Method and system for reliably detecting objects in the surroundings of a motor vehicle - Patents.com

Abstract

The present invention relates to a method for reliably detecting objects in the surroundings of a motor vehicle, the motor vehicle having a plurality of different surrounding sensors, the method (1) comprising the following steps: (2) detecting sensor data by at least one of the plurality of different surrounding sensors; (3) detecting, for each of the at least one of the plurality of different surrounding sensors, an object in the corresponding detected sensor data; (4) identifying at least one distinctive feature for object detection based on the detected object; and (5) checking, for each of the at least one distinctive feature, a corresponding distinctive feature for reliably detecting objects in the surroundings of the motor vehicle.

Description

[Technical Field]

[0001] The present invention relates to a method for reliably detecting objects in the vicinity of a motor vehicle, and in particular to a method for detecting objects in the vicinity of a motor vehicle, which can ensure a high identification rate of objects in the vicinity of the motor vehicle while reducing the number of incorrectly identified objects. [Background technology]

[0002] Comprehensive identification of the vehicle's environment and, in particular, objects, such as other road users, in the vehicle's environment provides the basis for many driver assistance systems and automated driving functions of the vehicle. In this case, the vehicle typically has a plurality of sensor systems, each tasked with identifying objects in its respective detection area. Each sensor system in this case includes one surroundings sensor and a processing unit, which is configured to analyze data provided by the surroundings sensors and, for example, detect objects in the data. In this case, the surroundings sensors may be, for example, video cameras, radar, LIDAR, or ultrasonic sensors.

[0003] In this case, each individual one of these ambient sensors is very limited in terms of its respective field of use and may not provide all the required information about the vehicle's surroundings, depending on, for example, the installation location or the current weather conditions. In this case, by combining or fusing the information from different ambient sensors, it is possible to increase the reliability when identifying the vehicle environment and thus when detecting objects around the vehicle, thereby again avoiding corresponding safety-critical situations when controlling driver assistance systems or driving functions based on the detected objects.

[0004] In known approaches, in this case, objects are detected separately in the sensor data detected by different surrounding sensors, and then each detection result is fused by a fusion approach.However, especially in complex scenarios where visibility is poor, it is difficult to ensure a high identification rate of the objects around the automobile and at the same time to suppress the number of incorrectly identified objects to a low level.Therefore, there is a need for an improved method for reliably detecting objects around the automobile.

[0005] WO 2013 / 056966 discloses a method for checking the validity of a sensor signal and a method and a device for outputting a trigger signal, wherein a first sensor element detects at least one first physical quantity and outputs it as a first sensor signal, and a second sensor element detects a second physical quantity correlated to the first physical quantity and outputs it as a second sensor signal, and wherein at least the first sensor element has at least one first certainty region with an upper and / or lower limit, which first certainty region is associated with the second physical quantity detected by the second sensor element, and the actual value of the first physical quantity detected by the first sensor element is identified as valid if the actual value of the second physical quantity detected by the second sensor element is within the corresponding first certainty region of the first sensor element. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] International Publication No. 2013 / 056966 Summary of the Invention [Problem to be solved by the invention]

[0007] The problem underlying the present invention is therefore to provide an improved method for reliably detecting objects in the surroundings of a motor vehicle. [Means for solving the problem]

[0008] The above problem is solved by a method for reliably detecting objects in the surroundings of a motor vehicle according to the characterizing part of claim 1.

[0009] The above problem is further solved by a system for reliably detecting objects in the surroundings of a motor vehicle according to the characterizing part of claim 8.

[0010] Disclosure of the Invention According to one embodiment of the present invention, the above problem is solved by a method for reliably detecting objects in the surroundings of a motor vehicle, the motor vehicle having a plurality of different surrounding sensors, the method comprising the following steps: detecting sensor data by at least one of the plurality of different surrounding sensors; detecting, for each of the at least one of the plurality of different surrounding sensors, an object in the corresponding detected sensor data; identifying at least one distinctive feature for object detection based on the detected object; and for each of the at least one distinctive feature, checking a corresponding distinctive feature for reliably detecting objects in the surroundings of the motor vehicle.

