in probability theory and statistics to measure the overall error of two variables, and variance is a special case of covariance, i.e. when two variables are the same. Covariance represents the error of the sum of two variables, as opposed to variance which represents the error of only one variable. If the two variables have the same trend, i.e. if one of them is greater than its expected value and the other is also greater than its expected value, the covariance between the two variables is positive. If the two variables have opposite trend, i.e. one of them is larger than the expected value of itself and the other is smaller than the expected value of itself, the covariance between the two variables is negative. I.e. when multiple variables are independent, the variance is used to evaluate the difference in this effect; when multiple variables are correlated, the variance is used to evaluate the difference in this effect. The application range of covariance is very wide, and in agricultural science experiments, the situation that controllable quality factors and uncontrollable quantity factors influence the experimental results at the same time often occurs, and at this time, a statistical processing method of covariance analysis needs to be adopted to comprehensively consider the quality factors and the quantity factors (also called covariates). For example, the actual effect of three fertilizers on apple yield is researched, and the first-year 'basal yield' of each apple tree is inconsistent, but has a certain influence on the test result. To eliminate the influence of this factor, the yield of each apple tree in the first year needs to be used as a covariate to perform covariance analysis, so as to obtain a correct experimental result.

Covariance matrix: in statistics and probability theory, each element of the covariance matrix is the covariance between the elements of the respective vectors, a natural generalization from scalar random variables to high-dimensional random vectors. Covariance matrices are found everywhere in statistics and machine learning and can be generally considered as two-part components of variance and covariance, i.e., variance constitutes diagonal elements and covariance constitutes off-diagonal elements. The covariance matrix can be used to represent the probability density of the multidimensional random variable, so that the study of the multidimensional random variable can be achieved through the covariance matrix.

The embodiments of the present application will be described below with reference to the drawings.

Referring to fig. 1, fig. 1 is a schematic flow chart of a method for determining covariance according to an embodiment of the present application, the method includes, but is not limited to, the following steps:

step 101: position measurement data from at least one sensor is acquired.

Specifically, one or more position measurement data is obtained from the sensor, and the position measurement data is used for describing the position information of the measured object.

The sensors in this embodiment may include a large class of sensors based on time of flight (TOF) measurement, such as radar like millimeter wave radar, ultrasonic radar/sonar or laser radar, and the position measurement data of such sensors is usually recorded in polar or spherical coordinates. For example, the above-mentioned sensors are widely deployed in Advanced Driver Assist Systems (ADAS) or Automated Driving (AD) systems or unmanned aerial vehicle systems or smart systems such as robots for sensing ambient environment information, for example, a typical vehicle-mounted millimeter wave radar may provide the following position measurement data with respect to the sensors: distance, azimuth, or azimuth and pitch, etc. However, in practical applications, the position measurement data often need to be converted into a rectangular coordinate system for use more conveniently, because the rectangular coordinate system is more favorable for describing the motion modeling of the measured target and is suitable for the tracked scene of the measured target, and therefore, the polar coordinate or spherical coordinate position measurement data is usually converted into a rectangular coordinate for use.

The position measurement data is rectangular coordinate position data; it can be obtained based on the measured data of the sensor, such as the distance, azimuth angle or distance, azimuth angle and pitch angle measured by the sensor, or can be directly output after the sensor utilizes the above data conversion. And are not limited herein. The position measurement data may also be polar or spherical coordinate position data, such as from a distance, an azimuth angle or a distance, an azimuth angle and a pitch angle measured by a sensor, without limitation.

It is noted that the individual measurement errors of polar or spherical coordinate measurement data are typically statistically independent, but the corresponding converted measurement errors after conversion to a rectangular coordinate system are typically statistically dependent. Specifically, referring to fig. 2, fig. 2 is a schematic coordinate diagram of position measurement data according to an embodiment of the present disclosure. As shown in fig. 2 (a), the distance and the azimuth are measured data of the position of the measured object relative to the sensor in a polar coordinate system, the sensor is located at the origin of the polar coordinate system, the polar axis is a ray with the sensor as an end point, and at this time, the position coordinate of the object is (d, θ), where d represents the distance between the object and the sensor, and θ represents the angle (azimuth) of the object relative to the sensor; in consideration of convenience in practical applications, such as modeling of motion state, the position measurement data usually needs to be converted into a rectangular coordinate system for use. As an implementation manner, the polar coordinate position may be directly converted into rectangular coordinates by using a transformation relationship between polar coordinates and rectangular coordinates, as shown in (b) of fig. 2, in order to measure the position of the measured object relative to the sensor in a planar rectangular coordinate system, the sensor is located at the origin of the planar rectangular coordinate system, and at this time, the position coordinates of the object are (x ', y'), where x 'is the abscissa of the object and y' is the ordinate of the object. Converting from the position measurement data in (a) in fig. 2 to the position measurement data in (b) in fig. 2 requires the following steps:

as can be seen from the above coordinate transformation relation, each measurement error (distance d, azimuth angle θ) of the polar coordinate measurement data is statistically independent, and the corresponding transformed measurement error after transformation to the rectangular coordinate system is statistically correlated, the transformed measurement error of x 'is correlated to distance d and azimuth angle θ, and the transformed measurement error of y' is also correlated to distance d and azimuth angle θ.

It should be noted that the present invention is not limited to the above-mentioned transformation relationship between polar coordinates and rectangular coordinates. For example, the coordinate transformation relationship may take into account the effects of measurement errors. For example, as another implementation, the position measurement data in rectangular coordinates can also be derived from the position data in polar coordinates according to the following relation:

wherein the content of the first and second substances,

is the variance of the measurement error of the azimuth angle of the corresponding sensor of the target.

Similarly, as shown in fig. 2 (c), the sensor is located at the origin of the spherical coordinate system for measuring the position of the target relative to the sensor in the spherical coordinate system, and the position coordinate of the target is

Where d represents the distance of the target from the sensor, theta is the azimuth angle of the target from the sensor,

representing the pitch angle of the target relative to the sensor; for practical convenience, the position measurement data usually needs to be converted to a rectangular coordinate system for use. As an implementation manner, the spherical coordinate position may be directly converted into rectangular coordinates by using a transformation relationship between the spherical coordinates and rectangular coordinates, as shown in (d) of fig. 2, which is position measurement data of the measured object relative to the sensor in a spatial rectangular coordinate system whose origin is the sensor, and at this time, the position coordinates of the object are (x ', y', z '), where x', y 'and z' are coordinate components of the three coordinate axes of the three-dimensional rectangular coordinate of the object. Converting from the position measurement data in (c) in fig. 2 to the position measurement data in (d) in fig. 2 requires the following steps:

it should be noted that the present invention is not limited to the transformation relationship between the spherical coordinates and the rectangular coordinates. For example, the coordinate transformation relationship may take into account the effects of measurement errors. For example, as another implementation, the position measurement data of rectangular coordinates can also be obtained from the position data of spherical coordinates according to the following relation:

wherein the content of the first and second substances,

the variance of the azimuth angle measurement error of the corresponding sensor for the target,

the variance of the pitch angle measurement error of the corresponding sensor for the target.

