Method for determining a sensor degradation state

Abstract

A method for determining a sensor degradation state of a first sensor system is proposed, the first sensor system being set up to provide data for representing an environment of the first sensor system, the method having the steps: providing data of the first sensor system for representing the environment; providing data of a second sensor system for representing the environment; determining an individual blindness indicator of the first sensor system based on sensor data of the first sensor system only; determining at least one environment-related decision factor based on the provided data of the first sensor system; determining at least one environment-related decision factor based on the provided data of the second sensor system; determining a fusion blindness indicator by means of a comparison of the at least one environment-related decision factor based on the provided data of the first sensor system and the at least one environment-related decision factor based on the provided data of the second sensor system; determining the sensor degradation state of the first sensor system by means of the individual blindness indicator of the first sensor system and the fusion blindness indicator.

Description

Background Technology

[0001] Driving automation is closely linked to equipping vehicles with increasingly large and powerful sensor systems for environmental monitoring. In some cases, vehicle sensors redundantly cover 360° of the environment and different effective ranges through multiple sensors and modalities. For example, video sensors, radar sensors, lidar sensors, ultrasonic sensors, and microphone sensors are used as modalities.

[0002] Sensor data is combined to form a safe environmental model to represent the vehicle's environment. On the other hand, the requirements for the scope and quality of this environmental model depend on the driving functions implemented upon it. In autonomous vehicles, this environmental model is used to make comprehensive driving decisions and correspondingly control actuators.

[0003] Each of these sensors may be limited or “degraded” relative to its nominal performance due to dirt, signs of wear, damage, or other environmental factors such as rain, snow, and fog. The impact on sensor data and its characteristics are particularly dependent on the sensor mode. Summary of the Invention

[0004] Data from the sensor system is processed by different modules in various processing steps to represent the vehicle's environment. This data is further abstracted at each processing step and ultimately combined into a safe environmental model. Here, commonly used algorithms for different sensor modalities, such as object detection, object classification, object tracking, and distance calculation, are susceptible to degradation of the input data. In these cases, typical object detection and classification methods fail because they cannot identify false positives and false negatives arising from the degradation of the data.

[0005] In almost all cases, false negative degenerate identification is an unacceptable security risk because data from “blind” sensors, in the form of unidentified or misidentified objects, can unknowingly lead to incorrect security-related decisions.

[0006] On the other hand, false-positive degradation identification “only” reduces the system’s usability because system degradation derived from sensor degradation is not necessarily occurring. Since the responsibility can be returned to the driver in automation levels 1-3 (with a driver), false positives are more acceptable compared to levels 4-5 (without a driver), where excessive degradation limits the vehicle’s driving capabilities (until it comes to a stop), which is unacceptable for the application.

[0007] In particular, the false positive rate of degradation recognition is a challenge in the path to highly autonomous driving, which necessitates improvements in degradation recognition of sensor systems.

[0008] Errors in environmental detection by sensor systems can lead to incorrect decisions in higher-level systems. This can result in safety risks or decreased comfort, depending on the functionality being implemented. Sensor systems in vehicles can be equipped with separate blindness detection. Through a mechanism for identifying sensor blindness, functions implemented based on sensor data throughout the system can be marked as degraded according to the current sensor availability.

[0009] In other words, the sensor system generates data in order to enable system functionality. If the availability of the sensor system is known, for example through reliable blindness detection, then the availability of system functionality can be derived from this. That is, the system can diagnose or identify corresponding system function degradation simply by accurately identifying sensor degradation. Therefore, for example, if only part of the surrounding environment is recorded, due to partial blindness of the system, the corresponding functionality can only be provided in a relatively limited manner.

[0010] According to aspects of the invention, a method for determining the sensor degradation state of a sensor system, a method for providing control signals, an evaluation device, a computer program product, and a machine-readable storage medium, as characterized in the independent claims, are provided. Advantageous designs are the subject of the dependent claims and the description that follows.

