Method for checking a correction of a system model in a Kalman filter

Abstract

Method for checking a correction of a system model in a Kalman filter (1) which is part of a filter network (2) for determining localization data in a motor vehicle and uses a first set (3) of sensor data for determining localization data, wherein the method comprises at least the following steps: a) Detecting a standstill situation of the motor vehicle during operation with a standstill detection system (12), so that a standstill signal (13) is generated; b) Performing a calibration function for parameters of the system model during the standstill situation in the Kalman filter (1), whereby corrected parameters (5) are determined; c) Performing a check of the corrected parameters (5) by comparing the corrected parameters (5) with comparison parameters (6) when a standstill signal (13) is present, wherein the comparison parameters (6) are determined with another data source (10), the other data source (10) generating comparison parameters (6) using a second set (4) of sensor data, the second set (4) of sensor data being reduced compared to the first set (3) of sensor data. d) Execute an error function (7) if a check performed in step c) fails.

Description

State of the art

[0001] The invention relates to a novel method for checking a parameterization of a system model in a Kalman filter.

[0002] The Kalman filter (also Kalman-Bucy filter, Stratonovich-Kalman-Bucy filter or Kalman-Bucy-Stratonovich filter) is a mathematical filter model for the iterative estimation of system states based on faulty input data, which are in particular sensor data.

[0003] The Kalman filter is used to estimate system variables that cannot be measured directly, while optimally reducing errors in observations. Based on the input data, the Kalman filter typically maintains an internal mathematical model as a constraint. This model is incorporated into the parameter estimation and takes into account dynamic relationships between the system variables. For example, the mathematical model contains equations of motion that relate the changing positions and velocities, which are input data to the Kalman filter, to each other. This allows for precise estimates of positions and velocities to be generated from (error-prone) input data.

[0004] Kalman filters are particularly useful for the iterative estimation of system states based on observations that are typically subject to error. In this context, Kalman filters have proven especially advantageous for applications where sensor information from various sensors needs to be combined (or fused) with model information. Furthermore, Kalman filters are frequently used in embedded systems because their calculations are efficient, accurate, and robust. Microcontrollers can also perform Kalman filter calculations with advantageous efficiency.

[0005] Kalman filters are primarily used to fuse data from various sensors that determine vehicle positions, thereby generating highly precise localization data. Localization data here refers specifically to position data, as well as data concerning speed and acceleration. Sensors whose data can be processed or fused using Kalman filters include, for example, GNSS sensors for determining locations via GNSS satellites, inertial sensors, and wheel sensors and steering angle sensors used in motor vehicles to monitor the vehicle's movement via the chassis. These sensors thus generate a kind of precursor data for localization data, which is used as input data in a Kalman filter to determine highly precise localization data.By jointly considering this described data in a Kalman filter using the system models, parameters and / or system states stored in the Kalman filter, it becomes possible to generate highly precise localization data from the described input data.

[0006] Kalman filters are used in particular to obtain highly precise localization data for highly automated driving functions of motor vehicles and especially for autonomous driving functions.

[0007] In a Kalman filter, a continuous correction of internal parameters and system states is typically performed. The term "corrected parameters" is used here to refer to all corrections taking place within the Kalman filter. The quality of the internal parameters and / or system states in the Kalman filter is crucial for improving localization data through the use of the Kalman filter.

[0008] However, there is a fundamental need for improvement regarding the precision and confidence of the localization data obtained with Kalman filters, because for applications of highly automated and autonomous driving, very high precision and high confidence of the localization data used is generally desirable. Disclosure of the invention

[0009] Based on this, a particularly advantageous method for checking the design of the system model and the parameters of the system model in a Kalman filter will be described.

[0010] The invention relates to a method for testing a design of a system model in a Kalman filter, which is part of a filter network for determining localization data in a motor vehicle and uses a first set of sensor data for determining localization data, wherein the method comprises at least the following steps: a) Detecting a standstill situation of the motor vehicle during operation with a standstill detection system, so that a standstill signal is generated; b) Performing a calibration function for parameters of the system model during the standstill situation in the Kalman filter, whereby corrected parameters are determined; c) Performing a check of the corrected parameters by comparing the corrected parameters with reference parameters when a standstill signal is present, wherein the reference parameters are determined from another data source, the other data source generating reference parameters using a second set of sensor data, the second set of sensor data being reduced compared to the first set of sensor data. d) Execute an error function if a check performed in step c) fails.