As can be seen from the above conversion relation, the respective measurement errors (distance d, azimuth angle theta, pitch angle) of the spherical coordinate measurement data

) Is statistically independent, and the corresponding transformed measurement error after transformation to the rectangular coordinate system is statistically dependent, the transformed measurement error of x' being related to the distance d, the azimuth angle theta and the pitch angle

Correlation, y' with the converted measurement error and the distance d, azimuth theta and pitch

Correlation, the conversion of the measurement error of z' with the distance d and the pitch angle

And (4) correlating.

Due to the influence of non-ideal factors such as thermal noise and interference, the measurement data of the sensor usually has a certain measurement error. Although the errors of the individual measurement data of the sensors can be statistically independent, the statistics of the corresponding transformed measurement errors after the transformation into the cartesian coordinate system have a statistical correlation, and therefore it is necessary to determine the covariance of the cartesian transformed measurement errors of the position measurement data.

Step 102: and determining the covariance of the rectangular coordinate conversion measurement error of the position measurement data according to the position measurement data and the statistical characteristics of the rectangular coordinate conversion measurement error.

The statistical characteristics of the rectangular coordinate system conversion measurement error corresponding to the position measurement data may include variance, standard deviation, root mean square error, and the like. Specifically, the statistical characteristics of the rectangular coordinate conversion measurement error may include statistical characteristics of each component of the rectangular coordinate conversion measurement error. Variance is a measure of the degree of dispersion when probability theory and statistical variance measure a random variable or a set of data; the standard deviation is the arithmetic square root of the variance, reflecting the degree of dispersion of a data set; the root mean square error is a measure of the deviation of the observed value from the true value. The variance or standard deviation or root mean square error is essentially a specific representation of statistical characteristics, and for convenience of illustration, the variance is taken as an example in the present embodiment.

The position measurement data may be two-dimensional or three-dimensional rectangular coordinate position data, or two-dimensional data in a polar coordinate system, or three-dimensional data in a spherical coordinate system, which is not limited herein. The statistical properties of the rectangular coordinate conversion measurement error of the position measurement data are statistical properties of the respective components of the conversion measurement error, and are different depending on the situation of the position measurement data.

As one implementation, the position measurement data is two-dimensional data in a polar coordinate system, and the statistical characteristic of rectangular coordinate conversion measurement error is the variance of each component in (x, y) in a planar rectangular coordinate system, which is respectively a first variance R on the x component₁₁And a second variance R on the y component₂₂. Wherein the first variance R₁₁The second variance R₂₂May be obtained by a sensor. The invention is not limited to how the sensor obtains the rectangular coordinate conversion measurement error by using the polar coordinate or spherical coordinate data measured by the sensor.

Similarly, the position measurement data is three-dimensional data in a spherical coordinate system, and rectangular coordinate conversion thereofThe statistical characteristic of the measurement error is the variance of each error component in (x, y, z) rectangular coordinate system, and the first variance R of the error component in x coordinate axis₁₁Second variance R of error component of y coordinate axis₂₂And a third variance R of the error component in the z-coordinate axis₃₃. Wherein the first variance R₁₁The second variance R₂₂Third party diff R₃₃May be obtained by a sensor. The invention is not limited to how the sensor obtains the rectangular coordinate conversion measurement error by using the polar coordinate or spherical coordinate data measured by the sensor.

Based on the statistical properties of the position measurement data and the rectangular coordinate conversion measurement error, the covariance of the rectangular coordinate conversion measurement error of the position measurement data can be determined according to the diagonalized relationship of the covariance matrix of the rectangular coordinate conversion measurement error. The diagonalized relationship may be an eigenvalue decomposition of a covariance matrix of rectangular coordinate transformation measurement errors, such as R ═ UDU^TOr U is^TRU ═ D, or a reasonable variation of the above diagonalized version. Wherein R is a covariance matrix, D is a diagonal matrix, U is an orthogonal matrix determined from the position measurement data, U is a position measurement data^TIs the transpose of the orthogonal matrix. In the diagonalization relationship, the orthogonal matrix is determined according to the position measurement data, so that a covariance matrix can be obtained according to the position measurement data and the diagonalization relationship of the covariance matrix, and the covariance of the rectangular coordinate conversion measurement error of the position measurement data is determined.

Specifically, the position measurement data may be two-dimensional data or three-dimensional data, and the covariance matrices R thereof are as follows:

or

Wherein the position measurement data may be two-dimensional data, R₁₂Is a target covariance indicating that the position measurement data correspondsThe target covariance R is the covariance between the two components of the measurement error₁₂Based on the position measurement data, the first variance R₁₁And a second variance R₂₂Determining;

the position measurement data may be three-dimensional data, R₁₂Is a first target covariance representing a covariance between a first dimension component (x) and a second dimension component (y) of a rectangular coordinate conversion measurement error corresponding to the position measurement data, and is R₁₂Based on the position measurement data, the first variance R₁₁And a second variance R₂₂Determination of R₁₃A second target covariance representing a covariance between a first dimensional component (x) and a third dimensional component (z) of a rectangular coordinate conversion measurement error corresponding to the position measurement data, the second target covariance R₁₃Based on the position measurement data, the first variance R₁₁And a third variance R₃₃Determination of R₂₃A third target covariance representing a covariance between the second dimensional component (y) and the third dimensional component (z) of a rectangular coordinate transformation measurement error corresponding to the position measurement data, the third target covariance R₂₃Based on the position measurement data, the second variance R₂₂And a third variance R₃₃And (4) determining.

The covariance of the rectangular coordinate conversion measurement error of the position measurement data can be obtained through the diagonalization relation of the covariance matrix, the covariance estimation precision of the rectangular coordinate conversion measurement error of the sensor position measurement data can be improved through the method for determining the covariance, the statistical characteristics of the position measurement data can be described more accurately, and therefore the performance of functions of state estimation, data association, occupied grid estimation and the like of the position measurement data is greatly improved under application scenes of tracking, navigation, environmental perception and the like.

