Method for operating a sensor system, device for processing sensor data and sensor system

Abstract

The invention relates to a method for operating and calibrating an inertial sensor system. In particular, a concept is presented that enables the implementation of parameters for a correction model to compensate for sensor errors during operation. The parameters can be determined, in particular, by evaluating the course of a movement direction.

Description

Technical FieldThe present invention relates to a method for operating a sensor system, in particular an inertial sensor system. The present invention further relates to a device for processing sensor data, in particular sensor data of an inertial sensor system, and to such a sensor system.BackgroundNumerous modern applications, such as navigation systems, require a very precise determination of position data and trajectories. This position data can be determined, for example, by global navigation satellite systems (GNSS), such as GPS, Galileo, etc. Furthermore, approaches based on inertial sensor systems are also known, which can determine a movement of an object by means of acceleration sensors and rotation rate sensors and can derive position changes and thus movement paths from these sensor data.Furthermore, systems are increasingly being used which can determine the position and direction of movement as exclusively as possible on the basis of inertial sensor data.The documents DE 10 2022 205 457 A1 and each describe sensor system and computing units and methods for evaluating sensor data, wherein a drift of the sensor data is determined and removed from the measurement data.Disclosure of the InventionThe present invention provides a method for operating a sensor system, a method for processing sensor data and a sensor system having the features of the independent patent claims. Further advantageous embodiments are the subject matter of the dependent claims.Accordingly, the following is provided:A method for operating a sensor system, in particular an inertial sensor system, having the following steps. In a first step, an at least approximately constant speed of the inertial sensor system is detected. In this case, a value for this at least approximately constant speed can be estimated. The method further comprises a step for determining a current speed. The speed can be determined in particular using sensor data of the inertial sensor system. Furthermore, the method comprises a step for determining a speed profile of the inertial sensor system, the speed profile can be determined in particular using the determined current speed. In this case, both the magnitude and the direction can be evaluated for the consideration of the speed profile. The method further comprises a step of calculating an average speed for a predetermined time interval. The mean speed of the vehicle may be calculated using the determined speed profile. Finally, the method comprises a step for adapting parameters for a correction model for compensating sensor errors of the inertial sensor system. The parameters can be determined in particular using a minimization of the difference between the calculated mean speed and the estimated value of the at least approximately constant speed. The determination of the parameters for the correction model can be effected if an at least approximately constant speed has been detected.Furthermore, the following is provided:A device for processing sensor data, in particular sensor data from an inertial sensor system. For this purpose, the apparatus comprises a processing device. The processing device is designed to receive and process sensor data from sensors of the inertial sensor system. In particular, the processing device is designed to carry out the method according to the invention.Finally, the following is provided:A sensor system having an acceleration sensor, a rotation rate sensor and a device according to the invention for processing sensor data. The device is designed to receive and process sensor data from the acceleration sensor and the rotation rate sensor.Advantages of the InventionNumerous applications, such as navigation systems, fitness applications, etc., require a determination of movement data, for example a spatial position, a movement direction, a movement speed, an orientation in space or the determination of a movement path. In addition to systems based on satellite navigation systems (GNSS), there are also approaches with inertial sensors, such as acceleration sensors, rotation rate sensors, etc., which can derive the desired information using the sensor data.For example, information such as speed or the like can be derived from sensor data from acceleration sensors. Furthermore, for example, information about an orientation or orientation can be obtained from sensor data of rotation rate sensors. The desired information can then be derived from this data.However, it can be observed that, for example, inaccuracies or sensor errors, such as an offset (bias) or other errors (for example axis offset or sensitivity error), have an influence on the accuracy of the ascertained information. In particular, due to integration operations performed in determining the desired situations from the sensor data, possible sensor errors may accumulate. In this case, external influences, such as temperature fluctuations or the like, can also have an influence on the sensor errors, for example.For example, conventional approaches may use additional information from satellite navigation systems or the like to improve accuracy. In addition, for