[0011] The fundamental approach of the procedure described here is to use stationary phases of a vehicle to test and validate the self-correcting capabilities of the Kalman filter. The self-correcting capabilities of the Kalman filter relate in particular to its ability to correct its own internal parameters.

[0012] Stationary phases are typically used by navigation systems for the isolated calibration of individual sensors (e.g., for calibrating inertial sensors). This approach is based on the premise that, for example, an inertial sensor should not output any acceleration readings during a stationary phase. Faulty signals from inertial sensors can thus be detected during stationary phases, sensor zero-point drift can be identified, and offset parameters can be introduced, if necessary, to correct such zero-point drift.

[0013] The procedure described here goes a step further. According to step c), the filter's own system models are checked or validated in the Kalman filter as a whole when a standstill phase occurs.

[0014] Up to now, the system model of a Kalman filter has been validated primarily during development using reference systems (carried in the vehicle). A review of the variances estimated by the filter (and thus of the system model) is therefore only performed during development and not for navigation filters in the field. The method presented here allows validation to be performed for every navigation device in the field at runtime. This enables improved detection of both age-related changes in the measuring instruments and sensor errors.

[0015] Due to its inherent properties, a Kalman filter is typically calibrated continuously. The calibration function described in step b) is preferably a (common) self-calibration of the Kalman filter, which is performed continuously during operation. The calibration according to step b) is a process performed by the Kalman filter itself, both when it is idle and preferably also during other operating phases (non-idle phases). This calibration is, in particular, an automatic error compensation resulting from the structure of the Kalman filter itself. In other words, errors that occur are implicitly detected, lead to changes in the system model, are stored in the system model, and are used in the future to correct sensor signals and generate improved localization data.

[0016] Due to the structure of the Kalman filter, standstill phases play a crucial role because the relationship between the individual signal sources processed by the Kalman filter differs from that during phases of motion. For example, inertial sensors typically provide no data during standstill phases, or only data due to sensor inaccuracies. Consequently, the system states of the system model within the Kalman filter change, particularly during standstill phases, leading to corrections of the system model's parameters. Parameters of the system model that change due to the internal correction mechanisms of the Kalman filter are referred to collectively as corrected parameters.

[0017] The parameters of the Kalman filter (and thus also the corrected parameters) are usually the individual values ​​of the following matrices and vectors: - a covariance matrix of the Kalman filter; - an estimated state vector (x) of the Kalman filter ; - Parameters and / or a structure of a system model of the Kalman filter; and - a noise vector that describes a system noise of the Kalman filter.

[0018] Sets of sensor data, as defined in the first and second sets, are each data from sensors. These sets of data are preferably processed further using filters, such as the Kalman filter. The signals that flow into the first set preferably overlap with the signals that flow into the second set. Most preferably, the second set is a subset of the first set.

[0019] The test performed in step c) does not involve any intervention in the Kalman filter, unlike the calibration described in step b). The test does not change the internal parameters (system states) of the Kalman filter. Rather, it compares the internal (corrected) parameters of the Kalman filter with other parameters. The test is preferably binary and can yield either a positive or negative result. Optionally, the test can also output more than two (binary) results. Preferably, however, there are negative results, which trigger the error function in step d), and positive results, which do not trigger the error function in step d). If the test is positive, the filter network is functioning as desired. The error function in step d) does not need to be triggered. If the test is negative, the filter network is not functioning as desired.In particular, the self-correction of the Kalman filter, which constantly corrects parameters or outputs corrected parameters, is then impaired.

[0020] The procedure aims to trigger the error function described in step d) if, according to step c), it is detected that the Kalman filter is not sufficiently self-correcting. The error function can, for example, detect that the localization data obtained with the Kalman filter is compromised overall and should be discarded or disregarded for highly automated or autonomous driving functions.