[0011] A sensor, also called a detector, measurand sensor or measurement sensor or (measurement) sensor, is understood to be a technical component that is able to detect certain physical or chemical properties and / or material characteristics in its surroundings either qualitatively or quantitatively as a measurand.

[0012] Furthermore, an ambient sensor is understood to be a sensor for detecting conditions that exist in the surroundings, for example around a motor vehicle.

[0013] Furthermore, a plurality of different ambient sensors is understood to mean a plurality of different types of ambient sensors or ambient sensors based on a plurality of different techniques, for example video cameras, radar, LIDAR or ultrasonic sensors.

[0014] Furthermore, detecting an object in the sensor data means that the object or thing represented in the object data is detected by object detection or by an algorithm for object detection, regardless of its type; the object may be, for example, a pedestrian, a further vehicle, a traffic guidance system, or a further object in the surroundings of the vehicle.

[0015] Furthermore, a distinctive feature is understood to be an object detection result that may indicate an object but that should or should be checked again, in particular to avoid a corresponding safety-critical situation when controlling a driver assistance system or driving function accordingly. For example, such a distinctive feature may be present when an object is detected in sensor data detected by one surroundings sensor, but this same object is not detected in other sensor data detected, for example, by another different surroundings sensor.

[0016] Furthermore, in this case, checking a distinctive feature means verifying the object underlying the corresponding distinctive feature, i.e., checking whether the distinctive feature really represents the object.

[0017] This allows for increased reliability or accuracy in detecting objects around the vehicle, thereby again avoiding corresponding safety-critical situations when controlling driver assistance systems or driving functions based on the detected objects. In particular, in this case, the discrimination ability or performance of the corresponding object detector can be increased, thereby again improving the performance of environmental perception. This allows for a high discrimination rate to be guaranteed while simultaneously reducing the number of incorrectly identified objects. Furthermore, it allows for reliable identification of blind areas of individual object detectors, i.e., situations in which it is difficult for the individual object detectors to detect objects.

[0018] Overall, therefore, an improved method for reliably detecting objects in the surroundings of a vehicle is presented.

[0019] In this case, the method may further include, for each of at least one of the plurality of different ambient sensors, pre-processing corresponding detected sensor data, and wherein, for each of at least one of the plurality of different ambient sensors, respectively detecting an object in the corresponding detected sensor data includes detecting the object in the corresponding pre-processed sensor data.

[0020] In this case, pre-processing of the sensor data means that the sensor data is prepared in a targeted manner for object detection, for example insufficiencies of the corresponding sensor can be compensated for.

[0021] This allows for even greater reliability in detecting objects in the vehicle's surroundings, and in this case the corresponding data can be optimally adapted to or prepared for object detection.

[0022] Further, the step of identifying at least one distinctive feature in object detection based on the detected object may include identifying the at least one distinctive feature based on the detected object and further information describing the situation.

[0023] In this case, information describing a situation is understood to be data representative of the corresponding situation, and the information may include, for example, context information or information regarding the values ​​of one or more vehicle parameters present in the corresponding situation.

[0024] This allows for even greater reliability in detecting objects around the vehicle by taking into account information about the current situation, e.g. current weather conditions, current light conditions and / or current vehicle speed.

[0025] Furthermore, the step of identifying at least one distinctive feature for object detection based on the detected object may also include applying a machine learning algorithm trained to identify distinctive features for object detection based on information about the detected object.

[0026] Machine learning algorithms are based on the use of statistical techniques to train data processing systems to perform specific tasks without the system being explicitly programmed for these tasks. The goal of machine learning is to build algorithms that can learn from data and make predictions. Such algorithms create mathematical models that can, for example, classify data.

[0027] This allows for even further optimization of the identification of distinctive features and even further improvement in the reliability or accuracy with which distinctive features are identified.

[0028] For each of the at least one distinctive feature, the step of checking a corresponding distinctive feature may further include applying an event camera and / or refining an object detection underlying the corresponding distinctive feature and checking the corresponding distinctive feature based on the refined object detection and / or detecting an object in sensor data detected by other ambient sensors of the plurality of different ambient sensors to check the corresponding distinctive feature.