example, even when detecting a completely stationary sensor system, the sensor system can be adjusted to exactly this known state of standstill.It is now an idea of the present invention to provide a concept by which it is made possible to increase the efficiency and accuracy of a sensor system based on inertial sensors, in particular acceleration sensors and rotation rate sensors. The concept according to the invention enables, on the one hand, efficient determination of parameters for compensating sensor errors. In addition, the concept enables the determination of parameters for the compensation of sensor errors during operational operation, for example during the determination of movement data or movement paths, to be carried out.This allows the accuracy of the ascertained information relating to a detected movement or position to be increased. In particular, the possibility of a dynamic adaptation of the parameters for the compensation of sensor errors also allows a good accuracy of the ascertained information to be achieved over a longer period of time. In addition, the concept presented also enables particularly efficient and thus resource-saving processing of the sensor data for position determination and for determination of the parameters for the error compensation.According to the invention, the compensation or correction of the sensor values can be carried out, for example, on the basis of an analysis of the speed. The basic principle of such an approach can be based, for example, on detecting a state of the sensor system in which the sensor system is moving at least approximately at constant speed and then comparing such a detected constant speed with an average speed during this phase of constant speed. From this comparison, parameters can be derived to adapt a correction model for compensating sensor errors. For example, the data can be evaluated by one or more sensors, in particular sensor data of the sensor system, in order to detect a predetermined state, such as an at least approximately constant speed of the sensor system. A course of the speed over a predefined period of time, at which the speed is within a predefined value range / tolerance band, can be considered as a constant speed, for example. For such an analysis of the sensor values for determining an at least approximately constant course of the speed, arbitrary suitable sensor data that are available can be evaluated in principle. In addition, any suitable methods, such as a frequency analysis of the available sensor data or the like, can also be used for evaluating an at least approximately constant speed. Additionally or alternatively, it is also possible to analyze the course of the speed using a trained neural network or in another suitable manner.If such a state is detected for an at least approximately constant speed, a value of this at least approximately constant speed can be determined for this state and this speed value can be used as a basis or reference value for the correction or compensation to be carried out subsequently. Furthermore, in the case of an at least approximately constant speed, a plurality of speed values can be determined over a period of time on the basis of the data from the sensor system, and a speed profile can be determined from this. An average value can be formed for this speed profile, so that fluctuations in the individual values of the speed profile are compensated for. This mean value of the speed curve can then be compared with the previously determined value (reference value) during the detection of the at least approximately constant speed for the sensor system. In particular, on the basis of this comparison, a correction model for compensating sensor errors can be created or adapted, so that a difference between the initially detected at least approximately speed and the mean value of the speed profile is minimized. For the creation or adaptation of the correction model, considerations on the reliability of the individual values and possibly error estimation or the like can also be taken into account.According to one embodiment, the adjustment of the parameters for the correction model takes place only when the speed is greater than a predetermined threshold value. This makes it possible to ensure that the underlying data are based on a stable state of the sensor system.According to one embodiment, the method comprises a step for determining uncertainties for the determined (estimated) constant speed of the sensor system and / or the mean speed, which has been determined on the basis of the sensor values. In this case, the adjustment of the parameters for the correction model can be done using these uncertainties.According to one specific embodiment, the correction model for compensating sensor errors includes a probabilistic filter, in particular a Kalman filter. The Kalman filter can be, for example, a nonlinear Kalman filter, for example an extended Kalman filter or a cube Kalman filter, in particular a filter designed as a square root cube Kalman filter.According to one specific embodiment, the detection of the at least approximately constant speed is carried out using the available sensor data. Additionally or alternatively, data of further sensors can also be used. For example, patterns or the like indicative of a stable state such as a constant speed may be identified in the available data. Moreover, it goes without saying that any other methods for detecting a constant speed and for estimating the value of this speed are also possible.According to one specific embodiment, the