WO2018091160A1

WO2018091160A1 - Functional monitoring of a sensor device of a motor vehicle

Info

Publication number: WO2018091160A1
Application number: PCT/EP2017/072846
Authority: WO
Inventors: Mohamed ABBAZ; Axel Klekamp
Original assignee: Valeo Schalter Und Sensoren Gmbh
Priority date: 2016-11-18
Filing date: 2017-09-12
Publication date: 2018-05-24
Also published as: DE102016122193A1

Abstract

The invention relates to a method for monitoring a function of a sensor device (2) of a motor vehicle (1), comprising (a) acquiring (10a-10g) object information (O_k(t_n;t_n+1)) about at least one object (k) external to the vehicle at a first time (t_n) and at a second time (t_n+1) following the first time (t_n), by at least one sensor unit (3, 4) of the sensor device (2), (b) storing (11) internal object information (O_k(t_n)), which corresponds to the object information (O_k(t_n)) acquired at the first time (t_n) in a computing unit (5) of the sensor device (2), (c) predicting (12) object information (O'_k(t_n+1)) for the second time (t_n+1) on the basis of the internal object information (O_k(t_n)) by means of the computer unit (5), (d) updating (13) the predicted object information ((O'_k(t_n+1)) relating to the new internal object information (O_k(t_n+1)) as a function of the object information (O_k(t_n+1)) acquired at the second time (t_n+1) by the computing unit (5), (e) comparing (17) the object information (O'_k(t_n+1)) predicted for the second time (t_n+1) with the updated object information (O_k(t_n+1)) and calculating a degree of innovation (l_k[O'_k(t_n+1); O_k(t_n+1)]) for the at least one object (k) as a function of a result of the comparison (17) by the computing unit (5), and (f) checking (18) a plausibility of the measure of innovation (l_k[O'_k(t_n+1); O_k(t_n+1)]) by means of the computing device (5), and (g) outputting (19) of a fault message by the computer device (5) if the checking (18) supplies a negative result, in order to make available a method of monitoring, in a way which is as simple and economical in terms of resources as possible, a function of the sensor device (2), in particular a sensor device with a plurality of sensor units (3, 4) whose data are merged, and in this way to meet the requirements made in respect of the functional reliability of the sensor device (2).

Description

Function monitoring of a sensor device of a motor vehicle

The invention relates to a method for monitoring a function of a

Sensor device of a motor vehicle, comprising (a) detecting object information about at least one external object at a first time and at a second time following the first time by at least one sensor unit of the sensor device, with a (b) storing an internal object information, which in a computing unit of the sensor device corresponds to (c) predicting object information for the second time based on the internal object information by the computing unit, with (d) updating the predicted object information to the new internal object information depending on recorded at the second time

Object information by the arithmetic unit, and with (e) comparing the object information predicted for the second time with the updated one

Object information and calculating an innovation measure for the at least one object as a function of a result of the comparison by the arithmetic unit. The invention also relates to a corresponding monitoring device for monitoring a function of a sensor device of a motor vehicle.

In motor vehicles, it is important to ensure the functional safety of various components. This also affects sensor devices with one or more sensors. To the functionality of one or more sensors

or to monitor a sensor device, special monitoring algorithms are usually applied to sensor data detected by the sensors or special error or monitoring signals are processed, which are generated or received by the respective sensor device or the respective sensors. These procedures require special algorithms and thus additional computational and storage capacities on the part of a corresponding

Arithmetic unit, for example, an electronic control unit (ECU) of

Sensor device or a corresponding monitoring device. Especially in the field of multisensory data fusion, in which data from a large number of sensors have to be combined and checked, this concept has its impact

Limits.

In US 746 0951 B2 it is disclosed that the future position of an object can be estimated by means of a Kalman filter based on fused sensor data. It is the object of the present invention to monitor, as far as possible, resource-saving monitoring of a function of a sensor device,

In particular, a sensor device with a plurality of sensor units whose data are fused to provide, and thus meet the demands on a functional safety of the sensor device.

This object is solved by the subject matters of the independent claims. Advantageous embodiments will become apparent from the dependent claims, the description and the figures.

