Identifying characteristics of a scene, with a high level of safety integrity, by comparing sensor data with an expectation
Patent Information
- Application Number
- EP2022830793
- Authority / Receiving Office
- EP · EP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2022-01-10
- Filing Date
- 2022-12-07
- Publication Date
- 2024-11-20
AI Technical Summary
Current detection systems for autonomous vehicles and robots, relying on neural networks, struggle with safety integrity due to limitations in recognizing unusual or rare objects and susceptibility to manipulation, leading to potential safety hazards.
A method that compares spatially resolved actual sensor data with a stored expectation from a map, using sensor data from various modalities like radar, lidar, and cameras, to determine the presence of specific properties in the environment, ensuring accurate recognition of passable areas and robustness against manipulation by checking consistency across different sensor modalities and allowing for tolerance and abstraction in comparisons.
This approach enhances safety integrity by reliably identifying passable areas and preventing unsafe navigation, even in the presence of unusual objects or manipulation attempts, by ensuring that only areas consistently matching the expectation are recognized as passable, thereby improving the appropriateness of vehicle or robot control.
Smart Images

Figure 1.1
Abstract
Description
[0001] Description
[0002] Title:
[0003] Detection with higher safety integrity of characteristics of a scene by comparing sensor data with an expectation
[0004] The present invention relates to the recognition of properties of a scene observed with at least one sensor, for example for the at least partially autonomous control of a vehicle or robot.
[0005] State of the art
[0006] Driver assistance systems and systems for the at least partially automated driving of vehicles or robots detect the environment of the vehicle or robot using one or more sensors and use this information to determine a plan for the future behavior of the vehicle or robot. Neural networks are often used to determine such a plan, or an environment representation as a preliminary product for the planning. DE 10 2018 008 685 A1 discloses a method for training a neural network to determine a path prediction for a vehicle.
[0007] Disclosure of the invention
[0008] The invention provides a method for evaluating spatially resolved actual sensor data recorded with at least one sensor. This sensor can, for example, be a mobile sensor carried by a person, a robot, or any land, water, or air vehicle. However, the sensor can also be, for example, a stationary sensor that monitors a busy intersection. Spatially resolved actual sensor data is understood to mean any sensor data that is assigned to specific locations during its recording. For example, radar and lidar data can be present as point clouds of measured values assigned to points in three-dimensional space from which the respective interrogation radiation was reflected. Images assign intensity values to the pixels of the sensor used for their recording, which are usually arranged in a regular grid.Depending on the perspective from which the image was taken and the optics used, each pixel corresponds to a location in space from which the incident light comes.
[0009] The location and orientation of the sensor at the time the sensor data is recorded are determined. This process is also called "registration" or "localization" of the sensor in the physical world. Any number of systems can be used for this purpose, either individually or in combination.
[0010] For example, navigation systems based on radio signals emitted by satellites or terrestrial stations can be used. Alternatively, or in combination, an inertial navigation system can be used. If the sensor is stationary, its registration or localization is particularly easy because its location and orientation are known in advance.
[0011] A spatially resolved expectation for the actual sensor data is retrieved from a spatially resolved map based on the location and orientation of the sensor. This expectation is stored in the spatially resolved map in association with at least one property of the observed scene. If the scene observed with at least one sensor actually exhibits this property stored in the map, the actual sensor data is expected to be consistent with the expectation retrieved from the spatially resolved map.
[0012] Conversely, this means that if the actual sensor data matches the expected value, the property stored in the map is also present. Checking whether this property is present is precisely the goal of the procedure.
[0013] Therefore, it is examined to what extent the actual sensor data are consistent with the expectation. At least for those locations where this is the case, it is determined that the scenery observed with at least one sensor exhibits the property stored in the map in connection with the expectation.
[0014] The expectation stored in the map represents a "fingerprint" of the scene observed with the sensor. This "fingerprint" can include any properties that can be analyzed from sensor data, such as geometry, texture, or multispectral response. Fundamental physical properties, such as magnetic resonance, can also be considered as analyzable properties.
[0015] The sensor data can, for example, be sensor data recorded by a sensor on a vehicle or robot. The vehicle can be any land, water, or air vehicle. An important application of the method in connection with the control of vehicles and robots is the examination of which locations can be freely accessed by the vehicle or robot. For this purpose, the property stored in the map can include a statement as to the extent to which locations to which the expectation refers can be freely accessed by the vehicle or robot. In the same way, the method can also be applied with mobile sensors that are carried by a person, for example to signal the accessible areas to a blind person.