detection of the uniform motion profile is carried out using a previously trained neural network. As a result, complex data structures can also be analyzed, classified and evaluated very easily in a relatively simple manner.According to one embodiment, the calculation of the mean speed takes place for a time window of one to three seconds. Such time windows in the range between one and three seconds have proven to be very well suited for averaging. Moreover, depending on the application, any other time windows for averaging are of course also possible.According to one embodiment, the calculation of the average speed is carried out using a predefined weighting of the individual speed values. For example, individual weighting factors can be assigned in each case in the individual sensor values within the noteworthy time window. In this way, the dynamics can be adjusted when calculating the averaged value.The above embodiments and developments can be combined with one another as desired, insofar as appropriate. Further embodiments, developments and implementations of the invention also include combinations of features of the invention described above or below with respect to the exemplary embodiments, which combinations are not explicitly mentioned. In particular, the person skilled in the art will also add individual aspects as improvements or additions to the respective basic forms of the invention.Brief Description of the DrawingsFurther features and advantages of the invention are explained below with reference to the figures. The following are shown: FIG. 1 : shows a schematic illustration of a sensor system according to one embodiment; FIG. 2 : shows a flow chart as it is used as a basis for a method for calibrating a sensor system according to one embodiment; FIG. 3 : shows a flow chart as it is used as a basis for a method for calibrating a sensor system according to a further embodiment; FIG. 4 : shows a flow chart as it is used as the basis for a method for processing sensor data according to one embodiment; and FIG. 5 : shows a flow chart as it is used as the basis for a method for processing sensor data according to a further embodiment.DESCRIPTION OF EMBODIMENTSFIG. 1 shows a schematic illustration of a sensor system 1 according to an embodiment. Sensor system 1 may include multiple inertial sensors, for example, at least one rotation rate sensor 11 and at least one acceleration sensor 12. The inertial sensors 11, 12 can be sensors based on microelectromechanical systems (MEMS), for example. In principle, however, the principle described here can also be applied to any other inertial sensors.The rotation rate sensor 11 can provide sensor data for a rotational movement (yaw rate) about a spatial axis, for example. In particular, the rotation rate sensor 11 can also provide sensor data for rotation rates of, for example, two or three spatial axes orthogonal to one another. The sensor data are generally provided with this sensor in relation to a reference system of the respective sensor and thus of the sensor system 1.Similarly, the acceleration sensor 12 can provide sensor data for a translatory acceleration along a spatial axis, for example. Here, too, the acceleration sensor 12 can provide acceleration values for two or three spatial axes orthogonal to one another, for example. These acceleration values are also generally provided with respect to a reference system of the respective sensor and thus of the sensor system 1.The sensorially detected values of the inertial sensors 11, 12 can be provided, for example, at an input interface 21 of a processing device 20. The sensor values can be provided in the form of analog or digital values. In particular, the sensor values can be provided at a predefined first sampling rate and / or captured by the input interface 21.The received sensor values can optionally be temporarily stored. Preprocessing, for example filtering or the like, is also possible if appropriate. This can take place either already in the input interface 21 or alternatively in the downstream processing device 22.The processing device 22 can process the sensor data from the acceleration sensor or sensors 12 in order to derive information about a movement, direction of movement or speed therefrom. For example, the acceleration values for the individual spatial directions can be integrated in order to calculate values for the respective speed components from this. By integrating the velocity components once again, distance travelled can also be determined.Analogously, the processing device 22 can also process sensor data from the rotation rate sensor or sensors 11 and can obtain information from this, for example, about an orientation / orientation of the sensors 11 and thus of the sensor system 1. For this purpose, too, the sensor values of the rotation rate sensors 11 can be integrated, for example.Due to inaccuracies or sensor errors, in particular offset and other errors, such as e.g. axis offset or sensitivity errors, the received sensor values may be error-prone. The integration of the sensor values described above has the result that, with increasing time of integration, the determined data such as speed, distance and orientation can increasingly deviate from the actual movement of the sensor system.Therefore, processing for detecting and correcting or compensating such inaccuracies can be provided in the processing device 22 of the processing apparatus 20.While conventional concepts for correction or compensation deviations are generally based on the use of additional information, for