Method for determining at least one system state using a Kalman filter - Patents.com

A Kalman filter assembly with multiple filters compensates for configuration-specific errors, enhancing robustness and efficiency in sensor data fusion for vehicle positioning and autonomous driving.

JP7587054B2Active Publication Date: 2024-11-19ROBERT BOSCH GMBH
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
JP2023551184
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2021-02-24
Filing Date
2022-01-25
Publication Date
2024-11-19
Estimated Expiration
2042-01-25

AI Technical Summary

Technical Problem

Kalman filters are complex and require significant configuration efforts, making them difficult to utilize for new applications and maintain existing ones, especially when dealing with nonlinear dynamic processes and adaptive filters.

Method used

A method using a Kalman filter assembly comprising multiple Kalman filters with different configuration parameters that operate independently and fuse their estimation results to form an overall estimation, compensating for configuration-specific errors and improving robustness to model and measurement inaccuracies.

Benefits of technology

The method enhances the robustness of Kalman filter systems to errors and reduces the complexity and cost of configuration, enabling more reliable and efficient sensor data fusion for applications like vehicle positioning and autonomous driving.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007587054000101
    Figure 0007587054000101
  • Figure 0007587054000102
    Figure 0007587054000102
  • Figure 0007587054000103
    Figure 0007587054000103
Patent Text Reader

Abstract

The invention relates to a method for determining at least one system state by means of a Kalman filter assembly, which is supplied with at least one measurement value measured by at least one sensor of the system, the method comprising at least the following steps: a) performing a first estimation of a system state by means of a first Kalman filter of the Kalman filter assembly, in which a first estimation result and at least one associated first information on the reliability of the first estimation result are output; b) performing a second Kalman filter of the Kalman filter assembly, in which a first estimation of a system state is output and at least one associated first information on the reliability of the first estimation result is output; a) performing a second estimation of the system state using a Kalman filter, where the second estimation and at least one associated second information on the reliability of the second estimation are output, where the second Kalman filter differs from the first Kalman filter in at least one setting parameter; and c) fusing the first and second estimations into an overall estimation of the system state, as well as fusing the first information on the reliability of the first estimation and the second information on the reliability of the second estimation into overall information on the reliability of the overall estimation.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a method for determining at least one system state using a Kalman filter, a computer program for implementing the method, a machine-readable storage medium having the computer program stored thereon, as well as a system for determining the position of a moving object, in particular a vehicle or the like, configured to implement the method. The present invention is particularly applicable in the context of at least partially automated or autonomous driving.

[0002] Prior Art Kalman filters are usually used for iterative estimation of system states based on error-laden observations. In this context, Kalman filters have proven to be particularly suitable for applications in which sensor information from various sensors must be integrated (or fused) with model information. Furthermore, Kalman filters are frequently used in embedded systems, since their calculations are preferably accurate and robust. Besides, microcontrollers can preferably efficiently perform the calculations of Kalman filters.

[0003] The Kalman filter equations can be written in matrix notation as follows:

number

number

number

number

number

[0004] In particular, if for each measurement variable there exists a corresponding model variable with the same scaling, and / or conversely, if for each model variable there exists a corresponding measurement variable with the same scaling, then explicit equations with the notations K, Σ', Σ0, Σ1, μ', μ0, μ1 can be used. Otherwise, e.g., with the notations H, K', P, R,

number

[0005] Equations GL1 and GL2 describe the iterative estimation process of the Kalman filter, while equations GL3-GL5 describe the correction or fusion of the iteratively estimated model values ​​with the sensorily captured measurements. For a more detailed explanation, please refer to the description of a typical structure of a Kalman filter in relation to FIG. 1.

[0006] However, the Kalman filter is relatively complex and requires a large number of configuration measures (in particular the system matrix F k , the variance matrix of the measurement noise R k , and the system noise variance matrix Q k ) which makes it difficult to apply Kalman filters to new applications and / or maintain existing applications.

[0007] In some cases, so-called extended Kalman filters are used to be able to model nonlinear dynamic processes and / or additional filters are used during the run time to calculate a number of set measures, in particular the matrix R for measurement and system noise. k and Q kTo be able to automatically adapt the value of , so-called adaptive Kalman filters (or ROSE filters) are used. However, a relatively large effort here consists in designing or configuring the additive filter, and in this case configuring the ROSE filter with its own additive filter is even more difficult.