[0016] This can be illustrated by a simple example in which camera images are used as sensor data. Camera images are then recorded on the map, and these images are labeled to indicate which areas are freely accessible to the vehicle and which are not. If, while the vehicle is driving, certain areas are recognized at the correct locations based on the camera images currently recorded by the vehicle, this guarantees that the information stored in the map about these areas, stating whether or not they are freely accessible to the vehicle, corresponds to the current state of these areas. If, for example, there is an obstacle at a certain point in the camera images recorded during the journey and this obstacle deviates from the same property in terms of at least one property, such as its optical appearance or its geometry, in the same way as the image stored in the map, this object can be reliably detected.
[0017] In this way, an open classification task as to whether an area is freely accessible (or whether any other property important for journey planning is present there) can be reduced to a comparison with known information stored in the map. It was recognized that the detection of freely accessible areas can be carried out with significantly better safety integrity in this way. If a spatial area at a specific location produces the same or at least similar sensor data as that stored as an expectation in the map, it necessarily follows that the area has the same occupancy of objects compared to its state when the map was created (e.g., it is free of objects and therefore accessible) and that there are no other objects between the sensor and this area.Any unplanned object in the area in question or in the line of sight between the sensor and the area destroys the correspondence between actual sensor data and the expected value and thus results in the area no longer being identified as freely passable.
[0018] Compared to an open classification task, such as that solved with neural networks, this has the advantage that a restricted area is recognized as restricted even if the reason for its non-accessibility is extremely unusual and therefore does not appear in the training data used to train the neural network. Events that are too improbable to appear in the training data are also referred to as "long tails."
[0019] For example, very unusual objects, such as furniture, large electrical appliances, skis, or bicycles, are sometimes lost on highways due to inadequate load securing. The appearance of such objects on the roadway traveled by one's own vehicle is a highly dangerous situation and therefore fortunately occurs only rarely. However, this also means that such examples are highly unlikely to occur when collecting real data for training neural networks during test drives. Consciously recreating such situations in public transport is impractical.
[0020] The same applies to the dreaded "blow-ups" that suddenly occur during heat waves, where the concrete roadway bulges or cracks. Such situations cannot be trained for, as the moment a "blow-up" appears in the camera image, an accident is almost impossible to prevent.
[0021] Furthermore, the detection of freely passable areas by recognizing them on the map is not susceptible to deliberate manipulation. Many image classifiers based on neural networks can be caused to output an incorrect classification by maliciously introducing noise patterns. For example, a stop sign can be manipulated by attaching a seemingly inconspicuous sticker to it, causing it to be classified as a "70 km / h" sign. Experiments have even shown that the ability of the downstream image classifier to detect pedestrians has been completely disabled by attaching a film with an inconspicuous, semi-transparent dot pattern to a camera lens. The pedestrians were classified as freely passable areas.
[0022] Any attempt at such manipulation is either completely ignored within the framework of the procedure presented here or, in the worst case, results in an otherwise freely passable area not being recognized as such. Any unexpected problem therefore leads to the affected area being avoided instead of being entered ("fail-safe").
[0023] Furthermore, the expectation can already be used to establish an upper limit for those areas that can be recognized as freely passable. The recognition of an area in the expectation can only trigger a classification of this area as freely passable if the area was marked as freely passable in connection with the expectation. Areas that have not been marked as freely passable, such as concrete barriers or trees at the edge of the road, can never be recognized as freely passable. The significantly improved safety integrity when determining properties of the scenery observed with at least one sensor means that the control of a vehicle or robot based on the properties determined in this way is more likely to be appropriate to the respective situation.Therefore, a control signal for the vehicle or robot is advantageously determined based on the determination of the locations for which the scene observed by the sensor exhibits the property stored in the map in conjunction with the expectation. The vehicle or robot is controlled with this control signal, so that the driving dynamics of the vehicle or robot are influenced according to the control signal.
[0024] The comparison of the actual sensor data with the expected value is not limited to 1:1 recognition. Instead, this recognition can be subject to any kind of tolerance, or even an abstraction based on specific features. For example, even two camera images of the same scene taken immediately after one another are usually not completely identical.