example from a satellite navigation system (GNSS) or the like, concepts are described below which also enable determination and adaptation of parameters for correction or compensation of system or sensor errors even in the case of sensor systems 1 in motion and in particular also during operation for determining navigation data such as speed, distance, direction of motion, etc.In one possible embodiment, the compensation or correction of the sensor values can be carried out, for example, on the basis of an analysis of the speed. The basic principle of such an approach can be based, for example, on the detection of a state of the sensor system 1, in which the sensor system 1 moves at least approximately at constant speed and then compare an average speed during such a phase with the current development of the speed. From this comparison, parameters can be derived to adapt a correction model for compensating sensor errors. The correction model can be used on the one hand to compensate possible errors in the raw data from the sensors 11, 12. In addition, the correction model can also be used to adapt or correct position or movement data already determined beforehand. Position or movement data which have been previously ascertained on the basis of data which have been ascertained in inertial sensors 11, 12 can thus be modified by means of this correction model in order to compensate for an offset, a drift or other deviation.FIG. 2 shows a flow chart as it can be used as the basis for a method for operating a sensor system 1, in particular an inertial sensor system, according to one specific embodiment.First, sensor data can be received from one or more inertial sensors 11, 12, in particular at least one acceleration sensor 12. Sensor data may be received by at least one acceleration sensor 12 and one rotation rate sensor 11.In step S 11, the data from one or more sensors, in particular sensor data of the sensor system 1, can be evaluated in order to detect a stable movement state of the sensor system 1. In this exemplary embodiment, this stable movement state can be, in particular, a state in which the mean orientation of the sensor system is at least approximately constant with respect to the mean movement direction of the sensor system. For this purpose, for example, the direction of movement and the orientation of the sensor system 1 can be considered and averaged over a predefined period of time. For such an analysis of the sensor values, it is possible in principle to evaluate any suitable sensor data which are available. In addition, any suitable methods, such as a frequency analysis of the available sensor data or the like, can also be used for evaluating an at least approximately constant direction of movement. Additionally or alternatively, it is also possible to analyze the course of the direction of movement using a trained neural network or in another suitable manner.On the basis of the received sensor data, a current movement, in particular the current vectorial speed of the sensor system 1, can be determined in a step S 12. Any suitable approaches, in particular for example an integration of acceleration values, can be used for this purpose. Furthermore, the orientation of sensor system 1 may be ascertained, for example, using sensor values of a rotation rate sensor 11In step S 13, the (vectorial) speed of the sensor system 1 determined can be monitored in order to determine a direction of movement over time from this. From this, an average direction of movement over a corresponding time interval can be determined. Analogously, the orientation of the sensor system 1 can be evaluated in order to determine from this an average orientation of the sensor system 1 in the time interval under consideration. For example, a time interval between 1 second and 3 seconds can be used as the time interval.From the values calculated in this way, a difference between the averaged change in the direction of movement and the averaged change in the orientation can be calculated in step S 14. Such a difference corresponds, for example, to a change in the orientation of the sensor system 1 with respect to the (averaged) direction of movement.If, according to the analysis carried out in step S 11, an at least approximately stable movement state is present, then in step S 15 parameters for a correction model for compensating sensor errors can be determined. For this purpose, for example, an approach can be used which adjusts the parameters for the correction model for compensating the sensor errors in such a way that the difference between the previously calculated mean change in the direction of movement and the value of the mean change in the orientation of the sensor system 1 becomes minimal, i.e. preferably approaches zero.The correction model for compensating the sensor errors can be, for example, a mathematical model based on a probabilistic filter. In particular, the correction model can comprise, for example, a so-called Kalman filter. In principle, however, any other suitable correction models or filter approaches are also possible.Because of system-related properties and disturbing influences, both the present sensor values from the inertial sensors 11 and 12 and the results in the calculation of the speed values may be subject to uncertainties. Therefore, these uncertainties can also be incorporated into the adaptation of the parameters for the correction model. For example, the input data such as acceleration, rotation rate, or the like may be used. The resulting uncertainty or accuracy can be assigned