[0021] For the comparison according to step c), comparison parameters are preferably determined. This is done using a different data source than the Kalman filter. In the simplest case, comparison parameters can be parameters that were directly determined from sensors (e.g., inertial sensors). Inertial sensors should not normally register any movements and / or accelerations or speeds of the vehicle during a (reliably detected) standstill phase. For this reason, deviations can be detected when data from inertial sensors are used directly as comparison parameters. Such deviations can provide clues to errors in the system model of the Kalman filter.

[0022] Preferably, the other data source is set up in such a way that the comparison parameters obtainable from this data source ensure easy comparability with the corrected parameters.

[0023] The basis of the described procedure is that in step a) a genuine and reliable standstill detection takes place.

[0024] It has been found that wheel sensor signals alone are usually insufficient to reliably detect a standstill for the described procedure. According to the described procedure, the corrected parameters of the Kalman filter are only checked when the vehicle is stationary.

[0025] The method is particularly advantageous if the first set of sensor data used by the Kalman filter includes at least GNSS signals from at least one GNSS sensor and inertial sensor signals from inertial sensors.

[0026] Particularly preferably, the first set of sensor data also includes WSS signals from wheel sensors and, particularly preferably, data from sensors that can be used to monitor the steering angle of a motor vehicle.

[0027] Furthermore, the method is advantageous if age-related changes in GNSS sensors and / or inertial sensors are detected by checking the corrected parameters in step c).

[0028] In particular, the described method can detect age-related changes in sensors. Specifically, it can also detect age-related changes in sensors that were not yet taken into account in the design of the Kalman filter.

[0029] Furthermore, the method is advantageous if data from GNSS sensors are not taken into account for the detection of a standstill situation in step a).

[0030] Furthermore, the method is advantageous if, in step a), sensor data from at least one inertial sensor and / or sensor data from wheel rotation sensors are used to detect a standstill situation.

[0031] To detect a standstill, a multiple sensor data points are preferably fused. For example, the following data can be considered for standstill detection: - Camera images, where, for example, a camera image that is completely or partially still can be interpreted as an indication of a standstill; - Time-based data, where, for example, certain times can be evaluated as an indication of a standstill; - Detecting a standstill in a workshop situation when the vehicle is in a workshop; and / or - Data relating to the condition of the motor vehicle, such as internal system flags indicating a deactivated ignition.

[0032] This list is not exhaustive. Many methods exist for detecting a vehicle's standstill. For the described procedure, it is crucial that the various data points are combined and linked in such a way as to reliably detect a standstill.

[0033] Furthermore, the procedure is advantageous if, when checking the corrected parameters checked in step c), at least the following parameters are checked: - a covariance matrix of the Kalman filter; - an estimated state vector of the Kalman filter; - Parameters and / or a structure of a system model of the Kalman filter; and - a noise vector that describes a system noise of the Kalman filter.

[0034] Furthermore, the method is advantageous if the other data source is a strapdown filter provided in the filter network in addition to the Kalman filter, wherein the strapdown filter uses data from at least one inertial sensor to generate comparison parameters for checking the corrected parameters in step c).

[0035] The strapdown filter is an additional filter to the Kalman filter in the filter network, which preferably processes the second set of sensor data in parallel with the Kalman filter.

[0036] Preferably, the strapdown filter also receives data from the Kalman filter as input data. Most preferably, the strapdown filter receives as input data the output data of the Kalman filter, which was previously determined (at an earlier time) using the Kalman filter.

[0037] The core of the invention is a validation approach performed at standstill that uses a parallel strapdown filter to validate the filter variances. The advantage lies in the ability to detect and respond to sensor errors or deviations from the filter model (e.g., due to sensor aging).

[0038] The strapdown filter preferably processes only data from inertial sensors. Preferably, the strapdown filter has a system model whose structure corresponds to that of the Kalman filter. This system model is regularly adapted to the Kalman filter's system model and, starting from a forward time, is updated or propagated using only the data from the second set (preferably exclusively data from inertial sensors). "Propagation" here means that the system model is updated only using inertial sensor data from the forward time. Specifically, "propagation" means that, at standstill, the system uses only the inertial sensor data to check how the state vector changes when updated by the strapdown filter. Changes should only occur within the expected range.In particular, a correlation between the increase of variances and their respective parameters is to be expected. In principle, the variances should always encompass the corresponding parameters. Preferably, the verification of the corrected parameters includes a comparison of parameters and their variances. A drift of errors or parameters away from the variances triggers the error function described in step d). The system models in the strapdown filter and the Kalman filter each model the drift of the error when no new input data is received. This drift can be measured during a standstill phase because it occurs here without any additional positional changes affecting the parameters.The described procedure, according to step c), essentially determines whether the actual propagation of the fault during standstill (which inevitably always occurs due to inaccuracies in the inertial sensors and their data processing) fits the system model or is correctly implemented in the system model. This allows it to be determined whether the system model is suitable for correctly accounting for the behavior of the inertial sensors during operation.