[0008] Furthermore, implementations of Kalman filters are known, such as ensemble Kalman filters, in which a matrix of possible state vectors is typically used instead of the state vectors and the covariance matrices belonging to the state vectors, and a matrix of possible measurement vectors is used instead of the measurement vectors and the covariance matrices belonging to the measurement vectors.

number

number

[0009] Disclosure of the Invention According to claim 1, a method is proposed for determining at least one system state by means of a Kalman filter assembly, the Kalman filter assembly being supplied with at least one measurement measured by at least one sensor of the system, the method comprising at least the following steps: a) performing a first estimation of a system state using a first Kalman filter of a Kalman filter assembly, in which a first estimation result and at least one associated first piece of information on the reliability of the first estimation result are output; b) performing a second estimation of the system state using a second Kalman filter of the Kalman filter assembly, whereby a second estimation result and at least one associated second piece of information on the reliability of the second estimation result are output, whereby the second Kalman filter differs from the first Kalman filter in at least one setting parameter; c) fusing the first estimation result and the second estimation result into an overall estimation result for the system state, and fusing first information regarding the reliability of the first estimation result and second information regarding the reliability of the second estimation result into overall information regarding the reliability of the overall estimation result.

[0010] The presented order of steps a), b) and c) is exemplary and may be performed at least once in the presented order, for example in a regular operating sequence, to perform the method. Alternatively or additionally, steps a), b) and c), in particular steps a) and b), may be performed at least partially in parallel or simultaneously. The method may be performed by a control device, such as a (micro)controller, which may be, for example, a component of a system also described herein.

[0011] The method advantageously allows for the identification or (automatically) correction of deviations from the reference situation during observation due to errors or large inaccuracies in model formation and / or configuration and / or faulty measurement values, even if the allocation of expected errors to model errors and / or measurement errors is unknown. This is achieved in particular in a Kalman filter assembly in that the variously configured Kalman filters first perform estimations independently of each other using measured values, in particular the same measured values, and then these estimation results are merged into an overall estimation result. This advantageously contributes to the fact that configuration-specific error sources, i.e. errors due to the configuration of the Kalman filter, can be compensated as much as possible, in particular by at least one further configured Kalman filter of the assembly.

[0012] The system may be a system for determining the position of an object, for example a vehicle (car). The system may be at least partially located inside or on the object, for example a vehicle. The vehicle may be configured for a preferably at least partially automated and / or autonomous driving mode, for example with a correspondingly configured control device. The control device may be connected to the system for receiving position data from the system. The system may include a plurality, in particular various or diverse sensors, or may be connected to the sensors of the vehicle. These sensors may include, for example, at least one GNSS sensor and one (optical or acoustic) surrounding environment sensor, for example a camera sensor, a LIDAR sensor, a RADAR sensor, an ultrasonic sensor, etc. The measurements of these sensors may be fused using the methods described herein or a Kalman filter assembly.

[0013] The method in particular contributes to (continuously) determining the system state by means of Kalman filter assemblies and depending on sensor data. These Kalman filter assemblies can be supplied with measurements from a number of different or diverse sensors of the system in order to carry out estimations taking these measurements into account. At least one system state can for example include the (instantaneous) (ego) position of an object or moving part that is moving, in particular movable along the surface of the earth, such as for example (automobiles), ships, planes, smartphones or smartwatches. Furthermore, the at least one system state can include the (instantaneous) speed, the (instantaneous) direction of movement and / or the (instantaneous) acceleration of the object. The use of the described method is fundamentally advantageous for any sensor data fusion task, such as for example vehicle positioning, object identification and / or driving dynamics control. Furthermore, the described method can be used for example for modeling the transmission characteristics of sensors to reduce measurement noise and other disturbing influences on the sensor signals.

[0014] The Kalman filter assembly may include a number of Kalman filters (each operating in a self-contained manner) that may operate independently of one another. In this case, these Kalman filters may operate independently of one another, but preferably on the same measurements. In particular, the Kalman filter assembly may operate at least partially in parallel or simultaneously with one another. This may in other words be described as, in particular, the Kalman filters in the assembly being arranged or connected in parallel with one another.