to inaccuracies or tolerances and derived therefrom. Additionally or alternatively, it is also possible to assign resulting variables such as speed, orientation, etc. to uncertainties or accuracy values. Furthermore, it is also possible, for example, to initially assume a large uncertainty due to a missing database at the beginning or during an initialization. As the operation continues, the uncertainty of the values can then decrease according to the obtained data or the underlying accuracy for the values can increase. In principle, any suitable approaches for evaluating the uncertainties or the underlying accuracy are possible for this purpose. This information can also be included in the determination of the parameters for the correction model. For example, the individual variables can be weighted according to the underlying uncertainties or accuracy values.As can be seen from these embodiments, the determination or adaptation of the parameters for the correction model for compensating the sensor values takes place in an operating state or an operating phase in which current measured values, i.e. speed, direction of movement, orientation, etc. can be determined. Thus, for dynamically adjusting the correction model, no interruption of the operational operation is required.By means of such a continuous dynamic adaptation of the correction model, stable, reliable and relatively accurate data about speed, direction of movement, position, etc. can thus also be obtained and provided over a significantly longer period of time. In particular, possible deviations such as an offset, a drift or the like in position or movement data already previously determined can also be (subsequently) corrected using this correction model.Analogously to an adaptation of parameters for a correction model for compensating sensor values on the basis of the monitoring of the direction of movement and the orientation of the sensor system 1, an adaptation of parameters for the correction model is also additionally or alternatively possible by evaluating the speed or a temporal profile of the speed. One possible concept is explained below with reference to FIG. 3.Sensor data is also received from one or more inertial sensors 11, 12. Sensor data can preferably be received both by at least one acceleration sensor 12 and a rotation rate sensor 11.In step S 21, the data from one or more sensors, in particular sensor data of the sensor system 1, can be evaluated in order to detect a predetermined stable movement state, such as an at least approximately constant speed of the sensor system 1. A course of the speed over a predefined period of time, at which the speed is within a predefined value range / tolerance band, can be considered as a constant speed, for example. For such an analysis of the sensor values for determining an at least approximately constant course of the speed, arbitrary suitable sensor data that are available can be evaluated in principle. In addition, any suitable methods, such as a frequency analysis of the available sensor data or the like, can also be used for evaluating an at least approximately constant speed. Additionally or alternatively, it is also possible to analyze the course of the speed using a trained neural network or in another suitable manner.On the basis of the received sensor data, a current speed of the sensor system 1 can be determined in a step S 22. For example, the values from the acceleration sensor 12 can be integrated for this purpose.In a step S 23, an average speed can be calculated for a predefined time window from the determined values for the speed of the sensor system 1. Any suitable time window can be used for this purpose. For example, a time period between 1 second and 3 seconds can be used as the time window.If an at least approximately constant speed is present, then in step S 24 parameters for a correction model for compensating sensor errors can be determined. The adjustment of the parameters for the correction model can be effected using a minimization of the difference between the calculated mean speed according to step S 23 and the speed from step S 21 estimated as the constant speed of the sensor system. In particular, the methods referenced in the introduction can also be implemented.For this purpose, for example, an approach can be used which adjusts the parameters for the correction model for compensating the sensor errors in such a way that the one difference becomes minimal, i.e. preferably approaches zero.The underlying correction model for compensating the sensor errors can also be, for example, a mathematical model based on a probabilistic filter. In particular, the correction model can comprise, for example, a so-called Kalman filter. In principle, however, any other suitable correction models or filter approaches are also possible.Likewise, data for the uncertainties of the sensor values and / or the variables resulting therefrom, such as speed, orientation, etc., can also be included in the determination of the parameters for the correction model.In the two methods described above for determining parameters for the correction model, mean values for speed, rotation rate and / or orientation are used, among other things. In addition to conventional averaging, in which all sensor values are weighted equally strongly within a predefined time interval, arbitrary other approaches are also possible. For example, the individual data points can be assigned an individual weighting in each case according to their temporal position in the time window with the values to be taken