The invention relates to a method for monitoring a function of a

Sensor device of a motor vehicle with a series of steps. To this

Steps include capturing object information about at least one

vehicle external object at a first time and at first

Time following second time by at least one sensor unit of

Sensor device. For example, as the object information about a

vehicle external object a distance d (t _n ) of a preceding vehicle (with the vehicle in front as vehicle external object) at a first time t _n and at a second time t _{n +} i are detected, for example, as a distance d (t _n ) = 5m and

Another part of the method is a storage of internal object information, which corresponds to the object information acquired at the first time, in the arithmetic unit of the sensor device. The internal object information may include or be the detected object information and / or one of the detected object information

Object information include or be derived object information. The internal

Object information can be stored, for example, in the form of a table or a database in which one or more associated information, for example about a position and / or a distance and / or a speed, is stored for respective vehicle-external objects. The deposit can take place in particular before the second time. For example, the above-mentioned distance d (t _n ) may be stored as internal object information before the second time in the arithmetic unit, for example in the form of a list in which one or more off-vehicle objects are kept. Another part of the method is a prediction of object information for the second time point on the basis of the internal object information for the first time, which is therefore assigned to the first time, by the arithmetic unit. This can

for example, by means of a Kalman filter. The forecasting can take place in particular before the second time. Thus, by means of, for example, the Kalman filter, starting from the distance d (t _n ) = 5m, a predicted distance d '(t _{n +} i) can be predicted and estimated for the second time t _{n +} i, for example to d' (t _{n +} i) = 3.5m. In forecasting, additional parameters such as a speed, a speed change and other aspects of the stored object information or object information can be taken into account.

Another step in the process is updating the forecast

Object information on the (then new) internal object information as a function of the object information acquired at the second time and in particular also in

Dependency of the predicted object information by the arithmetic unit. The non-updated predicted object information can continue to be stored in the arithmetic unit. In the above example, the internal object information stored as or with a distance d (t _n ) can be updated as a function of the object information acquired at the second time, in this case the distance d (t _{n +} i). For example, such a distance d (t _n + i) for the time t _{n +} i can be set to 3.0 m. In

Depending on the predicted object information and in particular a further information, for example a reliability of the corresponding sensor information, but could also deviates from the detected object information

Object information, for example, a distance d (t _{n +} i) = 3.1 m deposited.

Finally, a comparison of the predicted for the second time takes place

Object information with the updated, new (internal) object information and a calculation of a respective innovation measure for the at least one object in

Dependency of a result of the comparison by the arithmetic unit. In the comparison, for example, the predicted object information d '(t _{n +} i) = 3.5 m can be updated with the object information d "(t _{n +} i) = 3, 1 updated based on the object information d (t _{n +} i) = 3 m acquired at the second time m are compared, and the

Innovation measure l (t), for example in the form of a difference, in this case, for example, of l (tn ₊ i) = d '(t _{n +} i) -d "(t _{n +} i), calculated for this purpose l (t _{n +} i ) = 0.4m

Innovation refers here to innovation or change

predicted to updated object information and thus to the step of

Updating. It is thus used to check the update process, as in The following becomes clear. The innovation measure can be the so-called innovation known in the context of prediction using a Kalman filter, or it can also generally be a measure of a difference, for example also in a Euclidean difference of object information, for example as a value or vector, for a specific vehicle external Object to be used.

After comparing to respective object information, namely the

predicted and the detected object information at a same time, namely the second time relates, the method described can also be referred to as "in-sequence" monitoring or "in-sequence monitoring".

What is important here is that the degree of innovation by the computing device on a

Plausibility is checked. This can be done, for example, by means of a stored in the arithmetic unit information such as a list of plausible values for the innovation measure or calculation rule as a comparison with a stored comparison value. In the list, for example, a unique identification number can be assigned to each vehicle-external object, so that the components of the object information acquired at different times and / or by different sensor units can be assigned to the respective objects external to the vehicle without errors. Thus, for example, in the described example, an innovation measure I for the time t _{n +} i, which here corresponds to a temporal change of a distance of the object, which is, for example, 1 (t _{n +} i) = 30 m, can be recognized as implausible, since, in a real world Application (of course, depending on the two times t _n , t _{n +} i, or a difference between the two times) such a large change in distance in a short time interval, for example, as a motor vehicle external object of all experience is impossible. In this case, checking the plausibility of the innovation measure by the computing device would thus yield a negative result. The innovation measure of l (t _{n +} i) = 0.4m, however, could lead to a positive result for the time t _{n +} i. If the check yields a negative result, an error message is output by the computing device here. Thus, for example, a sudden change of the distance by 30m indicates that a function of the sensor device of the motor vehicle is disturbed. Thus, based on the error message, the function of the sensor device can be monitored.