[0025] In another particularly advantageous embodiment, the actual sensor data and the expected values are transferred into a common spatial reference system and / or a common workspace. The actual sensor data are compared with the expected values in this reference system or workspace. In this way, sensor data and expected values that were acquired using different modalities can also be compared. For example, the expected values
[0026] • a spatially resolved three-dimensional geometry, and / or
[0027] • a texturing, and / or
[0028] • a reflectance amplitude, and / or
[0029] • a multispectral response, and / or
[0030] • contain a magnetic resonance of the scene observed by at least one sensor. Such a geometry can be determined, for example, based on image recordings. Such a geometry can be easily labeled with regard to freely accessible areas or other characteristics of the scene. It can then be checked, for example, to what extent radar or lidar data are consistent with this geometry.
[0031] In addition to radar sensors and lidar sensors, stereoscopically arranged cameras or multi-camera systems are also considered for capturing sensor data. Such camera arrangements also provide depth information that can be checked against the expected geometry. Moving monocular cameras are also conceivable for generating depth information. Furthermore, multispectral cameras, time-of-flight (ToF) sensors, cameras with event-based pixels, ultrasonic sensors, or even magnetic sensors can be used.
[0032] For example, testing whether sensor data captured by a stereoscopic camera array is consistent with the expected three-dimensional geometry may involve testing the so-called stereo hypothesis. This test is based on the fact that the geometry of the scene couples the images provided by both cameras in the stereoscopic camera array. Thus, if one of the images and the expected geometry are present, the other image is at least largely determined.
[0033] Therefore, an image provided by the first camera of the camera array can be transformed based on the expected geometry into an expected image for the second camera of the camera array. It can then be checked to what extent this expectation is consistent with an image actually provided by the second camera of the camera array. The transformation can, in particular, include, for example, distorting the image provided by the first camera based on the geometry so that it matches the perspective of the second camera.
[0034] To compare the image provided by the second camera with the expected image, features can be extracted, for example, from the image provided by the second camera on the one hand and from the expected image for this image on the other. These features can then be compared with each other. This abstraction into features can smooth out insignificant differences for the comparison, for example, with regard to colors or lighting.
[0035] In particular, for example, a binary decision can be made for each feature to be compared as to whether a feature from the image is consistent with the corresponding feature from the expectation for that image. The number of consistent features can then be used to determine the degree of agreement between the image and the expectation. For example, a Hamming distance can be determined between the examined combinations of features. This distance increases the more features of the image and the expectation are inconsistent with each other.
[0036] In a further advantageous embodiment, an additional test is carried out to determine the extent to which a predetermined test image, which does not show the scene observed by the sensor, corresponds to the expectation for the image supplied by the second camera in the camera arrangement. This degree of correspondence is then used as the noise level for the determined correspondence between the image supplied by the second camera in the camera arrangement and the expectation for this image. In this way, a signal-to-noise ratio can be determined for the correspondence between the image supplied by the second camera and the expectation for this image. This signal-to-noise ratio is more meaningful than the correspondence alone. For example, the meaningfulness of the image may be reduced if large parts of it have become saturated due to over- or underexposure.
[0037] The method can also be generalized to the extent that actual sensor data is recorded using multiple sensor modalities and the results are then combined. Therefore, in a further advantageous embodiment, for actual sensor data recorded using multiple different sensors, a separate check is carried out to determine for each location for which this actual sensor data is consistent with the expectation retrieved from the map. It is then determined that the actual sensor data as a whole is consistent with the expectation only for those locations for which the actual sensor data from all sensors is consistent with the expectation. For example, an area is only considered freely accessible by a vehicle or robot if it has been independently identified as freely accessible on the basis of the sensor data provided by multiple sensors of different sensor modalities (e.g., lidar and stereoscopic camera).
[0038] For example, it can first be checked to what extent current lidar data are consistent with a geometry of the scene stored in the map. In parallel, it can be checked, for example, to what extent images provided by a stereoscopic camera arrangement are consistent with this geometry. For this purpose, the geometry can be transformed into the reference system of the first camera in the camera arrangement. The stereo hypothesis can then be tested, as described above, by transforming the image provided by the first camera in the camera arrangement based on the geometry into an expectation for the image provided by the second camera in the camera arrangement and comparing this expectation with the image actually provided by the second camera.Only for locations for which both the lidar data and the images provided by the stereoscopic camera array are consistent with the geometry stored as an expectation in the map can it then be determined that these locations have the desired property (such as the free accessibility of these locations) according to the map.