into account. In this way, the dynamics of the mean value formed can be adjusted accordingly.In order to determine whether the sensor system 1 is in a stable state of motion, various suitable approaches are possible in principle. For example, the variation or change in the translatory speed and the rotation rate can be compared with predetermined threshold values. In this case, translatory speed or rotation rate can be averaged, for example, over a predetermined period of time and these mean values can be analyzed accordingly. In addition, any other suitable, for example statistical, models are also possible. Likewise, the stable movement state can also be detected, for example, by means of simple machine-learning methods or the like.In particular, for example, a classification of the state of motion is possible. For example, a group of classes of different movement states can comprise, for example, a first class in which the sensor system 1 is completely at a standstill. A further, second class can be assigned, for example, to movement states in which the sensor system 1 is at a standstill averaged over predefined time intervals. In this case, the sensor system 1 can also move within this time interval, for example vibrate, oscillate back and forth, or the like. A third class of movement states can be assigned, for example, to movements in which the sensor system 1 is in a uniform movement. Finally, a fourth class of movement states can be assigned, for example, movements in which the sensor system 1 on average moves uniformly, but can also execute random, random or irregular movements for a short time. This may be the case, for example, when such a sensor system 1 is attached to the wrist of a user, wherein the user moves while swinging the arms, so that the sensor system swings back and forth on the wrist during the movement on the arm. The classification into four classes listed here, however, is to be understood here as merely exemplary. In addition, any other types of classifications for dividing the movement states, in particular for classifying different stable movement states, are also possible.For error compensation or generation of a correction model, in the case of a static state, the correction can be carried out, for example, on the basis of a probability of one or more of the following assumptions (it also being possible in principle, moreover, to make other assumptions): a) the translatory speed is at least approximately zero; b) the rotation rate is approximately zero or corresponds to an offset of the rotation rate sensor; c) an amplitude and / or direction of a value from the acceleration sensor corresponds to an amplitude or direction of the gravitational acceleration transformed into the coordinate system of the sensor.The corrections can be made according to measurements or estimates by means of a probabilistic filter or in any other manner. In particular, the correction can be carried out for all of the aforementioned classes of stable movement directions if the correction is applied here for a time window within which a stable movement state results averaged.In addition, further correction approaches, such as, for example, specifications for an average speed, an average acceleration or the like, can optionally also be used.Such a correction approach also has direct or indirect influence on the determined or corrected position, speed, alignment and sensor errors of acceleration sensor and rotation rate sensor. For example, if a correction is made for a dedicated speed of zero, the position and the linear acceleration are updated. Updating the linear acceleration results in error compensation of the acceleration and the orientation. The error compensation of the acceleration then leads to an error compensation of the acceleration sensor, and the correction of the orientation leads to an error compensation of the rotation rate, which in turn leads to an error compensation of the rotation rate sensor. Thus, all the states are corrected. The degree of the effect can be based here on the cross covariance estimated by the probabilistic filter, which cross covariance results from the noise propagation and the probabilities defined in the probabilistic filter.When processing sensor values for determining speed, direction of movement, path of movement, orientation, position etc., the sensor values are generally recorded, provided and processed at a first, relatively high sampling rate. This sampling rate may be in the range of a few 100 Hz to a few kilohertz. Processing all data at such a high sampling rate may require high computing power. In addition, transmission paths with a suitably high bandwidth must also be provided when forwarding the corresponding data and results.For further optimization and for increasing the efficiency, the sampling rate can be reduced in a first processing step, so that the subsequent steps need only be carried out at a second, lower sampling rate or the data can be forwarded at this second, lower sampling rate. The second sampling rate can be in the range of a few 10 Hz up to a few Hertz, for example.One possible method for reducing the sampling rate is described below in conjunction with FIG. 4. The method is based on the basic principle that the data with the first, higher sampling rate is first received and integrated over several iteration steps. Subsequently, the integrated values can be differentiated according to the time interval of the integration and these differentiated values can be output at the second sampling rate.At the beginning, an initialization