[0011] In this general description of the invention, the order of the method steps is presented to make the method easy to understand. However, those skilled in the art will recognize that multiple method steps in these steps can also be traversed in another order and yield the same or corresponding results. In this sense, the order of these method steps can be changed accordingly. Some features are equipped with numbers to improve readability or make the assignment more explicit, but this does not imply the existence of specific features.

[0012] According to one aspect of the invention, a method for determining the sensor degradation state of a first sensor system is provided, wherein the sensor system is configured to provide data representing the environment of the first sensor. The method comprises the following steps:

[0013] In one step of the method, data from a first sensor system is provided to represent the environment. In another step, data from a second sensor system is provided to represent the environment. In yet another step, an individual blindness index for the first sensor system is determined, based solely on sensor data from the first sensor system. In yet another step, at least one environment-related determinant is determined, based on data provided by the first sensor system. In yet another step, at least one environment-related determinant is determined based on data provided by the second sensor system. In yet another step, a fused blindness index is determined by comparing at least one environment-related determinant based on data provided by the first sensor system with at least one environment-related determinant based on data provided by the second sensor system. In yet another step, the sensor degradation state of the first sensor system is determined using the individual blindness index of the first sensor system and the fused blindness index.

[0014] Alternatively or additionally, a fusion blindness index of the first sensor system can be determined by comparing at least one environment-related determinant based on data provided by the first sensor system with at least one environment-related determinant based on data provided by the second sensor system. Then, in another step, the sensor degradation state of the first sensor system can be determined correspondingly by using both the individual blindness index of the first sensor system and the fusion blindness index of the first sensor system.

[0015] The methods used to determine sensor degradation state can be employed to determine the current sensor degradation state of a given sensor system. Here, sensor degradation state characterizes all performance-related damage that a sensor system may experience during its operation, such as due to rain, snow, scratches, fog, glare, low sunlight, leaves, and other contaminants. Since not every sensor mode is equally affected by the aforementioned causes, these damages may not necessarily be identified as damage for every sensor mode.

[0016] Here, the various method steps for determining the sensor degradation state can be determined by means of a system distributed across one or more control devices. Since the representation of the environment is typically determined using data from a sensor system, which traverses different processing levels until that representation is determined, this method for determining the sensor degradation state advantageously utilizes data and methods existing at the corresponding processing levels to determine the sensor degradation state. Here, the corresponding sensor degradation state is not yet determined at the sensor level, but rather existing information is carried up to the last level based on indicators, so as to advantageously add relevant information content for reliably and consistently determining the sensor degradation state. In particular, to determine the sensor degradation state, a fusion layer method can be included, in which data from different sensors, which may also have different modalities, are compared with each other, so that the determination of the degradation state of the corresponding sensor is no longer based solely on the data provided by the sensor itself. This determination of the sensor degradation state using this method can be distributed across multiple control devices and multiple electronic subunits of these control devices.

[0017] Reliable identification of degradation, especially due to a low false positive rate, ensures high availability of sensors and systems without compromising vehicle safety, i.e., maintaining a low false negative rate.

[0018] Here, the method can be applied to multiple sensors and sensor modes to determine degradation. By using methods that are used at different levels to determine the representation of the environment at different data processing levels, the corresponding metrics can be integrated into existing data processing in a resource-efficient manner for the representation of the environment.

[0019] By including a fusion layer, the strengths and weaknesses of different sensor modalities can be balanced. Furthermore, including a fusion layer allows for redundant verification of results from multiple sensors.

[0020] Especially for the higher requirements of vehicle automation levels 4 / 5, it is necessary to include the integration layer.

[0021] By integrating into the processing architecture that determines the representation of the environment, the determination of sensor degradation states can be combined and extended. Identified weaknesses can be targeted by integrating additional identification mechanisms without violating architectural principles.

[0022] If, especially when using neural networks, data from sensor systems is fully classified, then degradation categories can be additionally defined and integrated into the classification in a resource-efficient manner. Here, this classification can be integrated differently, either by overlaying or by substitution, depending on whether information from the environmental model should also be considered when identifying degradation. A prototype of this architecture is pixel-by-pixel semantic segmentation of camera images. The category "blind" can be introduced along with categories such as "empty land," "curbstone," and "person."