It should be noted that the implementation steps of the present invention do not need to know the elements of the diagonal matrix D in the diagonalized relation, that is, in the present embodiment, the process and the result of determining the covariance of the measurement error of the position measurement data transformation do not need to solve the diagonal matrix D as a necessary step. In contrast, the invention is used to determine synergyAfter the variance matrix, the diagonal matrix D may be determined further using the diagonalization relationship described above. Therefore, the diagonal matrix can be used for determining an elliptical area of a measurement error, and correspondingly, the diagonal matrix can be applied to the estimation of the occupied grid to improve the precision of the occupied grid, so that the method can be selected optionally based on the covariance matrix R, the orthogonal matrix U and the diagonalization relation D ═ U^TRU derives the diagonal matrix D, which determines the elliptical area of measurement error and may also reduce its performance loss in the application scenario.

Referring to fig. 3a, fig. 3a is a schematic flow chart of another method for determining covariance according to an embodiment of the present application, which includes, but is not limited to, the following steps:

step 301: position measurement data from at least one sensor is acquired.

In accordance with step 101 described above.

Step 302: and calculating to obtain the target covariance according to the position measurement data, the first variance and the second variance.

The position measurement data is two-dimensional data, and the embodiment provides a method for determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data.

Specifically, the position measurement data may be rectangular coordinate position data x and y or polar coordinate position data (d, θ), where d represents a distance of the object from the sensor and θ represents an azimuth angle of the object from the sensor, and accordingly, the statistical characteristic of the rectangular coordinate conversion measurement error of the position measurement data includes the first variance R₁₁And a second variance R₂₂The first variance R₁₁The second variance R is the variance of the first dimension component (x) of the rectangular coordinate transformation measurement error corresponding to the position measurement data₂₂The second dimension component (y) variance of the planar rectangular coordinate transformation measurement error corresponding to the position measurement data is shown. According to the aboveThe position measurement data and the first variance R₁₁And a second variance R₂₂Its target covariance can be determined as follows:

wherein R is₁₂For the target covariance sought, c_θ＝cosθ，s_θSin θ, θ is the azimuth angle of the target relative to the sensor, and

k is an integer. Alternatively, the first and second electrodes may be,

where x and y are the two components of the rectangular coordinate position data, respectively. Thus, c is_θAnd s_θCan be calculated by the angle value and also can be calculated by the rectangular coordinate value.

Step 303: and determining a covariance matrix of the rectangular coordinate conversion measurement error of the position measurement data according to the first variance, the second variance and the target covariance.

The first variance R can be found by the above steps₁₁The second variance R₂₂And a target covariance R₁₂Based on the position measurement data, a covariance matrix of rectangular coordinate transformation measurement errors of the position measurement data may be determined, where R is a covariance matrix as follows:

from the above covariance matrix, the data on the diagonal is the variance on each component, and the off-diagonal elements represent the covariance between the components of the transformed measurement error. According to the embodiment of the application, the position measurement data and the statistical characteristics of the rectangular coordinate conversion measurement error of the position measurement data are obtained, the covariance of the rectangular coordinate conversion measurement error of the position measurement data is determined by utilizing the position measurement data, and the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data is improved, so that the performance of functions such as state estimation, data association and occupied grid estimation of the position measurement data is greatly improved in application scenes such as tracking, navigation and environment perception.

On the other hand, the method for determining covariance provided in fig. 3a has a corresponding block diagram, and in particular, referring to fig. 3b, fig. 3b is a block diagram illustrating a method for determining covariance provided in an embodiment of the present application. As shown in fig. 3b, position measurement data is acquired, which contains rectangular coordinate position data describing the specific position information of the target relative to the sensor, such as (x, y); the rectangular coordinate position data may be converted from the polar coordinate position measurement data (d, θ) or directly output after the conversion of the sensor coordinates, although the measurement errors (distance d, azimuth θ) of the respective components of the polar coordinate position measurement data are statistically independent, the corresponding rectangular coordinate conversion measurement errors are statistically correlated, and the statistical characteristics (first difference R) of the respective components of the rectangular coordinate conversion measurement errors of the position measurement data are statistically correlated₁₁The second variance R₂₂) Insufficient to adequately describe the statistical properties between rectangular coordinate transformation measurement errors; according to the embodiment of the application, the target covariance R can be determined according to the position measurement data and the statistical characteristics of the rectangular coordinate conversion measurement error thereof₁₂Thereby obtaining a covariance matrix R of the rectangular coordinate conversion measurement error of the position measurement data.

Referring to fig. 4a, fig. 4a is a schematic flowchart of another method for determining covariance according to an embodiment of the present application, where the method includes, but is not limited to, the following steps:

step 401: position measurement data from at least one sensor is acquired.

In accordance with step 101 described above.

Step 402: and calculating to obtain a first target covariance according to the position measurement data, the first variance and the second variance.

The present embodiment provides a method for determining a covariance of a rectangular coordinate transformation measurement error of position measurement data, and a first target covariance representing a covariance between a first dimension component (x) and a second dimension component (y) of the rectangular coordinate transformation measurement error corresponding to the position measurement data can be determined according to the position measurement data and a statistical property of the rectangular coordinate transformation measurement error of the position measurement data.

Specifically, the position measurement data may be a three-dimensional rectangular coordinate position (x, y, z), where x, y, z are rectangular coordinate components of the target relative to the sensor; or a spherical coordinate position

Where d represents the distance of the target relative to the sensor, theta represents the azimuth angle of the target relative to the sensor,

the statistical characteristic of the rectangular coordinate conversion measurement error corresponding to the position measurement data includes a first variance R₁₁And a second variance R₂₂The first variance R₁₁The variance on the first dimension component (x) of the rectangular coordinate transformation measurement error corresponding to the position measurement data is shown, the second variance R₂₂The variance over the second dimension component (y) of the cartesian transformed measurement error for the position measurement data is shown. According to the position measurement data and the first variance R₁₁And a second variance R₂₂The first target covariance may be determined as follows:

wherein R is₁₂Is a first target covariance, c_θ＝cosθ，s_θ＝sinθ，c_2θIs cos (2 θ), θ is the azimuth angle of the target relative to the sensor, and

k is an integer. Alternatively, the first and second electrodes may be,

Step 403: and calculating to obtain a second target covariance according to the position measurement data, the first variance, the second variance and the third variance.

The statistical characteristic of the rectangular coordinate transformation measurement error of the position measurement data also comprises a third variance R₃₃The third variance R₃₃The variance of the third-dimensional component (z) of the rectangular coordinate transformation measurement error corresponding to the position measurement data is shown. According to the position measurement data and the first variance R₁₁The second variance R₂₂And a third variance R₃₃The second target covariance may be determined as follows:

wherein R is₁₃Is the second target covariance, R₁₃Showing the covariance between the first (x) and third (z) dimensional components of the cartesian transformation measurement error corresponding to the position measurement data,

c_θ＝cosθ，c_2θ＝cos(2θ)，

theta is the azimuth angle of the target relative to the sensor,

is the pitch angle of the target relative to the sensor, and

k and l are integers. Alternatively, the first and second electrodes may be,

where x, y and z are the three components of the rectangular coordinate position data, respectively. Thus, c is_θ、s_θ、

c_2θ、

Can be calculated by the angle value and also can be calculated by the rectangular coordinate value.