[0039] In preferred embodiments of the method, the strapdown filter operates permanently and independently of the standstill situation. However, the test described in step c) is then only performed in the standstill situation, which is preferably detected independently of the strapdown filter.

[0040] Strapdown is a fixed term for an algorithm in connection with fixed acceleration and rotation rate sensors.

[0041] Furthermore, it is advantageous if, in parallel with propagating the state vector of the Kalman filter with the strapdown filter, at least one covariance matrix of the Kalman filter is propagated with the strapdown filter as the propagated covariance matrix.

[0042] The strapdown filter preferentially corrects the inertial sensor signals (if available) using the estimated sensor errors from the state vector x of the Kalman filter. The corrected inertial sensor values ​​are then used in the strapdown filter to propagate the initial position, velocity, and orientation over time. Simultaneously, the covariance matrix is ​​propagated using the system model, the inertial sensor data, and the system noise.

[0043] The propagated covariance matrix is ​​then compared with the propagated states of position, velocity, and orientation.

[0044] Initialization errors, as well as measurement errors (inaccuracies) of the inertial sensors, cause the propagation errors of the states to increase. At standstill, the propagation error can be calculated by assuming constant position and orientation, as well as zero velocity, and by subtracting the initial values ​​from the current propagation state. With an ideally designed Kalman filter, three times the standard deviation (from the filter's covariance matrix) should encompass the estimation error. This cannot be verified with conventional systems without a reference system.

[0045] A comparison of the propagation errors described above and the propagated variance (also described above) allows for such a check. If this condition is violated over a longer period (tuning factor), it can be assumed that the system model no longer describes the sensors with sufficient accuracy. Depending on the magnitude of the deviation, this can be addressed by adjusting the parameters of the system model at runtime or by issuing an error signal in the case of larger deviations.

[0046] Furthermore, it is advantageous if, for the verification of the corrected parameters according to step c), a comparison is made between the covariance matrix and the state vector determined with the Kalman filter and the covariance matrix and the state vector propagated with the strapdown filter.

[0047] The comparison of covariance matrices and state vectors does not need to be complete; that is, not every element of the matrices and vectors needs to be compared. Individual values ​​can also be compared, for example, only the diagonal values ​​of the matrices. The diagonal values ​​of the covariance matrix typically contain variances. By comparing the variances, the test according to step c) can be performed.

[0048] This section will also describe a control unit, comprising a processor, which is adapted / configured to execute the described procedure.

[0049] Preferably, the described sensor data sources (GNSS sensors, inertial sensors, wheel sensors, etc.) are connected to the control unit.

[0050] Preferably, the control unit is configured to generate highly precise localization data from sensor data and to regularly execute the described procedure to ensure particularly high quality of the localization data and to execute the fault function described in step d) if the quality of the localization data cannot be ensured.

[0051] Furthermore, a computer program product, comprising instructions that, when executed by a computer, cause the computer to execute the described procedure, is to be described.

[0052] The computer program product is preferably installed on the described control unit.

[0053] The following will be described in more detail: a computer-readable storage medium comprising instructions which, when executed by a computer, cause it to execute the described procedure or the steps of the described procedure.

[0054] The invention and the technical context of the invention are explained in more detail below with reference to the figures. The figures show preferred embodiments to which the invention is not limited. It should be noted in particular that the figures, and in particular the proportions depicted in the figures, are only schematic. They show: Fig. 1: a filter network for carrying out the described procedure; Fig. 2: a flowchart of the described procedure; Fig. 3: an example of a discrepancy occurring between a state vector of a Kalman filter and a state vector propagated by a strapdown filter; and Fig. 4: another example of a discrepancy occurring between a state vector of a Kalman filter and a state vector propagated with a strapdown filter.