This has the advantage that the function of the sensor device with a low

Processor overhead and can be implemented in a particularly simple manner, without additional implementation effort or with little additional implementation effort. The is based on the fact that the invention essentially utilizes already existing data, in this case the detected and estimated, predicted object information or data, which is used by algorithms of the sensor device, in particular algorithms

Sensor data fusion of a sensor device with a plurality of sensors, already be calculated independently of the described method. This prevents or minimizes the generation of an additional processor load in a control unit or an ECU or a computing unit of the sensor device

Development costs reduced. At the same time while the reliability of

Sensor device monitors without the object information would need to be detected redundantly, or a sensor fusion redundant and thus with a large

Computing effort in the arithmetic unit would have to be realized. That's it

described method for monitoring the reliability not only suitable for multisensory data fusion, but also for monitoring the

Functionality of individual sensor elements or sensor devices with only one sensor or only sensors of a modality such as Lidar, radar or optical systems. The proposed method therefore uses data which in principle must already be generated and processed in any target-tracking or object-tracking sensor device or in the customary sensor devices with a multisensory data fusion anyway. The additional computational effort is therefore exhausted in carrying out a comparison of data sets which already exist, for example in the form of intermediate results.

In the method, therefore, three intermediate results, which are often already created in a normal sensor processing with one or more sensors: the internal object information, the predicted object information and the updated object information or the innovation measure, which is a measure of the difference between the predicted object information and of the updated object information. The degree of innovation is inherently already available in many updating algorithms used, for example in the Kalman filter or the various known variants such as the extended and the

uncentered Kalman filters and other variants. Decisive for the application of the method described here is only the use of the internal, the predicted and the updated object information and the calculation of a metric for comparing predicted and updated object information. Which methods for forecasting and updating are used computationally is irrelevant. By means of a simple logic, the differences between the three variables mentioned for a distance and / or a position and / or a ne speed and / or any other property of the corresponding vehicle external object, which is detected and its representation is updated in the sensor device on the object information, simply a limit are defined, beyond which the behavior of the sensor device is classified as implausible and the error message is issued.

In an advantageous embodiment, it is provided that the method is carried out iteratively or continuously. In particular, the second time of an iteration or an iteration stage of the method corresponds to the first time of a

following further iteration or iteration stage of the process. This has the advantage that the functional safety of the sensor device is continuously and thus constantly monitored. About this monitoring is done in a particularly efficient manner, since, for example, the updated object information can be used directly as internal object information of the subsequent iteration. This is the

Monitoring also realized in a particularly resource-saving way.

In a further, particularly advantageous embodiment, it is provided that the object information relates to a plurality of vehicle-external objects, that is to say refers to a plurality of vehicle-external objects or represents the properties of a plurality of vehicle-external objects. This has the advantage that the plausibility checking statistics can be improved by, for example, checking a pool of several respective innovation measures for plausibility, as explained below. Furthermore, this is also favorable for a sensor device with a plurality of different sensor units, since these can often detect different or at least partially different objects. Thus, in a real-world scenario in which a plurality of objects are typically present in an environment of the motor vehicle, the functional safety of the sensor device can be particularly simple and without, for example, a selection process, which includes the

Object information limited to a single vehicle external object, performed.

In a further advantageous embodiment it is provided that the

Object information is detected via a plurality of different sensor units, which in particular have one or more different modalities, and preferably each have a different modality. For example, the

Sensor device in each case one or more sensor units of a lidar, for example a lidar with a laser scanner or a lidar without a laser scanner, and / or a radar and / or a camera system (with one or more

Cameras as sensor units) and / or an ultrasound system (with one or more ultrasonic sensors as sensor units). Thus, a multisensory data fusion can be realized by the sensor device or even must be done for the best possible utilization of the object information. Accordingly, the proposed method is particularly advantageous here, since the conventional known approaches for monitoring the function of a sensor device in a scenario with a plurality of different sensor units are particularly expensive, that is, the above-mentioned disadvantages of a high implementation effort or additionally required sensors.

In another advantageous embodiment it is provided that the

Object information information about a position and / or about a

Speed and / or acceleration and / or distance and / or classification of the particular vehicle external object (s). For example, the respective information may be stored in the form of a vector for one or more, in particular all, vehicle-external objects that are represented in the object information. The choice of a vector as a representation is particularly advantageous here, since a change or a difference between different vectors is particularly easy to handle and quantify mathematically. The above information has the advantage that they characterize the vehicle-external objects particularly well and especially for this important information should be ensured in the context of functional safety in each case that they are correct, that is, for example, the updating of the object information works as intended.

In a particularly advantageous embodiment it is provided that the checking comprises a comparison of the calculated innovation measure with a limit value or a limit measure, for example also a vector or limit vector, and

in particular, a result of checking is negative if the calculated

Innovation measure is greater than the limit, and positive if the calculated

Innovation measure is smaller than the limit. The comparison can be realized here, for example, by differentiating as in the Kalman filter or by forming a Euclidean difference. This has the advantage that the plausibility of the innovation measure can be checked with a simple, low-resource-binding comparison. In another advantageous embodiment, provision is made for the innovation measure to be calculated for a plurality of vehicle-external objects which relate to the object information, in particular for all vehicle-external objects which concern the object information and / or for a plurality of points in time for which the object information is updated is, in particular for all times for which the object information is updated. A plurality of respective innovation dimensions l _k for a plurality of respective objects are thus calculated k. This has the advantage that a malfunction of the

Sensor device can be detected particularly reliable and efficient, since in all probability a functional malfunction does not occur only once at a single time and / or in a single vehicle external object, but typically repeats systematically and / or occurs simultaneously for several vehicles external objects. By means of corresponding correlations in the respective innovation measures l _k (t), a malfunction of the sensor device can thus be detected particularly well.