[0039] In a further advantageous embodiment, actual sensor data, the agreement with the expectation being checked, is checked for plausibility against actual sensor data recorded with another sensor. Agreement with the expectation is then only determined or maintained for those locations for which this plausibility check is positive. Thus, only one comparison instead of two takes place between one sensor modality and the expectation. Otherwise, the two sensor modalities are compared with each other. For example, an area can only be declared freely passable if, firstly, it turns out to be freely passable based on the comparison of lidar data with the geometry stored as an expectation and, secondly, if the lidar data in this area is consistent with images provided by a stereoscopic camera arrangement.
[0040] In a further advantageous embodiment, a determined location and / or a determined orientation is optimized with the aim of maximizing the agreement between the actual sensor data and the expectation. As explained previously, in order to compare actual sensor data with the expectation, it is important that the expectation for the correct location and the correct orientation of the sensor is retrieved from the map. Only then can the expectation be correctly recognized based on the current actual sensor data. However, every method for determining the location and orientation has a limited accuracy. If, for example, an additional shift of the determined location of the sensor and / or an additional tilting or rotating of the determined orientation of this sensor can significantly improve the agreement between the sensor data and the expectation, this indicates that the previously determined location orThe previously determined orientation of the sensor was not entirely accurate. Alternatively, or in combination with this, any other technique can be used to convert the actual sensor data and the map into a common reference system.
[0041] The method can, in particular, be fully or partially computer-implemented. Therefore, the invention also relates to a computer program with machine-readable instructions that, when executed on one or more computers, cause the computer(s) to execute the described method. In this sense, control units for vehicles and embedded systems for technical devices that are also capable of executing machine-readable instructions are also considered computers.
[0042] The invention also relates to a machine-readable data carrier and / or a downloadable product containing the computer program. A downloadable product is a digital product that can be transmitted over a data network, i.e., downloaded by a user of the data network, and which can be offered for immediate download, for example, in an online shop. Furthermore, a computer can be equipped with the computer program, the machine-readable data carrier, or the downloadable product.
[0043] Further measures improving the invention are presented in more detail below together with the description of the preferred embodiments of the invention with reference to figures.
[0044] Examples of implementation
[0045] It shows:
[0046] Figure 1 embodiment of the method 100 for evaluating spatially resolved actual sensor data 2;
[0047] Figure 2 Example application of method 100 to a street scene.
[0048] Figure 1 is a schematic flow diagram of an embodiment of the method 100 for evaluating spatially resolved actual sensor data 2 with regard to the locations at which the scenery observed by a sensor 1 has a property 5 of interest.
[0049] In step 105, sensor data 2 recorded by a sensor 1 on a vehicle 50 or robot 60 can be selected.
[0050] In step 106, sensor data 2 can be selected that was recorded with a radar sensor, a lidar sensor, and / or a stereoscopic camera arrangement.
[0051] In step 110, a location la and an orientation lb of the sensor 1 are determined at the time of recording the sensor data 2. In step 120, a spatially resolved expectation 4 is retrieved from a spatially resolved map 3 based on the location la and the orientation lb of the sensor 1.
[0052] Furthermore, a spatially resolved property 5 of interest is stored in map 3. Property 5 is linked to expectation 4 in such a way that, assuming that the scene observed by sensor 1 exhibits property 5 from a certain location onward, the actual sensor data 2 should be consistent with expectation 4.
[0053] In step 130, it is checked to what extent the actual sensor data 2 are consistent with the expectation 4.
[0054] In step 140, at least with respect to those locations for which the actual sensor data 2 are consistent with the expectation 4, it is determined that the scenery observed by the sensor 1 has the property 5 stored in the map 3 in connection with the expectation 4. This is illustrated in Figure 2 using an example.
[0055] In step 150, a control signal 150a for the vehicle 50 or the robot 60 is determined based on the determination of the locations for which the scenery observed by the sensor 1 has the property 5 stored in the map 3 in connection with the expectation 4.