[0029] Furthermore, an event camera is understood to be an imaging sensor that reacts to local brightness changes. Instead of supplying a sequence of images at a constant frequency, the event camera transmits only information from pixels that have a significant change in brightness. This allows for better detection of objects in the vehicle's surroundings, especially moving objects.

[0030] Furthermore, refining the object detection underlying the corresponding distinctive features and then checking the corresponding distinctive features based on the refined object detection is understood to mean that the entire computer resources at hand are dedicated or directed to resolving the distinctive features, i.e., the situation of the corresponding data processing system itself is taken into account and an alternative, possibly more accurate algorithm for object detection is selected, and then an attempt is made to resolve the distinctive features based on this more accurate algorithm, or that the object detection is adjusted so that non-distinctive areas or pixels that are not important for the corresponding distinctive features are omitted or not taken into account, in order to save computation time and be able to focus on resolving the corresponding distinctive features.

[0031] Furthermore, detecting objects in sensor data detected by other surrounding sensors among a plurality of different surrounding sensors in order to check for corresponding distinctive features means that these objects are verified by checking whether objects detected in the sensor data of one surrounding sensor are also detected in the sensor data of at least one other surrounding sensor.

[0032] This allows for even greater reliability in resolving distinctive features and thus in detecting objects in the vehicle's surroundings.

[0033] In one embodiment, the plurality of different ambient sensors includes video sensors and / or LIDAR sensors and / or radar sensors and / or ultrasonic sensors.

[0034] A video sensor is understood to be an image-based sensor that is able to record a moving scene and / or to evaluate the detected sensor data fully automatically.

[0035] A LIDAR sensor sends out a laser beam and detects the light backscattered at an object, from which the corresponding object can be inferred.

[0036] A radar sensor sends out electromagnetic waves, detects the waves reflected by an object, and can infer the corresponding object from the reflected waves.

[0037] Ultrasonic sensors again send out short, high frequency acoustic pulses and detect the pulses reflected off objects, from which the corresponding object can be inferred.

[0038] The different ambient sensors may therefore be sensors that are already installed in a typical vehicle or that are generally already on hand, and therefore the method can be implemented without the need for laborious and costly modifications.

[0039] However, including a plurality of different ambient sensors including a video sensor and / or a LIDAR sensor and / or a radar sensor and / or an ultrasonic sensor is merely one possible embodiment, and the plurality of different ambient sensors may instead include further sensors, such as acoustic sensors, that can detect objects around the vehicle.

[0040] According to a further embodiment of the present invention, there is also provided a method for controlling at least one function of a motor vehicle based on an object detected in the vicinity of the motor vehicle, the method comprising the following steps: detecting an object in the vicinity of the motor vehicle by the above-described method for reliably detecting an object in the vicinity of the motor vehicle; and controlling at least one function of the motor vehicle based on the detected object.

[0041] In this case, functions of the motor vehicle are understood to mean controllable functions of the motor vehicle, for example driver assistance systems or controllable driving functions.

[0042] Thus, a method for controlling at least one function of a vehicle is presented, based on an improved method for reliably detecting objects in the vehicle's surroundings. In particular, the method is based on a method for reliably detecting objects in the vehicle's surroundings, which can increase the reliability or accuracy of detecting objects in the vehicle's surroundings, thereby again avoiding corresponding safety-critical situations when controlling functions of the vehicle, such as driver assistance systems or driving functions, based on the detected objects. In particular, in this case, the discrimination ability or performance of the corresponding object detector can be increased, which again leads to improved performance of environmental perception. This can ensure a high discrimination rate while simultaneously reducing the number of incorrectly identified objects. Furthermore, blind areas of individual object detectors, i.e., situations in which it is difficult for the individual object detectors to detect objects, can be reliably identified.

[0043] According to a further embodiment of the present invention, there is also provided a system for reliably detecting objects in the surroundings of a motor vehicle, the motor vehicle having a plurality of different surrounding sensors, the system being configured to detect sensor data by at least one of the plurality of different surrounding sensors, the system having a detection unit for each of the at least one of the plurality of different surrounding sensors configured to detect an object in the corresponding detected sensor data, the system further having an identification unit configured to identify at least one distinctive feature for object detection based on the detected object, and a checking unit configured to check, for each of the at least one distinctive feature, a corresponding distinctive feature for reliably detecting objects in the surroundings of the motor vehicle.

