Redundancy information for an object interface for highly and fully automated driving

Abstract

A system for the reliability of objects for driver assistance or automated driving in a vehicle, the system including a plurality of sensors for providing sensor data for the objects, the plurality of sensors including one or more sensor modalities. An electronic tracking unit is configured to receive the sensor data in order to: determine a detection probability (p_D) for each object for each of the plurality of sensors, determine an existence probability (p_ex) for each object for each of the plurality of sensors, and provide a vector for each object based on the existence probability (p_ex) of the contributing sensors for each of the plurality of sensors for a particular object. The vector is provided by the electronic tracking unit for display as an object interface on a display device. The vector is independent of the sensor data from the plurality of sensors.

Description

[0001] Cross - reference to related applications

[0002] This application claims the priority of U.S. Provisional Application No. 62 / 854,729, filed on May 30, 2019, the disclosure of which is incorporated herein by reference in its entirety. Technical field

[0003] The present disclosure relates to driver assistance or autonomous driving. More specifically, the present disclosure relates to a method and system for use in driver assistance or autonomous driving of a vehicle for determining the reliability of sensed objects. Background art

[0004] In driver assistance and in automated driving, a representation of the vehicle environment is typically selected, among other things, in the form of a list of objects. These objects describe, among other things, other road users. Based on the object properties, a function decides whether and how a reaction to it should occur. An example in this regard is today's autonomous emergency braking system (AEB system), which identifies whether a collision with another road user is imminent and, if so, intervenes accordingly. Since environmental perception can be incorrect, a quality metric is calculated for the object, and this quality metric is used by the function to decide whether the object is reliable enough to trigger, for example, emergency braking. A typical metric used in today's driver assistance systems (DA systems) or arrangements is the object presence probability. Since false - positive interventions must be avoided in the AEB system, the function typically only reacts to objects with a sufficiently high presence probability and ignores all other objects. In addition, in AEB systems operating with multiple sensors, confirmation flags are frequently used. Emergency braking is only triggered when both sensors have confirmed the object.

[0005] Since both false - positive and false - negative reactions must be avoided, this tried - and - tested path for DA systems is no longer possible for autonomous driving. The trade - off between false positives (FP) and false negatives (FN) cannot be fixed, but rather this trade - off depends on the intervention severity.

[0006] Since self - driving cars have a redundant set of sensors, it is possible to keep a record of which sensors (e.g., radar sensors, video imaging sensors, Lidar sensors) have confirmed the object for each object. Depending on the appropriate manifestation of the trade - off between FP and FN, only objects seen by one sensor or by multiple sensors are considered.

[0007] A further motivation in this regard is the system response evaluated according to Automotive Safety Integrity Level D (ASIL D) as defined by ISO 26262 from the International Organization for Standardization (e.g., emergency braking from high speed in the case of high-speed deceleration). For example, from the perspective of electrical hardware errors, the information of a single ASIL B sensor is not reliable enough.

[0008] A disadvantage of the described method is the time aspect. In this regard, it may happen that the object is only incidentally measured by one of these sensors, or the measurement result associated with the object only imprecisely matches (e.g., deviates from the object type classification, deviates from the Doppler speed in the case of a radar sensor). In particular, in the case of a dynamic scenario, of interest is not only the presence of the object (i.e., whether the object is a phantom object or a real object), but also how consistently the different sensors have measured the object attributes (especially speed). Summary of the Invention

[0009] Embodiments herein describe a method and system for representing the reliability of an object, having the following properties:

[0010] 1. Estimate how consistent and reliable the dynamic state of the object is based on the existing sensor signals;

[0011] 2. Use a probabilistic representation: Instead of setting flags, calculate continuous values, and depending on the criticality of the system response, different thresholds can be fixed to these continuous values;

[0012] 3. Provide various sensor configurations, as different numbers of diverse sensor technologies are envisioned;

[0013] 4. Encapsulate sensor-specific knowledge so that the planning unit can evaluate the redundancy / reliability of the object independently of the knowledge about the sensors used and the sensor principle;

[0014] 5. Provide an object interface for customers who want to develop an independent electronic planning unit.

[0015] For this purpose, considering the detection probability of the sensor for each object, calculate the presence probability specific to the sensor type, and then convert this probability into a sensor-independent presence probability vector.

[0016] In addition, when evaluating the redundancy of the object, it is necessary to consider which sensors / measurement principles can actually measure the object (not only the visibility range, but also environmental conditions, sensor blindness, degradation, dynamic concealment, etc.).