[0023] Another example could be based on factoring in the erroneous associations of objects when fusing data from different sensor systems. If the processing steps in such fusion are performed by comparing or associating the current environmental model with sensor knowledge, then fusion degradation metrics can be determined here with minimal additional cost by factoring in the conflict calculations.

[0024] This method can be integrated into a processing architecture that operates distributed across multiple control devices (ECUs) while still ensuring that the corresponding sensor degradation state is determined while incorporating all relevant information sources. For example, complex preprocessing steps, such as determining indicators on the control device (ECU) and in memory using neural networks on images or point clouds, can be performed where the relevant data is locally available. This method allows for a high level of coverage of degradation causes and, if necessary, the best possible differentiation, because the final degradation decision is made only centrally while incorporating all relevant information.

[0025] In other words, the method for determining sensor degradation status can be integrated into a signal processing chain for environmental perception, for example, in the field of automated driving functions. Here, the sensor system for environmental perception can have one or more sensors from one or more sensor modalities. Typically, especially when using multiple sensors, environmental perception is determined in multiple processing steps and on multiple control devices (ECUs). In each of these processing steps, potentially useful indicators for degradation identification can be obtained. Furthermore, the data can exist in a suitable form between every two processing steps so that this data can be examined specifically regarding degradation. Here, it is not necessary to provide degradation indications in every processing step and on all forms of data.

[0026] The characteristic of a degradation index is that it condenses degradation information, responding to degradation scenarios or causes, and serves as a useful source of information for determining the sensor's degradation state. The degradation index can be transmitted to a central evaluation device, where the sensor's degradation state is determined.

[0027] With the help of this assessment device, other information that is important in subsequent systems can also be extracted and provided, such as the confidence level for determining the sensor degradation state, the contribution to the weather-environment model, and the requirements for the maintenance or cleaning process of the sensor system.

[0028] Here, sensor degradation states can be constructed at different granularities as needed, such as by corresponding solid angles and / or by corresponding effective range areas and / or by image regions and / or by specific use cases for at least partially automated vehicles, such as lane changing, following, or traffic light recognition. For example, ground recognition may still be effective, while pedestrian recognition may no longer be feasible.

[0029] According to one aspect, it is proposed that the fused blindness index be determined by means of a trained neural network or support vector machine and by utilizing at least one environment-related determinant based on data provided by a first sensor system and at least one environment-related determinant based on data provided by a second sensor system.

[0030] According to one aspect, it is proposed that: the individual blindness index of the first sensor system has a sensor blindness index and / or a perceptual blindness index, wherein the sensor blindness index is determined by means of sensor data provided solely by the first sensor system, and the perceptual blindness index is determined by means of sensor data provided solely by the first sensor system based on a method for determining environmentally relevant determinants, and the sensor degradation state of the first sensor system is correspondingly determined by means of the sensor blindness index and / or perceptual blindness index of the first sensor system and the fused blindness index of the first sensor system.

[0031] Here, the method for determining environment-related determinants evaluates data from a first sensor system to represent the environment of the first sensor system with respect to the measurement purpose. For example, image segmentation or the stixel or L-shaped pixels of a LiDAR system are evaluated with respect to the measurement purpose of object detection, such as identifying the object category of a car, measuring the object category of a car, and determining the location of the object category of a car.

[0032] According to one aspect, it is proposed that: the first sensor system is configured to provide sensor-specific degradation indicators; and the sensor degradation state of the first sensor system is additionally determined by means of the sensor-specific degradation indicators provided by the first sensor system.

[0033] These sensor-specific degradation metrics can typically be provided by each sensor system as a self-diagnostic measure, without comparison with other sensor systems. Below, examples of different self-diagnostics for different sensor modalities are presented, based on different strategies to determine sensor degradation:

[0034] - For ultrasonic sensors, blindness detection can reliably identify contact dirt, for example, by measuring the inherent frequency on the sensor membrane. This can be further extended by combining it with information provided by other sensor systems.