Step 404: and calculating to obtain a third target covariance according to the position measurement data, the first variance, the second variance and the third variance.

Similarly, based on the position measurement data and the first variance R₁₁The second variance R₂₂And a third variance R₃₃The third target covariance may be determined as follows:

wherein R is₂₃Is the third target covariance, R₂₃And a covariance between the second dimension component (y) and the third dimension component (z) representing a rectangular coordinate transformation measurement error corresponding to the position measurement data.

Step 405: and determining a covariance matrix of the rectangular coordinate conversion measurement error of the position measurement data according to the first variance, the second variance, the third variance, the first target covariance, the second target covariance and the third target covariance.

The first variance R can be found by the above steps₁₁The second variance R₂₂Third party diff R₃₃First target covariance R₁₂Second target covariance R₁₃And a third target covariance R₂₃From these known quantities, the diagonalized relation R ═ UDU of the covariance matrix can be used^TDetermining a covariance matrix of a rectangular coordinate transformation measurement error of the position measurement data, wherein R is the covariance matrix, D is a diagonal matrix, U is an orthogonal matrix obtained from the position measurement data, and U is a maximum value of the covariance matrix^TIs a transposed matrix of the orthogonal matrix. The covariance matrix is as follows:

from the covariance matrix, the diagonal data (R) is obtained₁₁、R₂₂、R₃₃) For variance on each component, data (R) on the off-diagonal₁₂、R₁₃、R₂₃) Representing the covariance between the different components of the cartesian transformation measurement error. According to the embodiment of the application, the statistical characteristics of the position measurement data and the rectangular coordinate conversion measurement error of the position measurement data are obtained, the covariance of the rectangular coordinate conversion measurement error of the position measurement data is determined according to the diagonalization relation of the covariance matrix, and the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data is improved, so that the performance of functions such as state estimation, data association and occupied grid estimation of the position measurement data is greatly improved under application scenes such as tracking, navigation and environmental perception.

On the other hand, the method for determining covariance provided in fig. 4a has a corresponding block diagram, and particularly, refer to fig. 4b, where fig. 4b is a block diagram of another method for determining covariance provided in the embodiment of the present application. As shown in FIG. 4bAcquiring position measurement data, wherein the position measurement data comprises rectangular coordinate position coordinates and is used for describing specific position information of the target, such as (x, y, z); the rectangular coordinate position data may be derived from spherical coordinate position measurement data

Converted or directly output after being converted by the sensor coordinate. Although the measurement error (distance d, azimuth angle theta, pitch angle) of each component of the spherical coordinate position measurement data

) Are statistically independent and the corresponding rectangular coordinate conversion measurement errors are statistically correlated, so that the statistical properties of the rectangular coordinate conversion measurement errors (first variance R) are₁₁The second variance R₂₂Third party diff R₃₃) Insufficient to adequately describe the statistical properties between rectangular coordinate transformation measurement errors; according to the embodiment of the application, the target covariance (R) is obtained according to the position measurement data and the statistical property of the rectangular coordinate conversion measurement error thereof₁₂、R₁₃、R₂₃) A covariance matrix R of the rectangular coordinate conversion measurement error of the position measurement data can be obtained.

Referring to fig. 5a, fig. 5a is a schematic flowchart of another method for determining covariance according to an embodiment of the present application, where the method includes, but is not limited to, the following steps:

step 501: position measurement data from at least one sensor is acquired.

In accordance with step 101 described above.

Step 502: and obtaining an angle matrix according to the position measurement data.

The position measurement data is three-dimensional data, and another method for determining covariance is provided in this embodiment, where an angle matrix may be obtained according to the position measurement data, where the angle matrix is used to determine covariance of rectangular coordinate transformation measurement error of the position measurement data, and the angle matrix may be as follows:

wherein A is an angle matrix,

to represent

c_θThe representation of the co s theta is,

to represent

s_θWhich is expressed in terms of sin theta and,

the pitch angle of the target relative to the sensor is shown, and theta is the azimuth angle of the measured target relative to the sensor; alternatively, the first and second electrodes may be,

Can be calculated by the angle value and also can be calculated by the rectangular coordinate value. The target may be an environment surrounding the sensor, such as an obstacle in the environment, or may be a moving target.

The angle matrix may also be a reasonable variation of the above equation a, which, as can be seen, mainly contains trigonometric values of the azimuth and elevation angles of the target with respect to the position of the sensor.

Step 503: and calculating to obtain the covariance of the rectangular coordinate conversion measurement error of the position measurement data according to the angle matrix, the first variance, the second variance and the third variance.

The angle matrix A is known through the above steps, and the statistical characteristics of the rectangular coordinate conversion measurement error of the position measurement data include the first variance R₁₁The second variance R₂₂And a third party difference R₃₃The first variance R₁₁The variance on the first dimension component (x) of the rectangular coordinate transformation measurement error corresponding to the position measurement data is shown, the second variance R₂₂The variance on the second dimension component (y) of the rectangular coordinate transformation measurement error corresponding to the position measurement data is shown, the third variance R₃₃The variance in the third dimension component (z) of the cartesian transformation measurement error for the position measurement data is shown. According to the angle matrix A and the first square difference R₁₁The second variance R₂₂And a third variance R₃₃The covariance of the rectangular coordinate transformation measurement error of the position measurement data may be determined as follows:

wherein A is an angle matrix and R₁₂Is a first target covariance, R₁₃Is a second target covariance, R₂₃A third target covariance, the first target covariance representing a covariance between the first dimensional component and the second dimensional component of the cartesian coordinate conversion measurement error corresponding to the position measurement data, the second target covariance representing a covariance between the first dimensional component and the third dimensional component of the cartesian coordinate conversion measurement error corresponding to the position measurement data, and the third target covariance representing a covariance between the second dimensional component and the third dimensional component of the cartesian coordinate conversion measurement error corresponding to the position measurement data.

Furthermore, a covariance matrix can be obtained by using a diagonalization relationship of a covariance matrix of a rectangular coordinate transformation measurement error of the position measurement data based on the first target covariance, the second target covariance, the third target covariance, the first variance, the second variance, and the third variance. The implementation mode of determining the covariance of the rectangular coordinate conversion measurement error of the three-dimensional position measurement data through the angle matrix can improve the covariance estimation precision of the rectangular coordinate conversion measurement error of the position measurement data, so that the accuracy of the rectangular coordinate conversion measurement error statistical characteristic description of the position measurement data is greatly improved.