[0015] The equations of the individual Kalman filters of the Kalman filter assembly, similar to equations (1)-(5), can be presented as follows, where index i represents the number of the respective Kalman filter:

number

number

number

number

number

[0016] In step a), a step of performing a first estimation of the system state using a first Kalman filter (index: i=1) of the Kalman filter assembly is performed, where the first estimation result (where:

number

number

number

number

[0017] In this case, the second Kalman filter differs from the first Kalman filter in at least one setting parameter. If more than two Kalman filters are provided, then for example each of the further Kalman filters may accordingly differ from the other Kalman filters in at least one setting parameter. As setting parameters, in particular individual parameters or multiple parameters of the setting means mentioned at the beginning, in particular the system matrix F k , the variance matrix of the measurement noise R k , and the system noise variance matrix Q k ) are taken into account. These setting parameters may, for example, correspond to the entries of the variance or covariance matrix of the Kalman filter. It is further conceivable that the setting parameters correspond to coefficients for scaling the corresponding matrices and thus act on the variance or covariance matrix or contribute to the setting of the corresponding Kalman filter.

[0018] In particular, in steps a) and b), a plurality of mutually independent Kalman filters, including at least a first Kalman filter and a second Kalman filter, are used, in particular, to calculate a plurality of individual covariance matrices (wherein P i,k ), but with different assumptions about the sensor and / or model variables that may be subject to error, in particular process noise (Q i,k ), and / or the measurement noise covariance matrix (where R i,k ) can be calculated with separate assumptions about their independent state vectors (

number

number

[0019] For example, for multiple Kalman filters or all Kalman filters in a Kalman filter assembly, the following variables: -i: Kalman filter index, -j: number of Kalman filters, -F k For example, the system state is expressed as k-1 From time t k The transition matrix propagating to -H k : an observation matrix that maps n values ​​of the system state to m observations, and / or -

number

[0020] Preferably, the following variables (of the second Kalman filter): -

number

number

number

number

number

number

number

number

number

number

number

[0021] In step c), the first estimation result (where:

number

number

number

number

number

number

[0022] According to a preferred embodiment, it is proposed that the first and second Kalman filters are supplied with the same measurements. For example, the first and second Kalman filters can have measurement entries connected to the same sensor output. This means, in other words, that in particular for the first and second Kalman filters, the (new) observations (measurement vector:

number

[0023] According to a further preferred embodiment, it is proposed that in step c) the first estimation result is weighted with a first information on the reliability of the first estimation result and the second estimation result is weighted with a second information on the reliability of the second estimation result. In particular, each estimation result (where:

number

number

[0024] For example, the covariance matrix

number

number

number

[0025] In this case, the fused covariance matrix

number

number

[0026] According to a further preferred embodiment, global information on the reliability of the overall estimation result (where:

number

number

number

number

number

[0027] For example, different estimation results or Kalman filter model values

number

number

[0028] Therefore, the discrepancy D is, for example, the variance

number

number

number

number

number

[0029] The measure of disagreement is further defined as the disagreement weight E D and / or may be weighted, for example, by the following formula:

number

number

[0030] For example, the discrepancy weight E D or the weights E for covariance P can be multiplied element-wise as a matrix, or it can be a scalar variable.

number

[0031] In particular, the measure of discrepancy includes the following elements: -First estimation result, - The second estimation result, - Overall estimation results, - information on the reliability of the first estimation result, - information on the reliability of the second estimation result, and - overall information on the reliability of the overall estimation results, can be determined or calculated using (see Equation 13).

[0032] Furthermore, it may be assumed that the disparity or a measure of disparity is temporally filtered and / or element-wise scaled.

[0033] Alternatively or additionally to the calculation according to equations (11) to (14), for example the following formula:

number

number

[0034] In this regard, the output covariance matrix

number

number

number

[0035] Compared to equations (11)-(14), error identification and compensation may be less significantly noticeable (deviations due to errors may lead to slightly smaller or even excessively small variances), but equations (15) and (16) have the advantage that it is not necessary to calculate the inverse matrix for them, which advantageously saves calculation time.

[0036] Weight π i can be chosen statically or calculated dynamically, in which case the weights π i In the dynamic calculation of

number

number

number

[0037] In particular: an overall estimation result is calculated using the estimation results and weights of the first and second Kalman filters; It may be assumed that the overall information on the reliability of the overall estimation result is calculated using the weighted estimation results of the first and second Kalman filters, the (weighted) information on the reliability of the estimation results of the first and second Kalman filters, the (unweighted) estimation results of the first and second Kalman filters, and the weights (see equations 15 and 16).