[0055] Fig. Figure 1 shows a filter network 2 with a Kalman filter 1 that processes a first set 3 of input data to determine localization data. The first set 3 of input data includes at least data from GNSS sensors 8, data from inertial sensors 9, and optionally additional data from wheel rotation sensors 11.

[0056] The Kalman filter 1 has an internal system model with a multitude of parameters, typically in the form of a state vector X, a covariance matrix 14, a noise vector 15, and possibly other parameters. The system model of the Kalman filter 1 serves to generate precise localization data from the input data. The Kalman filter continuously executes a calibration function that corrects the parameters of the system model, ensuring that precise localization data can still be generated even in the presence of systematic deviations in the input data. The Kalman filter 1 preferentially produces continuously corrected parameters 5 of the system model, which are primarily contained in the state vector X, the covariance matrix 14, and the noise vector 15.

[0057] In addition to the Kalman filter 1, the filter network 2 includes a strapdown filter 10, which is operated in parallel to the Kalman filter 1 and which processes a second set 4 of input data. The second set 4 of input data comprises data from inertial sensors 9 and, in particular, no data from GNSS sensors 8.

[0058] The strapdown filter 10 propagates the parameters of the Kalman filter 1 based on the second set 4 of input data. This means that the strapdown filter 10 propagates the parameters of the Kalman filter 1, in particular independently of the data from the GNSS sensors 8. Thus, comparison parameters 6 propagated by the strapdown filter 10 are generated, namely, in particular, a propagated state vector X, a propagated covariance matrix 16, and a propagated noise vector 17. Preferably, the propagation of the comparison parameters 6 by the strapdown filter 10 occurs continuously and in parallel with the operation of the Kalman filter 1.

[0059] The filter network 2 further features a standstill detection 12, which is based on (here in the Fig. The Kalman filter 1 detects standstill situations (within the scope of data not further defined) and outputs a standstill signal 13. When a standstill situation exists and the standstill signal 13 is output, a test function 20 is activated, which compares the corrected parameters 5 with the comparison parameters 6. In a standstill situation, only expected deviations should occur between the corrected parameters 5 and the comparison parameters 6, and in particular, no deviations exceeding limit values. In the case of unforeseen deviations, the test function 20 would return a negative result, and an error function 7 would be executed. If necessary, errors 19 estimated by the Kalman filter 1 can be transmitted to the strapdown filter 10, and corrections for inertial sensor signals 18 can be generated using the strapdown filter 10, which are then available for further processing.

[0060] Fig. Figure 2 shows a flow diagram of the described method. Step a) is shown, with which a standstill situation is detected. If a standstill situation has been detected, the result of step b) (corrected parameters of the Kalman filter 1) is then subjected to a test according to step c). Step c) is preferably divided into substeps c1), c2), and c3). Both step b) and step c1) are also executed permanently, independently of step a), if necessary, whereby the results of these steps (corrected parameters 5 of the Kalman filter 1 as the result of step b) and corresponding comparison parameters 6 of the strapdown filter 10 as the result of step c1) for steps c2) and c3) are preferably only used if step a) has shown that a standstill situation exists.

[0061] In step c2), a comparison of the corrected parameters 5 with the comparison parameters 6 is preferably carried out. Preferably, both the corrected parameters 5 and the comparison parameters 6 each comprise a plurality of individual parameters, wherein the comparison comprises in particular the comparison of corresponding individual parameters from the corrected parameters 5 and from the comparison parameters 6.

[0062] In step c3), variances of parameters are preferentially compared. If steps c2) and c3) show that a comparison was unsuccessful, an error function 7 is preferentially executed in step d).

[0063] The Fig. 3 and Fig. 4 each show how a deviation between corrected parameters 5 and comparison parameters 6 can develop during a standstill situation and which deviation can lead to the detection of an error or to a negative test result of the test of the corrected parameters 5 based on the comparison parameters 6.