On the other hand, fig. 5a also provides a corresponding block diagram of the method for determining covariance, and in particular, refer to fig. 5b, where fig. 5b is a block diagram of another method for determining covariance according to an embodiment of the present application. As shown in fig. 5b, position measurement data is obtained, which contains rectangular coordinate position data describing specific position information of the target, such as (x, y, z); the rectangular coordinate position data may be derived from spherical coordinate position measurement data

The data is obtained through conversion, or is directly output after the sensor coordinate conversion; obtaining an angle matrix a according to the position measurement data, specifically, the angle matrix a is described in detail in the step 502; despite the measurement errors (distance d, azimuth angle theta, pitch angle) in the spherical coordinate system on the respective components

) Are statistically independent, and the corresponding rectangular coordinate transformation measurement errors are statistically correlated, so that the statistical properties (first variance R) of each component of the rectangular coordinate transformation measurement errors₁₁The second variance R₂₂Third party diff R₃₃) Insufficient to adequately describe the statistical properties between rectangular coordinate transformation measurement errors; the embodiments of the present application may convert the statistical characteristics (R) of each component of the measurement error according to the position measurement data and the rectangular coordinates of the position measurement data₁₁、R₂₂、R₃₃) And the angle matrix A is used for obtaining the covariance [ R ] of the rectangular coordinate conversion measurement error of the position measurement data₁₂，R₁₃，R₂₃]Thereby obtaining a covariance matrix R of the rectangular coordinate transformation measurement error of the position measurement data.

The sensors involved in the embodiments of the present application may include a large class of sensors based on time of flight (ToF) measurements, such as radar like millimeter wave radar, ultrasonic radar/sonar, or lidar, etc. The embodiment of the application can be applied to the sensor, an advanced driving auxiliary system, an automatic driving system, an unmanned aerial vehicle system or other systems with the sensor, the sensor can be a data processing part of a sensing system, and a specific bearing platform of the sensor system can be vehicle-mounted such as an automobile, a motorcycle or a bicycle or airborne such as an unmanned aerial vehicle, a helicopter or a jet plane.

Referring to fig. 6, fig. 6 is a schematic view of a sensor application scenario provided in the embodiment of the present application. As shown in fig. 6, the vehicle supporting the automatic driving function includes a sensor system loaded on the vehicle, which includes a plurality of sensors designed based on the time-of-flight principle, specifically, a forward lidar, a forward millimeter wave radar, a lateral millimeter wave radar, a lidar, and a backward lidar, wherein the radars (sensors) work together to obtain position measurement data of obstacles around the vehicle, and the sensor system further includes front and rear cameras for collecting image information of the surroundings of the vehicle body and presenting the image information to a user through a display screen in the vehicle.

Referring to fig. 7, fig. 7 is a schematic view of a possible sensor imaging scene according to an embodiment of the present disclosure. As shown in fig. 7, a vehicle a shown in fig. 6 is automatically driven on the road, and the sensor system running on the vehicle detects that a vehicle B is running at a certain speed in front of the left side of the vehicle body, and at this time, the vehicle B is within the safety detection range of the vehicle a, and the sensor system of the vehicle a is determined as the detected object which may have a driving risk. The forward lidar configured on the vehicle a will emit a plurality of pulse laser signals and scan a plurality of detection regions of the object field at a plurality of scanning angles to form an image of the object field, and in particular, the forward lidar will also emit a plurality of pulse laser signals to the target (vehicle B) to be detected to acquire position measurement data of the vehicle B at a plurality of scanning angles, as shown in the following table:

the measured target image of the object space view field is formed by the position measurement data, and the specific process can be completed by a data processing module of the sensor system. On the other hand, in order to ensure that the position measurement data in the spherical coordinate system is accurate and the measured data in the spatial rectangular coordinate system can still be accurately located, the sensor needs to improve the covariance accuracy of the collected position measurement data, and the method for determining the covariance provided in fig. 1 or fig. 4a or fig. 5a may be used herein to improve the accuracy of the position measurement data.

The method of the embodiments of the present application is explained in detail above, and the apparatus of the embodiments of the present application is provided below.

Referring to fig. 8, fig. 8 is a schematic structural diagram of an apparatus for determining covariance according to an embodiment of the present application, where the apparatus for determining covariance may include an obtaining unit 801 and a determining unit 802, where the units are described as follows:

an acquisition unit 801 for acquiring position measurement data from at least one sensor;

a determining unit 802, configured to determine a covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the position measurement data and a statistical characteristic of the rectangular coordinate transformation measurement error.

In one possible embodiment, in determining the covariance of the cartesian-transformation-measurement error of the position measurement data, the determining unit 802 is specifically configured to determine the covariance of the cartesian-transformation-measurement error of the position measurement data according to a diagonalized relationship of a covariance matrix of the cartesian-transformation-measurement error, where the diagonalized relationship is determined according to the position measurement data.

In the embodiment of the present application, the covariance of the cartesian coordinate conversion measurement error of the position measurement data is determined according to a diagonalization relation of a covariance matrix of the cartesian coordinate conversion measurement error based on the statistical properties of the position measurement data and the cartesian coordinate conversion measurement error, wherein the diagonalization relation may be an eigenvalue decomposition form of the covariance matrix of the cartesian coordinate conversion measurement error, such as R ═ UDU ═ u^TOr U^TRU ═ D, or a reasonable variation of the above diagonalized form, where R is the covariance matrix, D is the diagonal matrix, U is the orthogonal matrix determined from the position measurement data^TIs a transpose of the orthogonal matrixAnd (4) matrix.

In one possible embodiment, the position measurement data is two-dimensional data, and the statistical characteristic includes a first variance or standard deviation or root mean square error of a first dimension component in the rectangular coordinate conversion measurement error of the position measurement data and a second variance or standard deviation or root mean square error of a second dimension component in the rectangular coordinate conversion measurement error of the position measurement data;

in terms of determining the covariance of the cartesian coordinate conversion measurement error of the position measurement data according to the position measurement data and the statistical characteristics of the cartesian coordinate conversion measurement error, the determining unit 802 is further configured to determine the covariance matrix of the cartesian coordinate conversion measurement error of the position measurement data according to the position measurement data, the first variance, standard deviation or root-mean-square error, and the second variance, standard deviation or root-mean-square error.