In a particularly advantageous embodiment, it is provided that the checking delivers a negative result, if the respective innovation measure

(In particular, for one or more predetermined times) a predetermined minimum number of vehicle-external objects and / or a predetermined

Minimum proportion of the vehicle-external objects represented in the object information is greater than a predetermined limit or the predetermined limit, and / or if the innovation measure for an object repeated, for example according to a recognizable pattern or even irregularly, is greater than the predetermined limit and / or if the number of vehicle-external objects changes by more than a predetermined number (which can also be zero) for which the respective innovation measure is greater than the predetermined limit value. It has been found that a malfunction of the sensor device can be monitored particularly reliably with the conditions mentioned for the degree of innovation, since systemic errors in the sensor device are reflected very rapidly or in a very obvious manner in a change of the named criteria.

In a further advantageous embodiment, it is provided that an additional comparison of the object information stored for the first point in time, ie the internal object information of the first time, with the updated object information (of the second time) takes place and calculating a deviation measure for the at least one vehicle external object Dependence of a result of the

Compare done by the arithmetic unit. The deviation measure serves this purpose Reviewing the forecast in a similar way as the innovation measure serves to validate the update. The deviation measure will therefore generally be greater than the innovation measure for a given updated object information.

Correspondingly, a plausibility of the deviation measure is checked by the computing device and an error message is output by the computer

Computation device, if the check provides a negative result, so that

Deviation measure is not plausible. In the example described above, for example, the distance d (t _n ) would be compared with the updated distance and thus calculated, for example, for the deviation measure A _k (t _{n +} i) = d (tn) -d (t _{n +} i) = 2 m.

This is particularly advantageous that, as already described in the introduction

Only the already existing object information is evaluated, so that no additional data must be generated or recorded. Since that

Deviation measure with a same or similar algorithm as that

Innovation measure can be calculated, only just starting from different object information, here are no new elaborate steps in the

Implementation necessary. For example, in the comparison, a difference as known in the Kalman filter for the innovation or a Euclidean difference can be calculated. Since the comparison is based on object information from different points in time or with different timestamps, this can be referred to as "out-of-sequence" monitoring or "out-of-sequence monitoring" in contrast to the checking based on the innovation measure.

For out-of-sequence monitoring, however, the benefits described at the outset for in-sequence monitoring also apply. Accordingly, similar to the embodiments of the method with respect to the comparison, calculation and checking of the innovation measure, the comparison and the calculation and the checking of the deviation measure can also be made more advantageous. This will be in

Set out below. The advantages achieved thereby mutatis mutandis correspond to the corresponding advantages in comparing, calculating or checking the innovation measure. These can be combined with the presence of

corresponding features also interact synergistically. Basically, the out-of-sequence monitoring would also be feasible independently of the in-sequence monitoring, but since the deviation measure itself provides a coarser monitoring as explained, the implementation together with the in-sequence monitoring makes sense, since errors in the forecasting process and in the Update process can be observed separately from each other in their influence.

In an advantageous embodiment, it is provided that the checking of the plausibility of the deviation measure is a comparison of the calculated

Deviation measure with a further limit, in particular a limit which is greater than the threshold for the innovation measure, and in particular a result of the check is negative if the calculated deviation measure is greater than the further limit, and positive if the calculated deviation measure is smaller as the further limit.

In a further, particularly advantageous embodiment, it is provided that the deviation measure is calculated for a plurality of, in particular all, vehicle-external objects which relate to the object information, and / or for a plurality, in particular all, times for which the object information is updated. Thus, a plurality of respective deviation measures A _k for the respective objects k are calculated.

In this case, it can be provided in a further advantageous embodiment that the check provides a negative result if the respective deviation measure (in particular for one or more times) for a predetermined minimum number of vehicle-external objects and / or a predetermined minimum proportion of the vehicle-external objects is greater than another predetermined further limit value or the predetermined further limit value and / or if the deviation measure for an object is repeated, for example according to a predetermined pattern or also irregularly, is greater than the predetermined further limit value and / or if the number of objects by more than one changed predetermined number for which the respective deviation measure is greater than the predetermined further limit.