[0035] For radar sensors, degradation metrics can be derived from the signal processing layer of the sensor itself. These degradation metrics can be filtered out over time in subsequent processing layers to suppress random errors.

[0036] For video sensors, degradation status can be determined using image content, such as by classifying the image content or evaluating optical flow. If degradation status cannot be determined using image content, the methods described herein can improve the determination of degradation status for such sensors.

[0037] Here, the degradation state can be derived by means of optical flow because some degradation of video sensors simply does not allow the optical flow to be determined (optical flow collapse), or subsequently inconsistent optical flow is determined, or, for example, the length of the optical flow is determined to be zero because the image content is static.

[0038] - Data from lidar sensors can be used to detect, differentiate, and quantify atmospheric phenomena, and this information can then be combined with other sensors.

[0039] In one aspect, it is proposed that for at least one of these degradation indices, an index confidence level is additionally calculated. For different degradation indices of the corresponding sensor system and / or combinations of different sensor systems, this index confidence level enables conflicting degradation indices and / or conflicting representations and / or environmentally relevant determinants of different sensor systems to be weighted accordingly in determining the sensor degradation state.

[0040] Such a confidence level can, for example, have values ​​from the range [0, 1], so that weighting of different blindness indicators related to the confidence level can be implemented in the detection step, the weighting being based on the quality or caliber by which these blindness indicators are determined.

[0041] In an additional step of the method, the corresponding blindness index and / or index confidence can be filtered over time, for example, especially by means of exponential filtering, in order to become robust to temporary outliers.

[0042] According to one aspect, it is proposed that: the first sensor system and the second sensor system have the same sensor mode; and / or the first sensor system and the second sensor system have different sensor modes.

[0043] Using two sensor systems with the same modality allows for easy identification of defects in terms of degradation of a single sensor system.

[0044] Sensor systems with different modalities can determine sensor degradation states with higher reliability, especially when cross-comparisons between sensor systems with different modalities are performed using model-based methods.

[0045] In one aspect, it is proposed that the comparison used to determine the fusion blindness index of the first sensor system is based on object and / or model.

[0046] To determine a representation of the environment using at least two sensor systems, a fused object with object probabilities is typically formed. Object-based comparisons used to determine fusion blindness metrics are based on a comparison of objects identified by at least two sensor systems, which may in particular have different modalities. Here, in various aspects, objects redundantly identified by at least two different sensor systems can be compared based on verification to determine fusion blindness metrics. In other words, object recognition from at least two different sensors is compared here.

[0047] - If, for example, an object has been identified by at least two sensors and the third sensor fails to recognize it even though the object should be within its field of view, this indicates sensor degradation or decalibration of the third sensor. This decision can be confirmed through statistical or other indicators.

[0048] - If at least two sensors conflict in their identification, this indicates that at least one of these sensors is faulty. Statistical analysis of multiple observations or multiple observation periods in combination, such as whether a particular sensor is always involved in the problem, or by combining with other indicators, such as sensor-based degradation indicators that apply to one of these sensors, can lead to degradation decisions.

[0049] In this object-based comparison, the sensors involved may have different sensor modes.

[0050] Furthermore, this object-based comparison can be performed such that degradation metrics are determined individually based on the angular and / or distance ranges from the perspective of the respective sensor system. For example, for lidar sensors, problems in object identification often occur within the "lower right" angular range or starting from a certain distance, which may cause degradation metrics that characterize localized fouling or loss of effective range, respectively.

[0051] In particular, these degradation metrics can have multiple sub-metrics or higher dimensions to describe the corresponding degradation of the sensor system.

[0052] Model-based comparisons used to determine fusion degradation metrics compare sensor data and / or the characteristics of the identified objects and / or the identified objects using model knowledge about the degradation phenomenon.

[0053] For example, when identifying weather phenomena, knowledge from one sensor system can be transferred to another, or the knowledge can be cross-validated or subjected to a plausibility check. It can also be considered that these sensor systems may degrade to varying degrees due to factors such as wet sensor surfaces during rain, depending on their current direction of travel and / or installation location. Furthermore, it can be considered that different sensor modes are inherently affected differently by the corresponding weather phenomena in terms of effective range, angular error, etc.