[0056] In step 160, the vehicle 50 or the robot 60 is controlled with this control signal 150a, so that the driving dynamics of the vehicle 50 or robot 60 are influenced in accordance with the control signal 150a.
[0057] According to block 111, a determined location la and / or a determined orientation lb can be optimized with the aim of maximizing the agreement of the actual sensor data 2 with the expectation 4.
[0058] According to block 121, the property 5 stored in the map 3 can, for example, include a statement as to the extent to which locations to which the expectation 4 refers can be freely accessed by the vehicle 50 or robot 60. According to block 122, the expectation 4 can, for example, include a spatially resolved three-dimensional geometry of the scenery observed by the sensor 1.
[0059] According to block 131, the actual sensor data 2 and the expectation 4 can be transferred into a common spatial reference system and / or a common workspace. According to block 132, the actual sensor data 2 can then be compared with the expectation 4 in this reference system or workspace.
[0060] According to block 133, to check whether sensor data 2 recorded with a stereoscopic camera arrangement is consistent with the three-dimensional geometry of expectation 4, an image provided by the first camera of this camera arrangement can be transformed, based on the geometry of expectation 4, into an expectation for the image provided by the second camera of the camera arrangement. According to block 134, it can then be checked to what extent this expectation is consistent with an image actually provided by the second camera of the camera arrangement.
[0061] This check, in turn, may include extracting features from the image provided by the second camera on the one hand and from the expectation for this image on the other hand, as per block 134a, and comparing these features with each other as per block 134b.
[0062] This comparison, in turn, may include, according to block 134c, making a binary decision for each feature as to whether a feature from the image is consistent with the corresponding feature from the expectation for that image, and, according to block 134d, determining a degree of agreement between the image and the expectation from the number of features that are consistent with each other.
[0063] According to block 135, it can additionally be checked to what extent a given test image, which does not show the scene observed by sensor 1, corresponds to the expected image provided by the second camera of the camera array. According to block 136, this degree of correspondence can be used as the noise level for the determined correspondence between the image provided by the second camera of the camera array and the expected image.
[0064] According to block 137, for actual sensor data 2 recorded with several different sensors 1, it can be checked separately for each location for which this actual sensor data 1 is consistent with the expectation 4 retrieved from the map 3. According to block 138, it can then be determined that the actual sensor data 2 as a whole is consistent with the expectation 4 only for those locations for which the actual sensor data 2 of all sensors 1 are consistent with the expectation.
[0065] According to block 141, actual sensor data 2, whose agreement with expectation 4 is checked, can be plausibilized against actual sensor data 2 recorded with another sensor 1. According to block 142, a match with expectation 4 can then be determined or maintained only with respect to those locations for which this plausibilization is positive.
[0066] Figure 2 shows an exemplary application of the method 100 to a street scene.
[0067] In this example, sensor 1 is carried by a vehicle (not shown). From a perspective determined by the location 1a and the orientation 1b of sensor 1, sensor 1 records actual sensor data 2 within its detection range 1c. In the example shown in Figure 2, the scene includes a road 10 with a vehicle 12 traveling ahead and a tree 11 at the roadside.
[0068] In the spatially resolved map 3, the road 10 and the tree 11 are also recorded, but the preceding vehicle 12 is missing. For the area of road 10, the interesting property 5 is stored, that this area is freely passable. The view and / or geometry of the scenery contained in map 3 is compared as expectation 4 with the actual sensor data 2. In this case, for the area of road 10, it is largely determined that the actual sensor data 2 are consistent with expectation 4 and, in connection with expectation 4, the property 5 is stored in map 3, that this is freely passable.
[0069] This area is therefore deemed to be open to traffic.
[0070] The only exception here is the area with the vehicle 12 in front. Since this is missing in map 3, the actual sensor data 2 here deviate from the
[0071] Expectation 4 is met. Accordingly, the area with the preceding vehicle 12 is not found to be freely passable.
Claims
Claims 1. Method (100) for evaluating spatially resolved actual sensor data (2) acquired with at least one sensor (1), comprising the steps: • a location (1a) and an orientation (1b) of the sensor (1) at the time of recording the sensor data (2) are determined (110); • from a spatially resolved map (3) a spatially resolved expectation (4) is retrieved based on the location (1a) and orientation (1b) of the sensor (1) (120); • It is checked (130) to what extent the actual sensor data (2) are consistent with the expectation (4); • at least with regard to those locations for which the actual sensor data (2) are consistent with the expectation (4), it is noted (140) that the scenery observed by the sensor (1) has a property (5) stored in the map (3) in connection with the expectation (4).