In a further advantageous embodiment, it is provided that both times, the first and the second time, are directly successive points in time. This not only has the advantage of close monitoring of the function of the

Sensor device, but also the advantage that the expected changes and thus the degree of innovation and also the deviation measure are smaller in magnitude than at times with a large time difference, and any measurement or prediction errors have less influence, so that the function of

Sensor device can be monitored very accurately and precisely. The invention also relates to a sensor device of a motor vehicle. The

In this case, the sensor device has a sensor unit for detecting object information about at least one vehicle-external object at a first time and at a further second time following the first time. It is the

Sensor device designed to store the object information detected at the first time as an internal object information in a computing unit of

Sensor device. The arithmetic unit is designed to predict object information for the second time based on the internal object information, as well as to update the predicted object information to the new internal

Object information as a function of the detected at the second time

Object information and for comparing the object information predicted for the second time with the updated object information and finally for calculating an innovation measure for the at least one object as a function of

Result of comparing. It is important here that the computing device is designed to check a plausibility of the innovation measure and to issue an error message if the check provides a negative result.

The invention also relates to a sensor device which is suitable for carrying out a

inventive method or an advantageous embodiment of the

inventive method is formed.

Advantages and advantageous embodiments of the sensor device according to advantages and advantageous embodiments of the method for monitoring a function of a sensor device of a motor vehicle.

The invention also relates to a motor vehicle with such a sensor device.

The features and combinations of features mentioned above in the description as well as the features and feature combinations mentioned below in the description of the figures and / or shown alone in the figures can be used not only in the respectively specified combination but also in other combinations, without departing from the scope of the invention , Thus, embodiments of the invention are to be regarded as encompassed and disclosed, which are not explicitly shown and explained in the figures, but which emerge and can be produced by separated combinations of features from the embodiments explained. There are also versions and

Feature combinations are disclosed as disclosed, which thus do not have all the features of an originally formulated independent claim. It's beyond that also embodiments and combinations of features, in particular by the above-mentioned embodiments, to be regarded as disclosed that go beyond the feature combinations set out in the back covers of the claims or deviate from these.

Embodiments of the invention are explained in more detail below with reference to schematic drawings.

Showing:

Fig. 1 is a schematic representation of a motor vehicle with a

exemplary embodiment of a sensor device; and

Fig. 2 is a schematic representation of a flowchart of an exemplary

Embodiment of a method for monitoring a function of a sensor device of a motor vehicle.

Identical or functionally identical features are provided in the figures with the same reference numerals.

1 shows a schematic representation of a motor vehicle with an exemplary embodiment of a sensor device. In the present case, the motor vehicle 1 comprises a sensor device 2 which has at least one, in the present case two sensor units 3, 4 of different modality. In the present case, the first sensor unit 3 is designed as an ultrasonic sensor unit for detecting a distance d to an object external to the vehicle k, and the second sensor unit 4 as a camera for detecting the distance d and a position of a vehicle-external object k. The sensor device 2 also has a computing unit 5, in which an internal object information O _k (FIG. 2), which corresponds to a detected object information o _k (FIG. 2), can be stored. Furthermore, the arithmetic unit 5 for predicting an object information 0 ' _k (FIG. 2) is designed on the basis of the internal object information O _k and for

Updating the predicted object information 0 ' _k (FIG. 2) to the new internal object information O _k as a function of a detected object information o _k . In this case, by the arithmetic unit 5, a comparison of the predicted object information 0 ' _k with the updated stored object information O _k and a calculation of a Innovation measure l _k (Fig. 2) are calculated for the at least one of the object information represented object k in dependence on a result of the comparison. It is important that the computing device 5 is designed to check a plausibility of the innovation measure l _k and to issue an error message if the check provides a negative result, so as to realize a monitoring of the functional safety of the sensor device 2.

The sensor device can thus be suitable for one of the methods as explained with reference to FIG. 2.

2 shows a schematic flowchart of the example method for monitoring a function of a sensor device. First takes place here detecting 10a - 10g object information o _k (t _n t _{n +} i) on at least one vehicle external object k. Detecting 10a-10g takes place at least at a first time t _n and at a second time t _{n +} i following the first time t _n . In this case, for example, a first detection 10a can take place by a first sensor unit 4 (FIG. 1), for example a camera, and, for example, a further, in this case sixfold, detection 10b-10g can be performed by a second sensor unit 3 (FIG can be designed as an ultrasonic sensor unit. Thus, a multimodal data fusion can be realized in the further process.