[0054] For example, fog can be identified using a lidar system. If the lidar system uses radiation close to the wavelength of visible light, such as 900 nm, the measured atmospheric absorptivity can be transmitted to, for example, a camera system at a substantially 1:1 ratio.

[0055] For example, model-based comparisons may include: depending on the weather phenomenon, some sensors are advantageous in identifying and measuring the phenomenon when necessary due to their installation location. Other sensors may still degrade due to the same weather phenomenon. In this case, knowledge from sensors advantageously installed regarding the weather phenomenon can be transferred to all affected sensors. Thus, for example, daytime fog can be well identified by a road-oriented camera system through changes in brightness along the road, and fog density can also be measured using such a camera system. Data from side-facing sensors, which are also affected by fog, can be correspondingly evaluated using knowledge from front-facing sensors to determine fusion blindness indices.

[0056] The same applies to rain detection, because a side-facing camera system may have more difficulty identifying rain than a front-facing sensor system, which can clearly show rain.

[0057] According to one aspect, it is proposed that in object-based comparisons, the at least one environment-related determinant is the object of the environment of the first sensor system and the second sensor system.

[0058] In one aspect, it is proposed that: regarding the model consideration of the degradation process of the first sensor system and the second sensor system, a model-based comparison is performed on at least one environmentally relevant determinant based on data from the first sensor system and at least one environmentally relevant determinant based on data from the second sensor system.

[0059] In one aspect, it is proposed to compare the different effects of environmental determinants determined by a first sensor system in a first mode and a second sensor system in a second mode on the degradation process acting on the two sensor systems.

[0060] Based on one aspect, it is proposed that the first mode is different from the second mode.

[0061] According to one aspect, it is proposed that: sensor-specific degradation indicators are determined by an evaluation device for a first sensor system; and / or perceptual blindness indicators are determined by a perception module of the evaluation device; and / or fusion blindness indicators are determined by a fusion module of the evaluation device; and / or degradation values ​​of the first sensor system are determined by a detection module of the evaluation device.

[0062] Here, these different modules can be configured to use this clock rate to determine the corresponding blindness index, so that the sensor degradation state can be determined for each provided data block in order to avoid delays caused by the architecture.

[0063] A method is proposed in which control signals for operating at least a partially automated vehicle are provided based on sensor degradation states of a first sensor system determined according to one of the methods described above; and / or alarm signals for alerting vehicle occupants are provided based on sensor degradation states of the first sensor system.

[0064] The term "based on" should be understood broadly regarding the characteristic of "providing a control signal based on the sensor degradation state of the first sensor system." This term should be understood as meaning that the sensor degradation state of the first sensor system is used in any determination or calculation of the control signal, which does not exclude the use of other input parameters for such determination of the control signal. This correspondingly applies to the provision of alarm signals.

[0065] Depending on the degree of degradation of the sensor system, this control signal can be used to respond differently to sensor degradation. Thus, for example, when a hold or cleaning function is present, this control signal can be used to activate cleaning of varying intensities, such as using nozzles and / or windshield wipers on the sensor surface, to maintain system availability. Furthermore, highly automated systems may use this control signal to initiate a transition to a safe state. For example, for vehicles with at least partially automated driving, this could result in a slow stop on the shoulder.

[0066] An evaluation device is proposed, which is configured to perform one of the above methods to determine the sensor degradation state.

[0067] According to one aspect, a computer program is described that includes instructions, which, when executed by a computer, cause the computer to perform one of the methods described above. This computer program enables the use of the described methods in various systems.

[0068] A machine-readable storage medium is described, on which the aforementioned computer program is stored. This machine-readable storage medium allows for the transfer of the aforementioned computer program. Attached Figure Description

[0069] refer to Figure 1 Embodiments of the invention are shown and described in more detail below. Wherein:

[0070] Figure 1 An evaluation device with multiple sensors is shown. Detailed Implementation

[0071] Figure 1 An assessment device with multiple sensors 110 to 150 is schematically depicted, which provide corresponding individual blindness indicators 110b to 150b in addition to their sensor signals 110a to 150a.