can take place, in which all speed values and rotation rate values are reset, for example set to 0.In step S 31, sensor data are initially received for this purpose from the sensors 11, 12 of the inertial sensor system 1 at the first sampling rate.Optionally, in step S 32, the received sensor values can then be corrected, if appropriate, according to a correction model. For example, a suitable scaling can be effected. Additionally or alternatively, an offset can also be adjusted or removed.In step S 33, the received sensor data is integrated, i.e. summed, with the first sampling rate. In this case, for example, the data can be processed in a reference system of the sensor system 1. Since the sensor system 1 can move over time and, in particular, a rotation of the sensor system 1 can also take place, the corresponding reference system of the sensor system 1 can change over time. In order to take this change of the reference system into account, it is possible, for example, to transform the already integrated data into the respectively current reference system of the sensor system 1. Then, the current sensor data may be added to the transformed data for integration. In this way, the integration result is always present in the form of the reference system according to the current orientation of the sensor system 1. Alternatively, it is also possible in principle to carry out the integration on the basis of any fixed reference system. For example, the original reference system can be used as a basis during an initialization of the sensor system 1. The sensor data can then be transformed to this fixed reference system before integration. In particular, for example, a transformation from the reference system of the sensor system 1 to a global reference system (for example, with respect to the gravitational field of the earth) is also possible. For the integration and in particular for the transformation of the reference system, a linear approximation can be used here, for example. In particular, it is possible in this case for trigonometric functions or nonlinear formulae with one or more trigonometric functions to be approximated at least in sections by suitable linear equations.After the sensor data have been integrated over at least two integration steps, for example over a predefined time interval or for a predefined number of integration or iteration steps, the result of this integration can be differentiated in step S 34 according to the integration time interval. In this way, after each differentiation, new data for sensor values corresponding to a second, significantly lower sampling rate can be obtained. A suitable linear approximation can also be used if necessary for this differentiation of the sensor values. For this purpose, different linear approximations can be used, in particular, as a function of the sensor data. For example, a different linear approximation can be used for high rotation rates than for low rotation rates. Moreover, it is also possible to understand that any other distinctions for selecting a suitable linear approximation are possible. By such linear approximation to, simpler computing structures can also be used which are not capable of carrying out complex, partially trigonometric functions. In addition, the linear approximation can reduce the required computing power and computing time.If the sensor values were corrected in accordance with a correction model in optional step S 32, the errors compensated previously can be added again to the sensor values with the second sampling rate, which is now lower, in likewise optional step S 35. In this way, sensor values with the second sampling rate are available for the further processing, which sensor values still have the original error properties. The sensor values provided at the second sampling rate can thus also be used to generate parameters for models for compensating errors. In particular, the sensor data provided at the second sampling rate can also be used to carry out the previously described methods for adapting the parameters for the correction models.The first sampling rate and the second sampling rate can be fixed in principle. In addition, however, it is also possible to dynamically adapt the first sampling rate and / or the second sampling rate during operation. For example, the first sampling rate can be adapted as a function of a data rate by the sensor system 1. In addition, any other desired approaches for varying the first sampling rate are also possible. The second sampling rate can be adjusted or predefined, for example, by a downstream processing device. This makes it possible for the data to be provided at a second sampling rate which corresponds to the requirements for further processing. For example, the second sampling rate can be adapted as a function of a data rate for transmission of the output data. It is likewise possible to adapt the second sampling rate as a function of the processing power of a downstream processing device. In addition, any other desired criteria for dynamically adapting the second sampling rate are also possible for the second sampling rate.The components and method elements described above can be used, for example, in a sensor system 1 having inertial sensors 11, 12 in order to increase the accuracy, long-term stability and processing speed. This allows sensor systems to be realized which can provide position information, movements, movement profiles, etc. in an efficient and precise manner.As has also already been explained above, the parameters for compensating sensor