A next step in this case is a storage 1 1 of an internal object information 0 (t _n ) which corresponds to the object information o (t _n ) acquired for the first time t _n , that is to say, for example, or is derived therefrom, in the arithmetic unit 5 of the sensor device 2. This is followed by a prognostic 12 of an object information 0 '(t _{n +} i) for the second time t _{n +} i on the basis of the internal object information 0 (t _n ) by the arithmetic unit 5. In a further step, then updating 13 of the predicted or predicted Object information 0 '(t _{n +} i) to the new internal

Object information 0 (t _{n +} i). This updating is performed in dependence of the 13 at the second time t _{n +} i o detected object information (t _{n +} i) by the computing unit 5, and provides an updated object information 0 (t _{n +} i). This updated object information 0 (t _{n +} i) can be provided by the computing unit 5 to further units, as symbolized by the arrow 14. At the same time, in an iterative or continuous execution of the method, the updated internal object information 0 (t _{n +} i) can also be used as new output object information for the next iteration, as symbolized by the arrow 15. However, this is not mandatory since the detection 10a-10g both at the first time t _n as also takes place at the second time t _{n +} i. Thus, at each instant t _{n +} i, both detected object information o (t _n ), o (t _{n +} i) are always available from the time t _n and t _{n +} i, as symbolized by the arrow 1 6. The update 13 may be calculated based on the Kalman algorithm. The object information 0 '(t _{n +} i) predicted for the second time instant t _{n +} i is then compared with the updated object information 0 (t _{n +} i) and an innovation measure l _k [0' _k (t _{n +} i) is calculated. ;

O _k (t _{n +} i)] for the at least one object k as a function of a result of

Comparing by the arithmetic unit 5. Then carried out by the computing device 5, a check 18 a plausibility of the innovation measure l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i)] as well as output 1 9 an error message if the check 1 8 gives a negative result.

For example, in the checking 18, a deviation between the

predicted object information 0 '(t _{n +} i) and updated object information 0 (t _{n +} i) are compared with a predetermined limit value and an error message is output if the deviation is greater than the predetermined limit value. Alternatively or additionally, other criteria can also be checked or specified. Thus, the additional expense which must be operated for monitoring the sensor device 2, only a simple comparison 1 7 with a minimum of additional computational effort. Since the comparison 17 here refers to object information O, O 'of the same time t _{n +} i, this monitoring can be referred to as "in-sequence" monitoring.

In in-sequence monitoring, the innovation measure _lk [ _0'k (tn ₊ i); _Ok (tn ₊ i)]

presently calculated for all objects k that are affected by the updating 1 3. For example, this may already be implicit in the form of calculating the so-called innovation when using the Kalman algorithm already in the context of forecasting 12. Therefore, the innovation measure l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i)] is available for every detected object k for each sub-step t. For every measure of innovation l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i)] of an object k at time t _{n +} i can now be checked, for example, whether it is above a limit value Di. For example, a counter N may count the number of objects k at a time t _{n +} i for which the innovation measure l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i)] is greater Di. Thus it can be checked, for example, for each time step t, what proportion of the objects k is an innovation measure l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i)] is greater than D, and / or what is the total number of objects k with such a large deviation. Based on these data, it can now be determined whether the innovation measure innovation measure l _k [0'k (t _{n +} i); O _k (t _{n +} i)] or the innovation measures l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i)] are plausible. The plausibility can therefore not only for every object k in the individual time step t _n be checked, but in addition to the following time steps t _{n +} i, t _{n + 2} , etc.

For example, checking 18 the innovation measures _lk [ _0'k (tn ₊ i); O _k (t _{n +} i)] be negative, if an identical object k, which already in a previous time step t, an innovation measure l _k [0 ' _k (ti); O _k (t,)] is greater than D, this also has t _{i +} i, t _{i + j} in one or more subsequent time steps. It can also be checked independently or in addition, whether the total number N changes over time. With these two measures it can be decided whether a processing of the

provided or acquired object information o _k (t _n ; t _{n +} i), in particular a fusion in the of different sensor units 3,4 provided or detected

Object information o _k (t _n t _{n +} i), is carried out correctly or not. This can be recorded and quantified both for individual time steps or times and over time. Thus, a gradual deterioration of a processing of the object information o _k (t _n ; t _{n +} i), for example, a sensor fusion over time can be monitored and detected.

Similarly, in addition to the in-sequence monitoring so-called out-of-sequence monitoring can be realized. For this purpose, an additional comparison 20 of the internal object information O _k (t _n ) of an object k takes place at the time t _n with the

updated object information O _k (t _{n +} i) of the object k at time t _{n +} i. In this case, depending on the comparison 20, a deviation measure A _k [O _k (t _n ); O _k (t _{n +} i)]. This can be done in a similar manner and for example even with the same or similar algorithm or with an identical or similar algorithmic one

Framework are like the innovation measure l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i)] is calculated. It should be noted that the deviation measure A _k [O _k (t _n ); O _k (t _{n +} i)] is larger than the innovation measure l _k [0 ' _k (t _{n +} i) assigned to this object k, in particular for this object k; _Ok (tn ₊ i)]. Accordingly, it is advantageous if a corresponding

Check 21 the plausibility of the deviation measure A _k [O _k (t _n ); O _k (t _{n +} i)] comprises a comparison with a further limit value, to select this further limit value D _A greater than the limit value D, for the innovation measure l _k [0 ' _k (t _{n +} i); _Ok (tn ₊ i)]. Deviation measure A _k [O _k (t _n ); O _k (t _{n +} i)] and innovation measure l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i)] can be used for the

Error handling or error analysis can be evaluated in combination.