[0044] Thus, an improved system for reliably detecting objects in the vicinity of a motor vehicle is presented. In particular, the reliability or accuracy of detecting objects in the vicinity of the motor vehicle can be increased, thereby again avoiding corresponding safety-critical situations when controlling a driver assistance system or driving function based on the detected objects. In particular, in this case, the discrimination ability or performance of the corresponding object detector can be increased, thereby again improving the performance of environmental perception. This can ensure a high discrimination rate while simultaneously reducing the number of incorrectly identified objects. Furthermore, blind areas of individual object detectors, i.e., situations in which it is difficult for the individual object detectors to detect objects, can be reliably identified.

[0045] In this case, the system may further comprise a respective pre-processing unit for each of at least one of the plurality of different surrounding sensors configured to pre-process corresponding detected sensor data, and for each of at least one of the plurality of different surrounding sensors, the detection unit is respectively configured to detect objects in the corresponding pre-processed sensor data, thereby enabling even greater reliability in detecting objects in the vehicle's surroundings, and in which the corresponding data can be optimally adapted or prepared for object detection.

[0046] Furthermore, the identification unit may be configured to identify at least one distinctive feature based on the detected object and further information describing the situation, thereby taking into account information about the current situation, e.g. current weather conditions, current light conditions and / or current vehicle speed, thereby enabling even greater reliability when detecting objects in the vehicle's surroundings.

[0047] Furthermore, the identification unit may be configured to identify at least one distinctive feature by applying a machine learning algorithm trained to identify distinctive features in object detection based on information about the detected object.

[0048] In this case, information relating to detected objects is understood to be data representative of the corresponding detected objects.

[0049] This allows for even further optimization of the identification of distinctive features and even further improvement in the reliability or accuracy with which distinctive features are identified.

[0050] Furthermore, the checking unit may be configured for each of the at least one distinctive feature to apply an event camera to check the corresponding distinctive feature, and / or to refine an object detection on which the corresponding distinctive feature is based and to check the corresponding distinctive feature on the basis of the refined object detection, and / or to check the corresponding distinctive feature by detecting an object in sensor data detected by other surrounding sensors of the plurality of different surrounding sensors, thereby providing even greater reliability in resolving the distinctive feature and thus in detecting objects in the vehicle's surroundings.

[0051] In one embodiment, the plurality of different ambient sensors includes video sensors and / or LIDAR sensors and / or radar sensors and / or ultrasonic sensors, and thus the plurality of different ambient sensors may be sensors already installed in a typical automobile or sensors that are generally already on hand, thus enabling the system to be implemented without the need for tedious and costly modifications.

[0052] However, including a plurality of different ambient sensors including a video sensor and / or a LIDAR sensor and / or a radar sensor and / or an ultrasonic sensor is merely one possible embodiment, and the plurality of different ambient sensors may instead include further sensors, such as acoustic sensors, that can detect objects around the vehicle.

[0053] According to a further embodiment of the present invention, there is also presented a system for controlling at least one function of a motor vehicle based on an object detected in the vicinity of the motor vehicle, the system comprising a receiving unit for receiving information about an object detected in the vicinity of the motor vehicle, the object having been detected by the above-mentioned system for reliably detecting objects in the vicinity of the motor vehicle, and a control unit configured to control at least one function based on the provided information.

[0054] In this case, information about detected objects in the surroundings of the motor vehicle is again understood to be data representative of the corresponding detected objects.

[0055] Thus, a system for controlling at least one function of a vehicle is presented, based on an improved system for reliably detecting objects in the vehicle's surroundings. In particular, the system is based on a system for reliably detecting objects in the vehicle's surroundings, which can increase the reliability or accuracy of detecting objects in the vehicle's surroundings, thereby again avoiding corresponding safety-critical situations when controlling functions of the vehicle, such as driver assistance systems or driving functions, based on the detected objects. In particular, in this case, the discrimination ability or performance of the corresponding object detector can be increased, which again leads to improved performance of environmental perception. This can ensure a high discrimination rate while simultaneously reducing the number of incorrectly identified objects. Furthermore, blind areas of individual object detectors, i.e., situations in which it is difficult for the individual object detectors to detect objects, can be reliably identified.