In one possible embodiment, the covariance matrix of the rectangular coordinate transformation measurement error of the position measurement data includes:

In one possible embodiment, the R group₁₂The method comprises the following steps:

In one possible embodiment, the position measurement data is three-dimensional data, and the statistical characteristics include a first variance or standard deviation or root mean square error of a first dimension component in the position measurement data rectangular coordinate conversion measurement error, a second variance or standard deviation or root mean square error of a second dimension component in the position measurement data rectangular coordinate conversion measurement error, and a third variance or standard deviation or root mean square error of a third dimension component in the position measurement data rectangular coordinate conversion measurement error;

in terms of determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the statistical characteristics of the position measurement data and the rectangular coordinate transformation measurement error, the determining unit 802 is further configured to determine the covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the position measurement data, the first variance, the first standard deviation, the second variance, the second standard deviation, the third variance, the third standard deviation, the third square deviation.

In a possible embodiment, in determining the covariance of the cartesian coordinate transformation measurement error of the position measurement data according to the position measurement data, the first variance, the second variance, the third variance, the first target covariance, the second target covariance, and the third target covariance, the determination unit 802 is further configured to determine a covariance matrix of the cartesian coordinate transformation measurement error of the position measurement data according to the first variance, the second variance, the third variance, the first target covariance, the second target covariance, and the third target covariance;

In a possible embodiment, in determining the covariance of the cartesian-transformation measurement error of the position measurement data according to the position measurement data, the first variance, the second variance, the third variance, and an angle matrix, the angle matrix being determined by trigonometric values of an azimuth angle and/or a pitch angle, the determining unit 802 is further configured to determine the covariance of the cartesian-transformation measurement error of the position measurement data according to the first variance, the second variance, the third variance, and the angle matrix.

In one possible embodiment, the angle matrix comprises:

wherein A is the angle matrix, the

To represent

C is mentioned_θRepresents cos θ, said

To represent

S is_θDenotes sin θ, said

Is a pitch angle and theta is an azimuth angle.

In one possible embodiment, the covariance of the rectangular coordinate transformation measurement error of the position measurement data includes:

In a possible implementation, the determining unit 802 is further configured to determine a diagonal matrix according to the covariance matrix and an orthogonal matrix, wherein diagonal elements of the diagonal matrix are variances of statistically independent errors, and the orthogonal matrix is determined by the position measurement data.

In one possible embodiment, the position measurement data is data in one of the following coordinate systems: rectangular coordinate system, polar coordinate system, spherical coordinate system.

In one possible embodiment, the position measurement data includes a distance, an azimuth angle, and a pitch angle; alternatively, the position measurement data includes range and azimuth.

According to the embodiment of the present application, the units in the apparatus shown in fig. 8 may be respectively or entirely combined into one or several other units to form a structure, or some unit(s) therein may be further split into multiple functionally smaller units to form a structure, which may achieve the same operation without affecting the achievement of the technical effect of the embodiment of the present application. The units are divided based on logic functions, and in practical application, the functions of one unit can be realized by a plurality of units, or the functions of a plurality of units can be realized by one unit. In other embodiments of the present application, the terminal-based terminal may also include other units, and in practical applications, these functions may also be implemented by being assisted by other units, and may be implemented by cooperation of multiple units.

It should be noted that the implementation of each unit may also correspond to the corresponding description of the method embodiments shown in fig. 1, fig. 3a, fig. 4a, and fig. 5 a.

In the device for determining covariance depicted in fig. 8, by acquiring statistical characteristics of position measurement data and a rectangular coordinate transformation measurement error, and determining covariance of the rectangular coordinate transformation measurement error of the position measurement data according to diagonalization relation of a covariance matrix, covariance estimation accuracy of the rectangular coordinate transformation measurement error of the position measurement data of the sensor is improved, and statistical characteristics of the position measurement data are described more accurately, so that performance of functions such as state estimation, data association, and occupancy grid estimation of the position measurement data is greatly improved in application scenarios such as tracking, navigation, and environmental awareness.

Referring to fig. 9, fig. 9 is a schematic structural diagram of an apparatus 90 for determining covariance according to an embodiment of the present application, where the apparatus 90 for determining covariance may include a memory 901 and a processor 902. Further optionally, a bus 903 may be included, wherein the memory 901 and the processor 902 are connected via the bus 903.

The memory 901 is used to provide a storage space, and data such as an operating system and a computer program may be stored in the storage space. The memory 901 includes, but is not limited to, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), or a portable read-only memory (CD-ROM).

The processor 902 is a module for performing arithmetic operations and logical operations, and may be one or a combination of plural kinds of processing modules such as a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a microprocessor unit (MPU), or the like.

The memory 901 stores a computer program, and the processor 902 calls the computer program stored in the memory 901 to perform the following operations:

acquiring position measurement data from at least one sensor;

In one possible implementation, in determining the covariance of the cartesian transformation measurement error of the position measurement data, the processor 902 is specifically configured to: and determining the covariance of the rectangular coordinate conversion measurement error of the position measurement data according to the diagonalization relation of the covariance matrix of the rectangular coordinate conversion measurement error, wherein the diagonalization relation is determined according to the position measurement data.

In the embodiment of the present application, the covariance of the cartesian coordinate conversion measurement error of the position measurement data is determined according to a diagonalization relation of a covariance matrix of the cartesian coordinate conversion measurement error based on the statistical properties of the position measurement data and the cartesian coordinate conversion measurement error, wherein the diagonalization relation may be an eigenvalue decomposition form of the covariance matrix of the cartesian coordinate conversion measurement error, such as R ═ UDU ═ u^TOr U^TRU ═ D, or a reasonable variation of the above diagonalized form, where R is the covariance matrix and D is the diagonal matrixU is an orthogonal matrix determined from position measurement data, U^TIs a transpose of the orthogonal matrix.

in determining the covariance of the cartesian coordinate transformation measurement error of the position measurement data according to the position measurement data and the statistical property of the cartesian coordinate transformation measurement error, the processor 902 is specifically configured to: and determining a covariance matrix of rectangular coordinate conversion measurement errors of the position measurement data according to the position measurement data, the first variance or standard deviation or root mean square error and the second variance or standard deviation or root mean square error.

In one possible embodiment, the position measurement data is three-dimensional data, and the statistical characteristics include a first variance or standard deviation or root mean square error of a first dimension component in the position measurement data rectangular coordinate conversion measurement error, a second variance or standard deviation or root mean square error of a second dimension component in the position measurement data rectangular coordinate conversion measurement error, and a third variance or standard deviation or root mean square error of a third dimension component in the position measurement data rectangular coordinate conversion measurement error:

in determining the covariance of the cartesian coordinate transformation measurement error of the position measurement data according to the position measurement data and the statistical property of the cartesian coordinate transformation measurement error, the processor 902 is specifically configured to: and determining the covariance of the rectangular coordinate conversion measurement error of the position measurement data according to the position measurement data, the first variance or standard deviation or root mean square error, the second variance or standard deviation or root mean square error and the third variance or standard deviation or root mean square error.