[0038] According to a further preferred embodiment, the second Kalman filter is adapted to filter the measurement noise (R i,k It is proposed that the second Kalman filter differs from the first Kalman filter in at least one setting parameter of the covariance matrix of the measurement noise. This setting parameter can for example be an entry of the covariance matrix of the measurement noise or a coefficient for scaling the covariance matrix of the measurement noise. The setting parameter is in particular selected such that the covariance matrix of the measurement noise of the first Kalman filter differs from the covariance matrix of the measurement noise of the second Kalman filter (within the same time step).

[0039] According to a further preferred embodiment, the second Kalman filter is a process noise (Q i,k It is proposed that the second Kalman filter differs from the first Kalman filter in at least one setting parameter of the covariance matrix of the process noise. This setting parameter can be, for example, an entry of the covariance matrix of the process noise or a coefficient for scaling the covariance matrix of the process noise. The setting parameter is in particular selected such that the covariance matrix of the process noise of the first Kalman filter differs from the covariance matrix of the process noise of the second Kalman filter (within the same time step).

[0040] In this regard, the configuration of the Kalman filter of the Kalman filter assembly is determined by, among other things, the measurement noise and / or the process noise Q i,k Covariance matrix R i,kIn particular, for sensor values ​​whose errors cannot be identified and compensated for in advance with such certainty that their effect on the Kalman filter calculation is negligible, the individual covariance matrix R of the measurement noise i,k For example, in this regard, a separate Kalman filter having the measurement noise covariance matrix R (in a Kalman filter provided for this purpose) can be obtained or added to the assembly. i,k The diagonal entries of can be augmented by the squared error value.

number

number

[0041] By way of example, the process noise Q i,k The corresponding diagonal elements of can be increased.

[0042] For example, if only one potentially erroneous sensor value or only one potentially erroneous model value is to be compensated, it may be advantageous or sufficient to form a Kalman filter assembly having, illustratively, only two Kalman filters, where a first Kalman filter configuration is formed assuming that no faults are present and a second Kalman filter configuration is formed assuming that faults are present. For example, if erroneous values ​​are to be identified from a plurality of sensor and / or model values, in addition to each Kalman filter for each potentially erroneous sensor value, it may also be advantageous to implement one further Kalman filter for each possible combination of the plurality of erroneous sensor values.

[0043] According to a further preferred embodiment, it is proposed that at least the first information on the reliability of the first estimation result or the second information on the reliability of the second estimation result is corrected using a discrepancy between at least one model value belonging to the corresponding estimation and at least one measurement value belonging to the corresponding estimation. This is achieved by using the covariance matrix calculated by the Kalman filter for the calculation of the assembly

number

[0044] For example, Carmen filter individual discrepancies D i Before calculating the assembly information or overall information on the reliability of the overall estimation result using the covariance matrix

number

number

[0045] In this case, sub-situation W i,0 and W i,1 is the following formula:

number

number

[0046] In this case, for calculations of assemblies using equations (11)-(13) or global information on the reliability of the global estimation results, the covariance matrix

number

number

number

[0047] Furthermore, it is advantageous to further element-wise PT1 filter the discrepancies, so that the output covariance matrix

number

[0048] According to a further aspect, a computer program for implementing the method presented herein is proposed, in other words it particularly relates to a computer program product comprising instructions which, when said program is executed by a computer, cause said computer to carry out the method described herein.

[0049] According to a further aspect, a machine-readable storage medium is proposed, having stored thereon a computer program as proposed herein. Typically, a machine-readable storage medium is a data carrier that can be read by a computer.

[0050] According to a further aspect, a system for determining the position of a moving object, such as a vehicle, is also proposed, configured to perform the method described herein. The system for determining the position of a moving object may for example be envisaged and configured to determine the ego-position of the moving object and / or to perform position measurements, particularly relative to other moving moving objects, such as other road users. The system may for example include a motion and position sensor configured to perform the method described herein. Furthermore, the motion and position sensor may for example receive GNSS data and / or environmental sensor data (from environmental sensors of the moving object or the vehicle). The system may for example include a computing device, such as for example a (micro)controller, having access to a computer program, also described herein, for performing the method. In this connection, a storage medium may also for example be a component of the system or connected to the system.

[0051] The details, features and preferred embodiments described in relation to the method may also occur in the computer program and / or storage medium and / or system presented herein accordingly, and vice versa, to which extent reference is made in full to the description set forth herein for a more detailed characterization of the features.