[0056] In summary, it can be seen that the present invention presents a method for detecting objects in the vicinity of a motor vehicle that can guarantee a high identification rate while reducing the number of incorrectly identified objects.

[0057] The described embodiments and developments can be combined with one another in any desired manner.

[0058] Further possible embodiments, developments and implementations of the invention also include not explicitly mentioned combinations of the features of the invention that are explained above or below with reference to the examples.

[0059] The accompanying drawings provide a further understanding of embodiments of the present invention, the drawings illustrating the embodiments and, in connection with the present specification, are used to explain the principles and concepts of the present invention.

[0060] Other embodiments and many of the above advantages will become apparent in view of the drawings, in which elements are not necessarily drawn to scale relative to each other. [Brief explanation of the drawings]

[0061] [Figure 1] 1 is a flowchart of a method for reliably detecting objects in the surroundings of a motor vehicle according to an embodiment of the present invention. [Figure 2] 1 is a schematic block diagram of a system for reliably detecting objects in the surroundings of a motor vehicle according to an embodiment of the present invention;

[0062] In the various figures of the drawings, identical or functionally equivalent elements, components or elements are designated by the same reference numerals unless otherwise stated. DETAILED DESCRIPTION OF THE INVENTION

[0063] FIG. 1 shows a flow chart of a method for robustly detecting objects in the surroundings of a motor vehicle 1 according to an embodiment of the invention.

[0064] In particular, FIG. 1 shows a flow chart of a method for reliably detecting objects in the surroundings of a motor vehicle 1 according to an embodiment of the invention, the motor vehicle having a number of different surrounding sensors.

[0065] Comprehensive identification of the vehicle's environment and, in particular, of objects, such as other road users, in the vehicle's environment provides the basis for many driver assistance systems and automated driving functions of the vehicle. In this case, the vehicle typically has a plurality of sensor systems, each tasked with identifying objects in its respective detection area. Each sensor system in this case includes one surroundings sensor and a processing unit, which is configured to analyze data provided by the surroundings sensor and, for example, detect objects in the data. In this case, the surroundings sensor may be, for example, a video camera, a radar, a LIDAR, or an ultrasonic sensor.

[0066] In this case, each individual one of these ambient sensors is very limited in terms of its respective field of use and may not provide all the required information about the vehicle's surroundings, depending on, for example, the installation location or the current weather conditions. In this case, by combining or fusing the information from different ambient sensors, it is possible to increase the reliability when identifying the vehicle environment and thus when detecting objects around the vehicle, thereby again avoiding corresponding safety-critical situations when controlling driver assistance systems or driving functions based on the detected objects.

[0067] In known approaches, in this case, objects are detected separately in the sensor data detected by different surrounding sensors, and then each detection result is fused by a fusion approach.However, especially in complex scenarios where visibility is poor, it is difficult to ensure a high identification rate of the objects around the automobile and at the same time to suppress the number of incorrectly identified objects to a low level.Therefore, there is a need for an improved method for reliably detecting objects around the automobile.

[0068] FIG. 1 shows method 1, which in this case includes step 2 of detecting sensor data by at least one of a plurality of different ambient sensors, step 3 of detecting an object in the corresponding detected sensor data for each of the at least one of the plurality of different ambient sensors, step 4 of identifying at least one distinctive feature for object detection based on the detected object, and step 5 of checking a corresponding distinctive feature for each of the at least one distinctive feature to reliably detect an object in the vicinity of the vehicle.

[0069] In this case, method 1 has the advantage of increasing the reliability or accuracy of detecting objects in the vehicle's surroundings, thereby again avoiding corresponding safety-critical situations when controlling a driver assistance system or driving function based on the detected objects. In particular, in this case, the discrimination ability or performance of the corresponding object detector can be increased, which again leads to an improved performance of environment perception. This can ensure a high discrimination rate while simultaneously reducing the number of incorrectly identified objects. Furthermore, blind areas of individual object detectors, i.e., situations in which it is difficult for the individual object detectors to detect objects, can be reliably identified.