2. Method (100) according to claim 1, wherein sensor data (2) are selected (105) which were recorded by a sensor (1) on a vehicle (50) or robot (60).
3. Method (100) according to claim 2, wherein the property (5) stored in the map (3) comprises a statement (121) as to what extent places to which the expectation (4) relates can be freely traversed by the vehicle (50) or robot (60).
4. Method (100) according to one of claims 2 to 3, wherein • taking into account the determination for which locations the scenery observed by the sensor (1) corresponds to the expectation (4) in the card (3) has stored property (5), a control signal (150a) for the vehicle (50) or the robot (60) is determined (150) and • the vehicle (50) or the robot (60) is controlled by this control signal (150a) (160), so that the driving dynamics of the vehicle (50) or robot (60) are influenced according to the control signal (150a).
5. Method (100) according to any one of claims 1 to 4, wherein the actual sensor data (2) and the expectation (4) are transferred into a common spatial reference system, and / or into a common workspace (131) and wherein the actual sensor data (2) are compared with the expectation (4) in this reference system or workspace (132).
6. Method (100) according to any one of claims 1 to 5, wherein the expectation (4) • a spatially resolved three-dimensional geometry, and / or • a texturing, and / or • a reflectance amplitude, and / or • a multispectral response, and / or • includes a magnetic resonance of the scenery observed by the sensor (1) (122).
7. Method (100) according to any one of claims 1 to 6, wherein sensor data (2) are selected (106) which were recorded with at least one radar sensor and / or at least one lidar sensor, and / or at least one camera.
8. Method (100) according to claims 6 and 7, wherein the test to determine whether sensor data (2) recorded with a stereoscopic camera arrangement are consistent with the three-dimensional geometry of the expectation (4) comprises, • to transform an image supplied by the first camera of the camera arrangement into an expectation for the image supplied by the second camera of the camera arrangement using the geometry of the expectation (4) (133) and - 18 - • to check (134) to what extent this expectation is consistent with an image actually supplied by the second camera of the camera arrangement. Method (100) according to claim 8, wherein • features are extracted from the image supplied by the second camera on the one hand and from the expectation for this image on the other hand (134a) and • these features are compared with each other (134b). Method (100) according to claim 9, wherein • for each feature, a binary decision is made (134c) as to whether a feature from the image is consistent with the corresponding feature from the expectation for that image; and • a degree of correspondence between the image and the expectation is determined from the number of features that are consistent with each other (134d). Method according to any one of claims 8 to 10, wherein • Additionally, it is checked (135) to what extent a given test image, which does not show the scene observed by sensor (1), is consistent with the expectation for the image supplied by the second camera of the camera arrangement, and • this degree of agreement is used as the noise level for the determined agreement between the image supplied by the second camera of the camera arrangement and the expectation for this image (136). Method (100) according to any one of claims 1 to 11, wherein • for actual sensor data (2) recorded with several different sensors (1), it is checked separately (137) for which locations these actual sensor data (1) are consistent with the expectation (4) retrieved from the map (3), and • determined only for those locations where the actual sensor data (2) of all sensors (1) are consistent with the expectation - 19 - will (138) that the actual sensor data (2) are in agreement with the expectation (4) overall.
13. Method (100) according to any one of claims 1 to 12, wherein • Actual sensor data (2), whose agreement with the expectation (4) is checked, are checked against actual sensor data (2) that were recorded with another sensor (1), and are made plausible (141) • agreement with the expectation (4) is only established or maintained with respect to those places (142) for which this plausibility check is successful.
14. Method (100) according to any one of claims 1 to 13, wherein a determined location (1a), and / or a determined orientation (1b), are optimized (111) with the aim of maximizing the agreement of the actual sensor data (2) with the expectation (4).
15. Computer program containing machine-readable instructions which, when executed on one or more computers, cause the computer(s) to execute the method (100) according to any one of claims 1 to 14.
16. Machine-readable data carrier and / or download product containing the computer program according to claim 15.
17. One or more computers with the computer program according to claim 15, and / or with the machine-readable data carrier and / or download product according to claim 16.