[0052] In the following, the solutions presented in this specification and their technical environment will be described in more detail based on the drawings. Please note that the present invention should not be limited to the illustrated embodiments. In particular, unless otherwise explicitly indicated, partial aspects of the examples described in the drawings can be extracted and combined with other components and / or findings from other drawings and / or this specification. [Brief description of the drawings]

[0053] [Figure 1] FIG. 1 shows a schematic diagram of a typical signal flow diagram of a Kalman filter according to the prior art; [Diagram 2] FIG. 2 shows a schematic diagram of an exemplary sequence of the method presented herein. [Diagram 3] FIG. 2 shows a schematic diagram of a more detailed illustration of the method presented herein. [Figure 4] FIG. 2 shows a schematic diagram of an example of a probability density distribution that can be realized using the methods presented herein. [Diagram 5] FIG. 1 illustrates a schematic diagram of an exemplary system for determining a vehicle position.

[0054] A typical structure of a prior art Kalman filter is shown in FIG. 1. The Kalman filter equation underlying this structure is:

number

number

number

number

number

[0055] In this case, equation (GL1) represents the state vector of the preceding time step (iterative estimation).

number

number

number

number

number

[0056] Equation (GL2) relates to the estimated state vector

number

[0057] Equation (GL3) describes the so-called Kalman gain K or Kalman gain matrix K', which is a function of the covariance matrix P k , the observation matrix H k , and the measurement noise covariance matrix R k It is formed based on the covariance matrix P k is the observation matrix H k Together, we can form the covariance matrix Σ0 of the model value vector μ0.

[0058] Equation (GL4) expresses the measurement vector z k or the state vector estimated using the measurements denoted by μ

number

number

[0059] Equation (GL5) is the state vector

number

number

[0060] Thus, equations (GL1) and (GL2) describe the iterative estimation process of the Kalman filter, which is marked with reference number 10 in FIG. 1. Equations (GL3)-(GL5) describe the subsequent correction or fusion of the iteratively estimated model values ​​with the measurements captured by the sensors, which is marked with reference number 20 in FIG. 1. This modified or fused (new) model value can be used in the subsequent iterative steps of the estimation process 10, which is indicated by the return arrow in FIG. 1.

[0061] 2 shows a schematic diagram of an exemplary sequence of the method presented herein, which is used for determining at least one system state using a Kalman filter assembly, which is fed with at least one measurement measured by at least one sensor of the system. The system can be, for example, a system for determining the position of a vehicle.

[0062] The order of steps a), b) and c) indicated by blocks 110, 120 and 130 is exemplary and may be performed, for example at least once, in the order shown for carrying out the method. Furthermore, steps a), b) and c), in particular steps a) and b), may also be performed at least partially in parallel or simultaneously.

[0063] In block 110, step a) is performed to perform a first estimation of the system state using a first Kalman filter of the Kalman filter assembly, whereby the first estimation result and at least one associated first information regarding the reliability of the first estimation result are output.

[0064] In block 120, step b) involves performing a second estimation of the system state using a second Kalman filter of the Kalman filter assembly, where the second estimation result and at least one associated second piece of information regarding the reliability of the second estimation result are output, where the second Kalman filter differs from the first Kalman filter in at least one setting parameter.

[0065] In this case, the first Kalman filter and the second Kalman filter may be fed with the same measurements.

[0066] For example, in this case, the second Kalman filter may differ from the first Kalman filter in at least one configuration parameter of the covariance matrix of the measurement noise.

[0067] Alternatively or cumulatively, in this case the second Kalman filter may differ from the first Kalman filter in at least one setting parameter of the covariance matrix of the process noise.

[0068] In this connection, it may be envisaged that at least one different setting parameter of the second Kalman filter acts on the covariance matrix of the measurement noise of the second Kalman filter and / or on the covariance matrix of the process noise of the second Kalman filter, for example in the type of coefficients for scaling the corresponding matrices.

[0069] In block 130, step c) includes fusing the first estimation result and the second estimation result into an overall estimation result for the system state, as well as fusing first information regarding the reliability of the first estimation result and second information regarding the reliability of the second estimation result into overall information regarding the reliability of the overall estimation result.

[0070] Here, the first estimation result may be weighted with first information regarding the reliability of the first estimation result, and the second estimation result may be weighted with second information regarding the reliability of the second estimation result.