[0064] On the parameter axis 22, a variance 23 and the parameter 24, which is monitored for an error, are plotted over the time axis 21. For simplified understanding: The parameter 24 is, for example, a position estimate. The variance 23 describes the uncertainty of this position estimate 24. It can be seen that the parameter 24 changes according to Fig. 3 behaves in accordance with the expected variance 23 and remains below the variance 23. This shows that the parameter 24 can be estimated based on the variance 23. Correct or permissible behavior can be determined. The test according to step c) of the described procedure would result in a positive test. According to Fig. 4, a deviation 25 occurs, which indicates a problem or a systematic error in the filter model, or which may mean a negative test result of the test of the corrected parameters 5 based on the comparison parameters 6. Here, the parameter 24 increases more rapidly than the variance 23. The test according to step c) of the described procedure would result in a negative test result.

Claims

[1] Method for testing a correction of a system model in a Kalman filter (1) which is part of a filter network (2) for determining localization data in a motor vehicle and uses a first set (3) of sensor data to determine localization data, the method comprising at least the following steps: a) detecting a standstill situation of the motor vehicle during operation with a standstill detection device (12), so that a standstill signal (13) is generated; b) performing a calibration function for parameters of the system model during the standstill situation in the Kalman filter (1), whereby corrected parameters (5) are determined; c) carrying out a test of the corrected parameters (5) by comparing the corrected parameters (5) with comparison parameters (6) when a standstill signal (13) is present, wherein the comparison parameters (6) are determined using another data source (10), wherein the other data source (10) creates comparison parameters (6) using a second set (4) of sensor data, wherein the second set (4) of sensor data is reduced compared to the first set (3) of sensor data, d) Executing an error function (7) if a test carried out in step c) is negative. [2] The method of claim 1, wherein the first set (3) of sensor data used by the Kalman filter (1) comprises at least GNSS signals from at least one GNSS sensor (8) and inertial sensor signals from inertial sensors (9). [3] Method according to one of the preceding claims, wherein age-related changes of GNSS sensors (8) and / or inertial sensors (9) are detected by checking the corrected parameters (5) in step c). [4] Method according to one of the preceding claims, wherein data from GNSS sensors (8) are not taken into account for detecting a standstill situation in step a). [5] Method according to one of the preceding claims, wherein sensor data from at least one inertial sensor (9) and / or sensor data from wheel rotation sensors (11) are used to detect a standstill situation in step a). [6] Method according to one of the preceding claims, wherein, when checking the corrected parameters (5) checked in step c), at least the following parameters are checked: - a covariance matrix (14) of the Kalman filter (1); - an estimated state vector (x) of the Kalman filter (1); - parameters of a system model of the Kalman filter (1); and - a noise vector (15) describing a system noise of the Kalman filter (1). [7] Method according to claim 6, wherein the other data source (10) is a strapdown filter (10) provided in the filter network (2) in addition to the Kalman filter (1), the strapdown filter (10) using data from at least one inertial sensor (9) to generate comparison parameters (6) for testing the corrected parameters (5) in step c). [8] Method according to claim 7, wherein the comparison parameters (6) generated with the strapdown filter (10) comprise at least one propagated state vector (y) which corresponds to the state vector (x) estimated with the Kalman filter (1) and propagates this state vector (x). [9] Method according to one of claims 6 to 8, wherein, in parallel with the propagation of the state vector (x) of the Kalman filter (1) with the strapdown filter (10), the at least one covariance matrix (14) of the Kalman filter (1) is propagated with the strapdown filter (10) as a propagated covariance matrix (14). [10] Method according to one of claims 6 to 9, wherein for the testing of the corrected parameters (5) according to step c), a comparison of the covariance matrix (14) determined with the Kalman filter (1) and the state vector (x) determined with the Kalman filter (1) with the covariance matrix (14) propagated with the strapdown filter (10) and the state vector (y) propagated with the strapdown filter (10) is carried out. [11] Control device comprising a processor adapted / configured to carry out the method according to any one of claims 1 to 10. [12] A computer program product comprising instructions which, when the computer program product is executed by a computer, cause the computer to carry out the method according to any one of claims 1 to 10. [13] A computer-readable storage medium comprising instructions which, when executed by a computer, cause the computer to carry out the steps of the method according to any one of claims 1 to 10.

Images