errors can be determined and adapted during operational operation, i.e. in parallel with the determination of the movement data. As a result, such sensor systems 1 are not limited to static, predefined correction models. By integrating the compensation of sensor errors for the processing of the sensor data, it is also possible to dynamically adapt the parameters for the correction models during operation. A possible sequence for such a method for processing sensor data of a sensor system 1 having a plurality of inertial sensors 11, 12 is illustrated, for example, in FIG. 5.In step S 41, sensor data can be received from sensors 11, 12, in particular one or more rotation rate sensors 11 and one or more acceleration sensors 12.In step S 42, sensor errors of the received sensor data are corrected by a dynamically adaptable correction model. The parameters for this correction model can be dynamically adjusted using currently determined values for an orientation, speed and / or position. For this purpose, in particular the above-described methods for determining or adapting parameters for correction models can be used.In step S 43, orientation, speed and / or position and optionally any other suitable parameter can be determined. Furthermore, position or movement data already determined beforehand can optionally also be modified or corrected on the basis of the adapted correction model.The determination of the parameters for the correction model and the determination of the speed can also be carried out here in particular with a reduced second sampling rate, wherein this second sampling rate, as already described above, is lower than the first sampling rate at which the sensor data are provided by the sensors 11, 12.In summary, the present invention relates to a method for operating and calibrating an inertial sensor system. In particular, a concept is presented which makes it possible to execute parameters for a correction model for compensating sensor errors during operational operation. The parameters can be determined in particular by evaluating a course of the speed.Additionally or alternatively, the parameters can be determined by evaluating a profile of a direction of movement.The invention furthermore relates to a method for reducing the data rate of sensor data in an inertial sensor system. For this purpose, the original sensor data can be preprocessed and integrated with a first sampling rate. The integrated values are then differentiated according to a second, lower sampling rate.References included in the specificationThis list of documents cited by the applicant has been produced in an automated manner and is only included for the better information of the reader. The list is not part of the German patent application or utility model application. The DPMA does not take any adhesion for any faults or omissions.Patent Literature citedDE 10 2022 205 457 A1

[0004]

Claims

Method for operating an inertial sensor system (1), having the steps: detecting (S 21) an at least approximately constant speed of the inertial sensor system (1); ascertaining (S 22) a speed profile of the inertial sensor system (1) using sensor data of the inertial sensor system (1); calculating (S 23) an average speed for a predetermined time interval using the ascertained speed profile of the inertial sensor system (1); and adapting (S 24) parameters for a correction model for compensating sensor errors of the inertial sensor system (1) using a minimization of the difference between the calculated average speed and the value of the at least approximately constant speed if an at least approximately constant speed has been detected.Method according to claim 1, wherein the adjustment (S 24) of the parameters for the correction model takes place only if the speed is greater than a predetermined threshold value.Method according to Claim 1 or 2, wherein the method comprises a step for determining uncertainties for the determined constant speed of the inertial sensor system (1) and / or the mean speed, and wherein the adaptation (S 25) of the parameters for the correction model takes place using the determined uncertainties.Method according to one of Claims 1 to 3, wherein the correction model for compensating sensor errors comprises a probabilistic filter, in particular a Kalman filter.Method according to one of Claims 1 to 4, wherein the detection (S 21) of the at least approximately constant speed is carried out using the sensor data and / or data of at least one further sensor.Method according to one of Claims 1 to 5, wherein the detection of the at least approximately constant speed is carried out using a previously trained neural network.Method according to one of Claims 1 to 6, wherein the calculation of the mean speed takes place for a time window of one to three seconds.Method according to one of Claims 1 to 7, wherein the calculation of the mean speed is carried out using a predefined weighting of the individual speed values.Device (20) for processing sensor data from an inertial sensor system (1), having a processing device (22) which is designed to receive and process sensor data from sensors (11, 12) of the inertial sensor system (1), wherein the processing device (22) is designed to carry out a method according to one of Steps 1 to 8.Sensor system (1), having: an acceleration sensor (11), a rotation rate sensor (12), and a device (20) for processing sensor data according to Claim 9, wherein the device (20) is designed to receive and process sensor data from the acceleration sensor (11) and the rotation rate sensor (12).

Images