[0072] Here, sensor 110 depicts a video system, sensor 120 depicts a radar system, sensor 130 depicts a lidar system, sensor 140 depicts an ultrasonic system, and sensor 150 depicts a sound conversion system.

[0073] Sensors 110 to 150 are configured to determine sensor-specific degradation indices 110b to 150b using sensor data provided solely by the respective sensor systems.

[0074] Sensor signals 110a to 150a used to represent the environment, together with individual blindness indicators 110b to 150b, are provided to the perception module 210 in order to determine the perceived blindness indicators.

[0075] The corresponding sensors 110-150 are configured using a corresponding evaluation device to calculate degradation indices using not only existing internal sensor information but also raw sensor data. For radar systems, this could be a calculation of signal quality; or for ultrasonic systems, it could be a measurement of the membrane's natural frequency. However, existing signals can also be used as degradation indices, such as optical flow for video sensors.

[0076] The sensing module 210 determines environmentally relevant determinants representing the environment using the provided sensor signals 110a to 150a. Additionally, for sensor systems 110 to 150, the sensing module 210 determines at least one perceptual blindness index of the respective sensor system based on a corresponding method for determining environmentally relevant determinants, using only the corresponding sensor data provided by the respective sensor system.

[0077] exist Figure 1 The image shows a perception blindness index 210a for the first sensor system and a perception blindness index 210b for the second sensor system among these sensor systems 110 to 150.

[0078] The perception module 210 uses the provided data and information to determine the corresponding degradation indices for the respective sensor systems 110-150. Here, the environmental determinants can be more abstract than the pure sensor data from sensor systems 110-150, and the determination of the perceptual blindness indices is based on the determination of these environmental determinants. For example, these environmental determinants can include objects, features, stixels, the size of the specific object, object type, three-dimensional "bounding box," object category, such as L-shapes and / or edges and / or reflection points in a lidar system, and several other determinants. Furthermore, the perception module can track objects, i.e., track the position of these objects over time. For radar systems, for example, multiple tracked objects can be used to determine the perceptual blindness indices. For video systems, for example, the output values ​​of classifications determined by means of a neural network can be used to determine the perceptual blindness indices.

[0079] The corresponding sensor signals 110a to 150a of the corresponding sensor systems 110-150 for representing the environment, the individual blindness indicators 110b to 150b of the corresponding sensor systems, and the perceived blindness indicators of the corresponding sensor systems 110 to 150 are provided to the fusion module 310 in order to determine the fused blindness indicators of the corresponding sensor systems 110 to 150.

[0080] Here, by comparing at least one environment-related determinant 320a based on data provided by the first sensor system among these sensor systems 110 to 150 with at least one environment-related determinant 320b based on data provided by the second sensor system among these sensor systems 110 to 150, the corresponding fusion blindness index 330 of the respective sensor system is determined.

[0081] These environment-related determinants are highly abstracted in the fusion module 310, enabling the formation of fused objects with object probabilities, for example, from objects or columnar pixels. Furthermore, environment maps can be generated and / or updated, and / or these environment maps can be compared with map information from a database. Additionally, abstract information from all sensor systems is present.

[0082] Individual blindness indices 110b to 150b of the corresponding sensor systems, perceptual blindness indices of the corresponding sensor systems 110 to 150, and fusion blindness indices of the corresponding sensor systems 110 to 150 are provided to the evaluation device 410, which uses these provided values ​​to determine the degradation values ​​of the corresponding sensor systems 110 to 150.