[0070] Overall, therefore, an improved method 1 for reliably detecting objects in the surroundings of a motor vehicle is presented.

[0071] Thus, method 1 allows a particularly improved environmental perception to be achieved, and individual sensor or detection inadequacies can be identified and subsequently compensated for accordingly.

[0072] As shown in FIG. 1 , the method 1 further includes, for each of at least one of the plurality of different ambient sensors, step 6 of preprocessing corresponding detected sensor data, respectively, and for each of at least one of the plurality of different ambient sensors, step 3 of detecting an object in the corresponding detected sensor data, respectively, includes detecting an object in the corresponding preprocessed sensor data.

[0073] Pre-processing the corresponding sensor data may in this case comprise one or more processing steps, for example, applying semantic segmentation, applying a shape estimator, low-pass filtering or averaging, or applying time windowing of the detected sensor data.

[0074] According to the embodiment of FIG. 1, step 4 of identifying at least one distinctive feature in object detection based on the detected object further comprises identifying at least one distinctive feature based on the detected object and further information describing the situation.

[0075] The further information describing the situation or corresponding metadata may in this case be one or more features of, for example, the speed of the detected object, the current light conditions, the current weather conditions, situational context information such as whether the detected object is present at a pedestrian crossing, position data of the vehicle, or data from the vehicle's navigation system, or data relating to the planned route guidance.

[0076] In the embodiment of FIG. 1, the step of identifying at least one distinctive feature for object detection based on the detected object further includes applying a machine learning algorithm trained to identify distinctive features for object detection based on information about the detected object.

[0077] The corresponding machine learning algorithm may be, for example, a neural network trained on the basis of corresponding labeled training data.

[0078] In particular, the step of identifying at least one distinctive feature during object detection may in this case comprise applying an occupancy grid, in particular a two-dimensional occupancy grid, which in this case is made up of a plurality of cells and represents a two-dimensional Cartesian grid of the vehicle's surroundings in a bird's-eye view, each cell of the occupancy grid being associated with a corresponding coordinate, and each detected object, i.e. an object detected in at least one sensor data detected by at least one of the plurality of different surrounding sensors, is associated with, i.e. occupies, a corresponding coordinate or cell.

[0079] Subsequently, distinctive features or regions of interest for object detection can be identified based on occupied cells of the occupancy grid. For example, if multiple objects are detected in the sensor data detected by multiple different ambient sensors, a distinctive feature may exist if an object is detected in the sensor data of only some of the multiple different ambient sensors, but not all of the ambient sensors. On the other hand, if multiple objects are detected in the sensor data detected by only one ambient sensor, the corresponding detected objects may be distinctive features to be verified by detecting objects in the sensor data detected by the other ambient sensors.

[0080] Subsequent checks for distinctive features can be concentrated on the respective regions of interest.

[0081] According to the embodiment of FIG. 1 , for each of the at least one distinctive feature, step 5 of checking a corresponding distinctive feature may include applying an event camera to check the corresponding distinctive feature, refining an object detection underlying the corresponding distinctive feature and checking the corresponding distinctive feature based on the refined object detection, and / or detecting an object in sensor data detected by other ambient sensors of the plurality of different ambient sensors.

[0082] However, it is also possible to adjust the object detection thresholds underlying the corresponding discriminative features or the thresholds of the corresponding algorithms in order to achieve improved object detection.

[0083] Additionally, it is also possible to optimally or improve the use of ambient sensor resources, for example, by utilizing beamforming techniques to align sensor perception to a particular region of interest, or by increasing the sensor sampling frequency of at least one ambient sensor of the plurality of ambient sensors for each particular region of interest.

[0084] Overall, therefore, Figure 1 illustrates a method 1 in which, for example, when distinctive features are identified in an object detected by one sensor modality, a parallel pathway is provided that acts redundantly to a classical multimodal perception approach, or that can adjust or modulate the classical pathway, or in which greater robustness is achieved by comparing an object detected by one sensor modality with an object detected by another sensor modality.