[0071] Furthermore, the overall information regarding the reliability of the overall estimation result can be refined using a measure of discrepancy between the first estimation result and the second estimation result.

[0072] Furthermore, it may be assumed that at least the first information regarding the reliability of the first estimation result or the second information regarding the reliability of the second estimation result is corrected using a discrepancy between at least one model value belonging to the estimation in question and at least one measured value belonging to the estimation in question.

[0073] 3 shows a schematic diagram of a more detailed example of the method presented herein. In the Kalman filter assembly, in step a) and in step b), one Kalman filter estimation is performed respectively, with the corresponding output: First estimation result

number

number

number

number

number

number

[0074] 3 may differ from one another in their configuration, e.g., by being configured or designed using different assumptions about the sensor accuracy. In this regard, the exemplary second Kalman filter may be configured to use the measurement noise covariance matrix R k The second Kalman filter may differ from the first Kalman filter in at least one configuration parameter of the first Kalman filter. Briefly, the first Kalman filter may assume that there is no sensor error, and the second Kalman filter may assume that there is a sensor error.

[0075] 4 shows a schematic diagram of an example of a probability density distribution that can be realized using the methods presented herein. The depiction is directed to the example Kalman filter assembly of FIG.

[0076] 5 shows a schematic diagram of an exemplary system 1 for determining the position of a moving object 2, such as a vehicle, which is envisaged and arranged to carry out the methods described herein.

[0077] The methods described herein as well as the systems described herein, inter alia, enable one or more of the following advantages: The Kalman filter assembly is preferably more robust to model and measurement errors. Cost savings in the development and parameterization of Kalman filters, especially in model formation, development and parameterization of sensor fault identification and compensation. The discrepancy calculation can preferably be applied to the output of the Kalman filter extensions EKF (Extended Kalman Filter) and UKF (Unscented Kalman Filter) as well as in the well-known Ensemble Kalman Filter EnKF (Ensemble Kalman Filter).

Claims

1. 1. A method for determining at least one system state using a Kalman filter assembly, the Kalman filter assembly being supplied with at least one measurement measured by at least one sensor of a system; The method comprises at least the following steps: a) performing a first estimation of the system state using a first Kalman filter of the Kalman filter assembly, whereby a first estimation result and at least one associated first piece of information regarding the reliability of the first estimation result are output; b) performing a second estimation of the system state using a second Kalman filter of the Kalman filter assembly, outputting a second estimation result and at least one associated second piece of information on the reliability of the second estimation result, the second Kalman filter differing from the first Kalman filter in at least one setting parameter; c) fusing the first estimation result and the second estimation result into an overall estimation result for the system state, and fusing the first information on the reliability of the first estimation result and the second information on the reliability of the second estimation result into overall information on the reliability of the overall estimation result; The method of claim 1, wherein the overall information regarding the reliability of the overall estimation result is modified using a measure of discrepancy between the first estimation result and the second estimation result.

2. The method of claim 1 , wherein the first Kalman filter and the second Kalman filter are provided with the same measurements.

3. 3. The method according to claim 1, wherein in step c) the first estimation result is weighted with the first information regarding the reliability of the first estimation result and the second estimation result is weighted with the second information regarding the reliability of the second estimation result.

4. 4. The method according to claim 1, wherein the second Kalman filter differs from the first Kalman filter in at least one setting parameter of a covariance matrix of the measurement noise.

5. 5. The method according to claim 1, wherein the second Kalman filter differs from the first Kalman filter in at least one setting parameter of a covariance matrix of the process noise.

6. 6. The method according to claim 1, wherein at least the first information on the reliability of the first estimation result or the second information on the reliability of the second estimation result is corrected using a discrepancy between at least one model value belonging to the estimation in question and at least one measured value belonging to the estimation in question.

7. A computer program for carrying out the method according to any one of claims 1 to 6.

8. 8. A machine-readable storage medium having stored thereon the computer program of claim 7.

9. A system (1) for determining the position of a moving object (2), said system (1) being configured to implement a method according to any one of claims 1 to 6.

Citation Information

Patent Citations

  • Interactive multi-model tracking method based on self-adaptive transition probability matrix

    CN104880707A

  • State estimating apparatus and method

    JP2009031096A

  • Method and apparatus for calculating position

    JP2010169640A

  • Multiple model radar tracking filter and systems and methods employing same

    US20050128138A1

  • System and method for fusing asynchronous sensor tracks in a track fusion application

    US20200312156A1