In this case, in calculating the deviation measure A _k [O _k (t _n ); O _k (t _{n +} i)] an additional consistency check between the respective internal object information O _k at time t _n and the internal object information O _k at time t _{n +} i provided be. This can be done according to a predetermined acceptance criterion.

For example, this review can be done simply by taking advantage of one

Association information of updating 13 done, which is already known for example from a Kalman filter. In the mentioned manner, other combinations of available data for the comparison for calculating the innovation measure _lk [ _0'k (tn ₊ i); O _k (t _{n +} i)] and the deviation measure A _k [O _k (t _n ); O _k (t _{n +} i)] conceivable. These then depend on specific conditions of the respective environment and on the available computing capacity of the computing device 5.

Claims

claims

1 . Method for monitoring a function of a sensor device (2) of a

Motor vehicle (2), with the steps:

a) detecting (10a-10g) object information (o _k (t _n ; t _{n +} i)) via at least one vehicle external object (k) at a first time (t _n ) and at one of the first time (t _n ) following second time (t _{n +} i) by at least one sensor unit (3, 4) of the sensor device (2);

b) depositing (1 1) an internal object information (O _k (t _n )) which corresponds to the object information (o _k (t _n )) acquired at the first time (t _n ) in a computing unit (5) of the sensor device ( 2);

c) predicting (12) of an object information (0 _'k (t _{n +} i)) for the second time (t _{n +} i) _k based on the internal object information (O (t _n)) (by the computing unit 5); d) updating (13) the predicted object information ((0 _'k (t _{n +} i)) for the new internal object information (O _k (t _{n +} i)) in dependence of at the second time (t _{n +} i) Object information detected (o _k (t _{n +} )) by the arithmetic unit (5);

e) comparing (17) the predicted for the second time point (t _{n +} i)

Object information (0 _'k (t _{n +} i)) with the updated object information (O _k (t _{n +} i)) and calculating an innovation dimension (l _k [0' _k (t _{n +} i) O _k (t _{n +} i)]) for the at least one object (k) as a function of a result of the comparison (17) by the arithmetic unit (5);

characterized by a

f) checking (18) a plausibility of the measure of innovation (l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i)]) by the computing device (5); and

g) outputting (19) an error message by the computing device (5) if the checking (18) delivers a negative result.

2. The method according to claim 1

characterized in that

the method is carried out iteratively and in particular the second time (t _{n +} i) of an iteration of the method corresponds to the first time (t _n ) of a subsequent further iteration of the method.

3. The method according to any one of the preceding claims,

characterized in that

the object information (o _k (t _n t _{n +} i), O _k (t _n), 0 _'k (t _{n +} i), O _k (t _{n +} i)) relates to a plurality of vehicle-external objects (k).

4. The method according to any one of the preceding claims,

characterized in that

the object information (o _k (t _n ; t _{n +} i)) is detected via a plurality of sensor units (3, 4), which in particular have a different modality.

5. The method according to any one of the preceding claims,

characterized in that

the object information (o _k (t _n t _{n +} i), O _k (t _n), 0 _'k (t _{n +} i), O _k (t _{n +} i)) information about a position and / or speed and / or an acceleration and / or a distance (d) and / or a classification of the relevant vehicle-external object (k).

6. The method according to any one of the preceding claims,

characterized in that

checking (18) comparing the calculated innovation measure

( _lk [ _0'k (tn ₊ i); _Ok (tn ₊ i)]) with a threshold (D,) and in particular a result of the check (18) is negative if the calculated innovation measure ( _lk [ 0 ' _k (tn ₊ i); _Ok (tn ₊ i)]) is greater than the threshold (D,), and positive if the calculated innovation measure ( _lk [ _0'k (tn ₊ i); _Ok (t _{n +} i)]) is less than the limit (D,).

7. The method according to any one of the preceding claims,

characterized in that

the Innovationsmaß (l _k [0 _'k (t _{n +} i) O _k (t _{n +} i)]) for a plurality of vehicle-external objects (k), in particular all vehicle-external objects (k) which is the object information (0' _k (t _{n +} i), O _k (t _{n +} i)), and / or for several times (t _n ; t _{n +} i), in particular all times (t _n ; t _{n +} i) for which the object information (O _k (t _n ), O _k (t _{n +} i)) is calculated.