[0085] According to the embodiment of FIG. 1, the plurality of different ambient sensors is again a video sensor, a LIDAR sensor, a radar sensor, and an ultrasonic sensor.

[0086] The corresponding reliably detected object can then be used, for example, to control functions of a vehicle, in particular an autonomous vehicle, where it is even possible to fuse the data of the individual ambient sensors in advance.

[0087] In this case, controlling the functions of the vehicle may include, for example, outputting a corresponding warning to the driver of the vehicle, intervening in longitudinal control and / or lateral control (e.g. initiating emergency braking), or outputting a corresponding warning to other road users, for example by automatic activation of the vehicle's signal generators, such as the horn or headlights.

[0088] FIG. 2 shows a schematic block diagram of a system for reliably detecting objects in the surroundings of a motor vehicle 10 according to an embodiment of the present invention.

[0089] In particular, FIG. 2 shows a schematic block diagram of a system for reliably detecting objects in the surroundings of a vehicle 10 according to an embodiment of the present invention, the vehicle having a number of different surrounding sensors 11, 12, 13, 14.

[0090] According to the embodiment of Figure 2, the system 10 is in this case configured to detect sensor data by at least one of a plurality of different ambient sensors 11, 12, 13, 14, and the system 10 has, for each of the at least one of the plurality of different ambient sensors, a detection unit 15 configured to detect an object in the corresponding detected sensor data, and the system 10 further has an identification unit 16 configured to identify at least one distinctive feature for object detection based on the detected object, and a checking unit 17 configured to check, for each of the at least one distinctive feature, a corresponding distinctive feature in order to reliably detect an object in the vicinity of the motor vehicle.

[0091] In this case, the detection unit, the identification unit and the check unit may each be realized based on corresponding code stored in a memory and executable by a processor, for example.

[0092] As shown in FIG. 2, the system further comprises, for each of at least one of the plurality of different ambient sensors 11, 12, 13, 14, a respective pre-processing unit 18 configured to pre-process corresponding detected sensor data, and for each of at least one of the plurality of different ambient sensors, a respective detection unit 15 configured to detect an object in the corresponding pre-processed sensor data.

[0093] In this case, the pre-processing unit may again be realized on the basis of corresponding code that is stored in a memory and that is executable by a processor, for example.

[0094] According to the embodiment of FIG. 2, the identification unit 16 is again configured to identify at least one distinctive feature based on the detected object and further information describing the situation.

[0095] Furthermore, the identification unit 16 is also configured to identify at least one distinctive feature by applying a machine learning algorithm trained to identify distinctive features in object detection based on information about the detected object.

[0096] According to the embodiment of Figure 2, the checking unit 17 is again configured, for each of the at least one distinctive feature, to apply an event camera to check the corresponding distinctive feature, and / or to refine the object detection underlying the corresponding distinctive feature and check the corresponding distinctive feature based on the refined object detection, and / or to check the corresponding distinctive feature by detecting an object in sensor data detected by other surrounding sensors of the plurality of different surrounding sensors.

[0097] The plurality of different ambient sensors 11, 12, 13, 14 are again a video sensor 11, a LIDAR sensor 12, a radar sensor 13 and an ultrasonic sensor 14.

[0098] Furthermore, the system 10 according to FIG. 2 is configured to implement the above-described method for reliably detecting objects in the surroundings of a motor vehicle.

Claims

1. 1. A method for reliably detecting objects in the surroundings of a motor vehicle, comprising: the vehicle has a plurality of different ambient sensors; The method (1) includes the following steps: (2) detecting sensor data by at least one of the plurality of different ambient sensors; - for each of at least one of the plurality of different ambient sensors, respectively detecting an object within the corresponding detected sensor data; - identifying (4) at least one distinctive feature for object detection based on the detected object; - for each of said at least one distinctive feature, checking each corresponding distinctive feature in order to reliably detect objects in the surroundings of said vehicle; A method comprising:

2. The method (1) further includes a step (6) of pre-processing the corresponding detected sensor data for each of at least one of the plurality of different ambient sensors; and for each of at least one of the plurality of different ambient sensors, the step (3) of respectively detecting an object in the corresponding detected sensor data includes detecting an object in the corresponding pre-processed sensor data. The method of claim 1.