In one possible implementation, in determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the position measurement data, the first variance, the second standard deviation, the third variance, the third standard deviation, the third root mean square error, the processor 902 is specifically configured to: determining a covariance matrix of the rectangular coordinate transformation measurement error of the position measurement data according to the first variance, the second variance, the third variance, a first target covariance, a second target covariance and a third target covariance;

In one possible implementation, in determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the position measurement data, the first variance, the second standard deviation, the third variance, the third standard deviation, the third root mean square error, the processor 902 is specifically configured to: and determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the first variance, the second variance, the third variance and an angle matrix, wherein the angle matrix is determined by the trigonometric function value of the azimuth angle and/or the pitch angle.

In one possible embodiment, the angle matrix comprises:

wherein A is the angle matrix, the

To represent

C is mentioned_θRepresents cos θ, said

To represent

S is_θDenotes sin θ, said

Is a pitch angle and theta is an azimuth angle.

In the embodiment of the present application, the position measurement data is three-dimensional position measurement data, and another implementation manner for determining the covariance of the rectangular coordinate transformation measurement error of the three-dimensional position measurement data is provided, that is, the position measurement data is three-dimensional position measurement data, according to a specific implementation manner of an angle matrix, where the angle matrix is determined according to a trigonometric function value of an azimuth angle and/or a pitch angle, the trigonometric function value of the azimuth angle and/or the pitch angle can be obtained according to the position measurement data, specifically, a pitch angle of a target relative to a sensor, and θ is an azimuth angle of the target relative to the sensor, and the target can be an environment around the sensor, such as an obstacle in the environment, or can be a moving target; further:

In a possible implementation, the processor 902 is further specifically configured to: and determining a diagonal matrix according to the covariance matrix and an orthogonal matrix, wherein diagonal elements of the diagonal matrix are the variances of statistically independent errors, and the orthogonal matrix is determined by the position measurement data.

It should be noted that the specific implementation of the apparatus for determining covariance may also correspond to the corresponding description of the method embodiments shown in fig. 1, fig. 3a, fig. 4a and fig. 5 a.

In the covariance determining apparatus 90 described in fig. 9, by obtaining statistical characteristics of the position measurement data and the rectangular coordinate transformation measurement error, and determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the diagonalization relationship of the covariance matrix, the covariance estimation accuracy of the rectangular coordinate transformation measurement error of the position measurement data of the sensor is improved, and the statistical characteristics of the position measurement data are described more accurately, so that the performance of functions of state estimation, data association, and occupancy grid estimation of the position measurement data is greatly improved in application scenarios such as tracking, navigation, and environmental awareness.

Embodiments of the present application further provide a computer-readable storage medium, in which a computer program is stored, and when the computer program runs on one or more processors, the method for determining covariance shown in fig. 1, fig. 3a, fig. 4a, and fig. 5a may be implemented.

Embodiments of the present application further provide a computer program product, which when run on a processor, can implement the method for determining covariance shown in fig. 1, 3a, 4a and 5 a.

Embodiments of the present application also provide a sensor system that includes at least one sensor based on time-of-flight measurements. The sensor may comprise the means for determining covariance shown in fig. 8, or the means for determining covariance shown in fig. 9. Further optionally, the sensor system may further comprise at least one of: the system comprises at least one camera, at least one millimeter wave radar and at least one ultrasonic radar.

An embodiment of the present application further provides a chip system, where the chip system includes at least one processor and an interface circuit. Optionally, the interface circuit and the at least one processor are interconnected through a line, and the interface circuit is used for connecting an external device to the processor. The chip system may further comprise at least one memory storing a computer program, or the interface circuit may be adapted to provide the at least one processor with a computer program stored in an external memory; the computer program, when executed by the at least one processor, is adapted to implement the method flows shown in fig. 1, 3a, 4a and 5 a.

The embodiment of the application further provides a terminal, which can be a transport means or an intelligent device and comprises an unmanned aerial vehicle, an unmanned transport vehicle, an automobile or a robot and the like, wherein the terminal comprises at least one of the covariance determination device, the sensor system, the chip system and the like.

Specifically, the terminal is a vehicle, and the vehicle further includes at least one of at least one sensor, a vehicle body, an engine, an energy source, wheels, a vehicle control system, peripheral devices (such as an on-board computer, a microphone, a speaker, and the like).

In summary, by obtaining the statistical characteristics of the position measurement data and the rectangular coordinate transformation measurement error, and determining the covariance of the rectangular coordinate transformation measurement error of the position measurement data according to the diagonalization relationship of the covariance matrix, the covariance estimation precision of the rectangular coordinate transformation measurement error of the sensor position measurement data is improved, and the statistical characteristics of the position measurement data are described more accurately, so that the performance of functions such as state estimation, data association, occupied grid estimation and the like of the position measurement data is greatly improved in application scenes such as tracking, navigation, environmental perception and the like.

One of ordinary skill in the art will appreciate that all or part of the processes in the methods of the above embodiments can be implemented by hardware associated with a computer program, which can be stored in a computer-readable storage medium, and when executed, can include the processes of the above method embodiments. And the aforementioned storage medium includes: various media that can store computer program code, such as a read-only memory ROM or a random access memory RAM, a magnetic disk, or an optical disk.

Claims

1. A method of determining covariance, comprising:

acquiring position measurement data from at least one sensor;

2. The method of claim 1, wherein determining the covariance of the Cartesian conversion measurement error of the position measurement data comprises:

3. The method of claim 1 or 2, wherein the position measurement data is two-dimensional data, and the statistical properties include a first variance or standard deviation or root mean square error of a first dimension component of the position measurement data rectangular coordinate converted measurement error and a second variance or standard deviation or root mean square error of a second dimension component of the position measurement data rectangular coordinate converted measurement error;

4. The method of claim 3, wherein the covariance matrix of the position measurement data rectangular coordinate transformation measurement errors comprises:

wherein R is a covariance matrix of the rectangular coordinate transformation measurement error, and R is₁₁Is the first variance, the R₂₂Is the second variance, the R₁₂Representing a covariance between the first dimension component and the second dimension component of the position measurement data rectangular coordinate transformation measurement error, the covariance between the first dimension component and the second dimension component of the rectangular coordinate transformation measurement error being derived from the position measurement data, the first variance, and the second variance.

5. The method of claim 4, wherein R is₁₂The method comprises the following steps:

wherein, R is₁₂C is a covariance between the first dimension component and the second dimension component representing a rectangular coordinate transformation measurement error of the position measurement data_θRepresents cos θ, said s_θDenoted sin θ, said angle θ being the azimuth angle.