Claims

1. A method for determining the sensor degradation state of a first sensor system (110, 120, 130, 140, 150), the first sensor system being configured to provide data representing the environment of the first sensor system, the method comprising the following steps: Provide data (110a, 120a, 130a, 140a, 150a) from the first sensor system (110, 120, 130, 140, 150) to represent the environment; Data (110a, 120a, 130a, 140a, 150a) from the second sensor system (110, 120, 130, 140, 150a) are provided to represent the environment; Based solely on sensor data from the first sensor system (110a, 120a, 130a, 140a, 150a), individual blindness indicators (110b, 120b, 130b, 140b, 150b, 210a) for the first sensor system (110, 120, 130, 140, 150) are determined. Based on the data (110a, 120a, 130a, 140a, 150a) provided by the first sensor system (110, 120, 130, 140a, 150a), at least one environmentally relevant determinant (320a) is determined. Based on the data provided by the second sensor system (110, 120, 130, 140, 150), at least one environmentally relevant determinant (320b) is determined. By comparing the at least one environment-related determinant (320a) based on the data provided by the first sensor system with the at least one environment-related determinant (320b) based on the data provided by the second sensor system (110, 120, 130, 140, 150), a fusion blindness index (330) is determined. Using the individual blindness indicators (110b, 120b, 130b, 140b, 150b, 210a) of the first sensor system and the fused blindness indicator (330), the sensor degradation state of the first sensor system (110, 120, 130, 140, 150) is determined. The comparison used to determine the fusion blindness index (330) of the first sensor systems (110, 120, 130, 140, 150) is based on object and / or model. In the object-based comparison, at least one environment-related determinant (320a, 320b) is an object of the environment of the first sensor system (110, 120, 130, 140, 150) and the second sensor system (110, 120, 130, 140, 150).

2. The method according to claim 1, wherein the individual blindness indices (110b, 120b, 130b, 140b, 150b, 210a) of the first sensor system have sensor blindness indices and / or perceptual blindness indices (210a), wherein the sensor blindness indices (110b, 120b, 130b, 140b, 150b) are determined by means of sensor data provided solely by the first sensor system (110, 120, 130, 140, 150), and the perceptual blindness indices (210a) are based on a method for determining environmentally relevant determinants by means of the first sensor system. The sensor data (110a, 120a, 130a, 140a, 150a) provided by the systems (110, 120, 130, 140, 150a) are determined accordingly, and the sensor degradation state of the first sensor system is determined accordingly by means of the sensor blindness index (110b, 120b, 130b, 140b, 150b) and / or the perception blindness index (210a) of the first sensor system (110, 120, 130, 140, 150) and the fusion blindness index (330) of the first sensor system (110, 120, 130, 140, 150).

3. The method according to claim 1 or 2, wherein the first sensor system (110, 120, 130, 140, 150) and the second sensor system (110, 120, 130, 140, 150) have the same sensor mode; or the first sensor system and the second sensor system have different sensor modes.

4. The method of claim 3, wherein, considering the model of the degradation process of the first sensor system (110, 120, 130, 140, 150) and the second sensor system (110, 120, 130, 140, 150), a model-based comparison is performed between at least one environmentally relevant determinant based on data from the first sensor system (110, 120, 130, 140, 150) and at least one environmentally relevant determinant based on data from the second sensor system (110, 120, 130, 140, 150).

5. The method of claim 4, wherein the environmental determinants determined by the first sensor system (110, 120, 130, 140, 150) of the first mode and the second sensor system (110, 120, 130, 140, 150) of the second mode are compared with respect to the different effects of the degradation processes acting on the two sensor systems.

6. The method of claim 5, wherein the first mode is different from the second mode.

7. A method for providing control signals for operating at least a partially automated vehicle and / or providing alarm signals for alerting vehicle occupants, wherein the method provides control signals for operating at least a partially automated vehicle based on a sensor degradation state of a first sensor system (110, 120, 130, 140, 150) as determined by any one of claims 1 to 6; and / or provides alarm signals for alerting vehicle occupants based on said sensor degradation state of the first sensor system (110, 120, 130, 140, 150).

8. An evaluation device configured to perform the method according to any one of claims 1 to 6 to determine the sensor degradation state.

9. A computer program product comprising a computer program, the computer program including instructions that, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 7.

10. A machine-readable storage medium having a computer program stored thereon, the computer program including instructions that, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 7.

Images