8. The method according to claim 7

characterized in that

checking (18) gives a negative result if the respective one

Innovation measure ( _lk [ _0'k (tn ₊ i); _Ok (tn ₊ i)]) of a given minimum number of Vehicle external objects (k) or a predetermined minimum proportion of the vehicle external objects (k) is greater than a predetermined limit (D,) and / or if the innovation measure (l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i )]) for an object (k) is repeatedly greater than the predetermined limit value (D,) and / or if the number of objects (k) changes by more than a predetermined number, for which the respective innovation measure (l _k [0 ' _k (tn ₊ i); _Ok (tn ₊ i)]) is greater than the given one

Limit value (D,) is.

9. The method according to any one of the preceding claims,

characterized by a

Comparing (20) the object information (O _k (t _n )) stored for the first time (t _n ) with the updated object information (O _k (t _{n +} i)) and calculating a deviation measure (A _k [O _k (t _n O _k (t _{n +} i)]) for the at least one vehicle-external object (k) as a function of a result of the comparison (20) by the arithmetic unit (5), as well as a

Checking (21) a plausibility of the deviation measure (A _k [O _k (t _n ); O _k (t _{n +} i)]) by the computing device (5); and

- Output (22) of an error message by the computing device (5) if the check (21) provides a negative result.

10. The method according to claim 9,

characterized in that

checking (21) comparing the calculated deviation measure

(A _k [O _k (t _n ); O _k (t _{n +} i)]) with a further limit value (D _A ) and in particular a result of the checking (21) is negative, if the calculated

Deviation measure (A _k [O _k (t _n ); O _k (t _{n +} i)]) is greater than the further limit value (D _A ), and positive if the calculated deviation measure (A _k [O _k (t _n ); O _k (t _{n +} i)]) is smaller than the further limit value (D _A ).

1 1. Method according to one of claims 8 to 10,

characterized in that

the deviation measure (A _k [O _k (t _n ); O _k (t _{n +} i)]) for a plurality of, in particular all, vehicle-external objects (k) containing the object information (O _k (t _n ), O _k (t _{n +} i)), and / or for several, in particular all, times (t _{n +} i) for which the

Object information (O _k (t _{n +} i)) is updated is calculated.

12. The method of claim 1 1

characterized in that

checking (21) gives a negative result if the respective one

Deviation measure (A _k [O _k (t _n ); O _k (t _{n +} i)]) for a predefined minimum number of vehicle-external objects (k) or a predetermined minimum proportion of the vehicle-external objects (k) is greater than another predetermined further limit value ( D _A ) and / or if the deviation measure (A _k [O _k (t _n ); O _k (t _{n +} i)]) for an object (k) is repeatedly greater than the predetermined further limit value (D _A ) and / or if the number of objects (k) changes by more than a predetermined number, for which the respective deviation measure (A _k [O _k (t _n ); O _k (t _{n +} i)]) is greater than the predetermined further limit value (D _A ) is.

13. The method according to any one of the preceding claims,

characterized in that

both times (t _n ; t _{n +} i) are directly consecutive times.

14. Sensor device (2) for a motor vehicle (1), with

- A sensor unit (3, 4) of the sensor device (2) for detecting (10a-10g) object information (o _k (t _n ; t _{n +} )) via at least one external vehicle object (k) at a first time (t _n ) and at a second time (t _{n +} i) following the first time (t _n ); in which

- The sensor device (2) is designed to store (1 1) the at the first time (t _n ) detected object information (o _k (t _n )) as an internal object information (O _k (t _n )) in a computing unit (5 ); and with

- the arithmetic unit (5) for predicting (12) an object information (0 ' _k (t _{n +} i)) for the second time (t _{n +} i) on the basis of the internal object information (O _k (t _n )), for updating (13) the predicted object information (0 _'k (t _{n +} i)) for the new internal object information (O _k (t _{n +} i)) in dependence of at the second time (t _{n +} i) object information detected (o _k (t _{n +} i)), and for comparing (17) the object information (0 ' _k (t _{n +} i)) predicted for the second time (t _{n +} i) with the updated object information (O _k (t _{n +} i)) and calculating an innovation measure (l _k [0 ' _k (t _{n +} i); O _k (t _{n +} i)]) for the at least one object (k) in response to a result of the comparison (17);

characterized in that

the computing device (5) is designed, a plausibility of the innovation measure (lk [0'k (tn ₊ i); Ok (tn ₊ i)]) and to issue an error message if the check (18) gives a negative result.

15. Motor vehicle (1) with a sensor device (2) according to claim 14.