6. The method of claim 1 or 2, wherein the position measurement data is three-dimensional data, and the statistical properties include a first variance or standard deviation or root mean square error of a first dimension component in the position measurement data rectangular coordinate conversion measurement error, and a second variance or standard deviation or root mean square error of a second dimension component in the position measurement data rectangular coordinate conversion measurement error, and a third variance or standard deviation or root mean square error of a third dimension component in the position measurement data rectangular coordinate conversion measurement error;

7. The method of claim 6, wherein determining the covariance of the Cartesian conversion measurement errors for the position measurement data based on the position measurement data and the first variance or standard deviation or root mean square error, and the second variance or standard deviation or root mean square error, and the third variance or standard deviation or root mean square error comprises:

8. The method of claim 7, wherein the covariance matrix of the position measurement data rectangular coordinate transformation measurement errors comprises:

9. The method of claim 6, wherein determining the covariance of the Cartesian conversion measurement errors for the position measurement data based on the position measurement data and the first variance or standard deviation or root mean square error, and the second variance or standard deviation or root mean square error, and the third variance or standard deviation or root mean square error comprises:

10. The method of claim 9, wherein the angle matrix comprises:

wherein A is the angle matrix, the

To represent

C is mentioned_θRepresents cos θ, said

To represent

S is_θDenotes sin θ, said

Is a pitch angle and theta is an azimuth angle.

11. The method of claim 9 or 10, wherein the covariance of the position measurement data rectangular coordinate transformation measurement error comprises:

wherein A is the angle matrix and R is₁₁Is the first variance, the R₂₂Is the second variance, the R₃₃Is the third variance, the R₁₂Is a first target covariance, said R₁₃Is a second target covariance, said R₂₃A third target covariance, the first target covariance representing a covariance between the first dimension component and the second dimension component in the position measurement data rectangular coordinate transformation measurement error; the second target covariance represents a covariance between the first dimension component and the third dimension component in the position measurement data rectangular coordinate transformation measurement error; the third target covariance represents the second dimension component in the Cartesian coordinate conversion measurement error of the position measurement dataAnd the covariance between the third-dimensional components.

12. The method of claim 2, further comprising:

and determining a diagonal matrix according to the covariance matrix and the orthogonal matrix, wherein diagonal elements of the diagonal matrix are the variances of statistically independent errors, and the orthogonal matrix is determined according to the position measurement data.

13. A method according to any of claims 1-12, characterized in that the position measurement data is data in one of the following coordinate systems: rectangular coordinate system, polar coordinate system, spherical coordinate system.

14. The method of any of claims 1-13, wherein the position measurement data includes a distance, an azimuth angle, and a pitch angle; alternatively, the position measurement data includes range and azimuth.

15. An apparatus for determining covariance, comprising:

16. The apparatus according to claim 15, wherein the determining unit is configured to determine the covariance of the cartesian coordinate transformation measurement error of the position measurement data based on a diagonalized relationship of a covariance matrix of the cartesian coordinate transformation measurement error, wherein the diagonalized relationship is determined based on the position measurement data.

17. The apparatus of claim 15 or 16, wherein the position measurement data is two-dimensional data, and the statistical characteristic comprises a first variance or standard deviation or root mean square error of a first dimension component of the position measurement data rectangular coordinate transformation measurement error and a second variance or standard deviation or root mean square error of a second dimension component of the position measurement data rectangular coordinate transformation measurement error;

the determining unit is specifically further configured to determine a covariance matrix of the rectangular coordinate transformation measurement error of the position measurement data according to the position measurement data, the first variance, the standard deviation, or the root-mean-square error, and the second variance, the standard deviation, or the root-mean-square error.

18. The apparatus of claim 17, wherein the covariance matrix of the position measurement data rectangular coordinate transformation measurement errors comprises:

19. The apparatus of claim 18, wherein R is₁₂The method comprises the following steps:

wherein, R is₁₂For representing said position measurement data rectangular seatsTransforming the covariance between said first dimension component and said second dimension component of the measurement error, c_θRepresents cos θ, said s_θDenoted sin θ, said angle θ being the azimuth angle.

20. The apparatus of claim 15 or 16, wherein the position measurement data is three-dimensional data, and the statistical properties comprise a first variance or standard deviation or root mean square error of a first dimension component of the position measurement data rectangular coordinate transformation measurement error, a second variance or standard deviation or root mean square error of a second dimension component of the position measurement data rectangular coordinate transformation measurement error, and a third variance or standard deviation or root mean square error of a third dimension component of the position measurement data rectangular coordinate transformation measurement error;

the determining unit is specifically further configured to determine a covariance of a rectangular coordinate transformation measurement error of the position measurement data according to the position measurement data, the first variance, the second standard deviation, the third variance, the third standard deviation, and the third root-mean-square error.

21. The apparatus according to claim 20, wherein the determining unit is further configured to determine a covariance matrix of the cartesian transformation measurement error of the position measurement data according to the first variance, the second variance, the third variance, a first target covariance, a second target covariance, and a third target covariance;

22. The apparatus of claim 21, wherein the covariance matrix of the position measurement data rectangular coordinate transformation measurement errors comprises:

23. The apparatus according to claim 20, wherein the determining unit is further configured to determine a covariance of the cartesian transformation measurement error of the position measurement data based on the first variance, the second variance, the third variance, and an angle matrix, the angle matrix being determined by trigonometric values of an azimuth angle and/or a pitch angle.

24. The apparatus of claim 23, wherein the angle matrix comprises:

wherein A is the angle matrix, the

To represent

C is mentioned_θRepresents cos θ, oThe above-mentioned

To represent

S is_θDenotes sin θ, said

Is a pitch angle and theta is an azimuth angle.

25. The apparatus of claim 23 or 24, wherein the covariance of the position measurement data rectangular coordinate transformation measurement error comprises:

26. The apparatus of claim 16, wherein the determining unit is further configured to determine a diagonal matrix according to the covariance matrix and an orthogonal matrix, wherein diagonal elements of the diagonal matrix are variances of statistically independent errors, and the orthogonal matrix is determined according to the position measurement data.

27. The apparatus according to any of claims 15-26, wherein the position measurement data is data in one of the following coordinate systems: rectangular coordinate system, polar coordinate system, spherical coordinate system.

28. The apparatus of any of claims 15-27, wherein the position measurement data comprises a distance, an azimuth angle, and a pitch angle; alternatively, the position measurement data includes range and azimuth.

29. A sensor comprising at least one of a sensing element for acquiring position measurement data from at least one sensor, a conversion element, a memory having a computer program stored therein, and a processor that invokes the computer program stored in the memory to perform the operations of:

acquiring position measurement data from at least one sensor;

30. A computer-readable storage medium, in which a computer program is stored which, when run on one or more processors, performs the method of any one of claims 1-14.