3. and wherein the step (4) of identifying at least one distinctive feature for object detection based on the detected object comprises identifying at least one distinctive feature based on the detected object and further information describing the situation.

3. The method according to claim 1 or 2.

4. and wherein the step (4) of identifying at least one distinctive feature for object detection based on the detected object comprises applying a machine learning algorithm trained to identify distinctive features for object detection based on information about the detected object.

4. The method according to any one of claims 1 to 3.

5. For each of the at least one distinctive feature, the step (5) of checking the corresponding distinctive feature comprises: applying an event camera to check the corresponding distinctive features; and / or refining the object detection underlying the corresponding distinctive features and checking the corresponding distinctive features based on the refined object detection; and / or detecting objects in sensor data detected by other ambient sensors of the plurality of different ambient sensors to check for corresponding distinctive features. Including, 5. The method according to any one of claims 1 to 4.

6. the plurality of different ambient sensors include video sensors and / or LIDAR sensors and / or radar sensors and / or ultrasonic sensors; 6. The method according to any one of claims 1 to 5.

7. 1. A method for controlling at least one function of a vehicle based on objects detected in the vehicle's surroundings, comprising: The method comprises the following steps: - detecting objects in the surroundings of a motor vehicle by a method for reliably detecting objects in the surroundings of a motor vehicle according to any one of claims 1 to 6; - controlling at least one function of the vehicle based on the detected object; A method comprising:

8. 1. A system for reliably detecting objects in the vicinity of a motor vehicle, comprising: the vehicle has a plurality of different ambient sensors; the system (10) is configured to detect sensor data by at least one of the plurality of different ambient sensors (11, 12, 13, 14); the system (10) comprises, for each of at least one of the plurality of different ambient sensors (11, 12, 13, 14), a respective detection unit (15) configured to detect an object in the corresponding detected sensor data; The system (10) comprises: an identification unit (16) configured to identify at least one distinctive feature in object detection based on the detected object; a checking unit (17) configured for each of said at least one distinctive feature to check a corresponding one of said distinctive features in order to reliably detect objects in the surroundings of said vehicle; further comprising system.

9. The system (10) further comprises, for each of at least one of the plurality of different ambient sensors (11, 12, 13, 14), a respective pre-processing unit (18) configured to pre-process the corresponding detected sensor data; for each of at least one of the plurality of different ambient sensors (11, 12, 13, 14), the detection unit (15) is respectively configured to detect an object in the corresponding pre-processed sensor data; The system of claim 8.

10. the identification unit (16) is configured to identify the at least one distinctive feature based on the detected object and further information describing the situation.

10. The system according to claim 8 or 9.

11. the identification unit (16) is configured to identify the at least one distinctive feature by applying a machine learning algorithm trained to identify distinctive features in object detection based on information about the detected object. A system according to any one of claims 8 to 10.

12. The checking unit (17) checks for each of the at least one distinctive feature: applying an event camera to check the corresponding distinctive features; and / or refining the object detection underlying the corresponding distinctive features and checking the corresponding distinctive features based on the refined object detection; and / or and checking the corresponding distinctive features by detecting the object in sensor data detected by other ambient sensors of the plurality of different ambient sensors. It is composed of 12. A system according to any one of claims 8 to 11.

13. The plurality of different ambient sensors (11, 12, 13, 14) includes a video sensor (11) and / or a LIDAR sensor (12) and / or a radar sensor (13) and / or an ultrasonic sensor (14).

13. A system according to any one of claims 8 to 12.

14. 1. A system for controlling at least one function of a vehicle based on objects detected in a surrounding of the vehicle, comprising: The system comprises: a receiving unit for receiving information about objects detected in the surroundings of the motor vehicle, the objects having been detected by a system for reliably detecting objects in the surroundings of a motor vehicle according to any one of claims 8 to 13; a control unit configured to control the at least one function based on the provided information; A system having:

Images