[0017] In one embodiment, a system for driver assistance or automated driving of a vehicle by detecting the reliability of detected objects includes: a plurality of sensors for providing sensor data for the objects, the plurality of sensors including different sensor modalities. The system includes an electronic tracking unit for receiving the sensor data. The electronic tracking unit is configured to process the sensor data to: determine a detection probability (p_D) for the objects for each of the plurality of sensors, and determine a presence probability (p_ex) for the objects for each of the plurality of sensors. The electronic tracking unit is further configured to: provide a vector for each object based on the presence probability (p_ex) for each object for each of the plurality of sensors, wherein the vector includes all the presence probabilities for each object for all contributing sensors among the plurality of sensors. The vector is a sensor-independent representation.

[0018] In another embodiment, a system for determining the reliability of objects detected for a driver assistance arrangement or an autonomous vehicle is provided. The system includes: a plurality of sensors for providing sensor data for the objects, the plurality of sensors including different sensor modalities; and an electronic tracking unit for receiving the sensor data. The electronic tracking unit is configured to process the sensor data to: determine a detection probability (p_D) for each object for each of the plurality of sensors, determine a presence probability (p_ex) for each object for each of the plurality of sensors, and provide a vector for each object based on the presence probability (p_ex) for each object for all contributing sensors among the plurality of sensors. A display device displays the vector as an object interface.

[0019] Other aspects, features, and embodiments will become apparent upon consideration of the detailed description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A block diagram of a vehicle equipped with a vehicle system according to one embodiment is illustrated.

[0021] Figure 2 An object interface of objects disposed around a vehicle according to one embodiment is illustrated.

[0022] Figure 3 Illustrated according to Figure 2 Object interfaces of objects disposed around a vehicle at different times. DETAILED DESCRIPTION

[0023] Before explaining any embodiments in detail, it is to be understood that the present disclosure is not intended to be limited in its application to the details of the construction and arrangement of components set forth in the following description or illustrated in the following drawings. Embodiments can have other configurations and can be practiced or carried out in various ways.

[0024] A variety of embodiments can be implemented using multiple hardware - and software - based devices and multiple different structural components. Additionally, embodiments can include hardware, software, and electronic components or modules, which for purposes of discussion can be illustrated and described as if most components are implemented only in hardware. However, those of ordinary skill in the art and upon reading this detailed description will recognize that in at least one embodiment, the electronic - based aspects of the present invention can be implemented in software (e.g., stored on a non - transitory computer - readable medium) executable by one or more electronic controllers. For example, the "units", "control units", and "controllers" described in the specification can include one or more electronic controllers, one or more memories including non - transitory computer - readable media, one or more input / output interfaces, one or more application - specific integrated circuits (ASICs) and other circuits, and various connections (e.g., wires, printed traces, and buses) connecting the various components.

[0025] Figure 1 A block diagram of a vehicle system 20 for tracking objects and for determining the reliability of object presence near a vehicle is shown. The vehicle system 20 includes an electronic tracking unit 30. The electronic tracking unit 30 includes an electronic processor 34, a memory 36, and an input / output (I / O) interface 38 connected to a communication bus 40. The memory 36 includes at least one or both of a random - access memory (RAM) and a read - only memory (ROM). The electronic processor 34 of the electronic tracking unit 30 is configured to execute a program for tracking objects as set forth below.

[0026] Figure 1 The communication bus 40 shown is a flex - ray bus, a CAN bus, or other type of communication link among multiple control units, sensors, and other devices. Figure 1 The user interface 44 shown enables a user to provide input to various devices in the vehicle system 20. The display device 48 provides a visual display of information to the user. In one embodiment, the display device 48 and the user interface 44 are combined in a touchscreen. In another embodiment, the user interface 44 includes a keyboard for receiving input. In one embodiment, a display device 48 for displaying an object interface is contemplated. In another embodiment, the user interface 44 includes a microphone and a voice analyzer for receiving voice commands or input.

[0027] Figure 1A plurality of video imaging sensors 50, such as video cameras, are shown for providing video images of objects around the vehicle. A plurality of radar sensors 54 are provided for providing radar sensing of objects around the vehicle. In addition, Doppler sensing is provided to determine the relative speed of the detected objects. A plurality of light detection and ranging (Lidar) sensors 60 are also provided for detecting objects around the vehicle and their distances. In one embodiment, the Lidar sensors 60 are mounted in a Lidar sensing unit above the vehicle roof, and the Lidar sensors are rotatable for scanning around the vehicle.

[0028] Figure 1 The vehicle system 20 in [description] includes a vehicle speed sensor 64 for providing the vehicle speed to the respective units. A global positioning signal (GPS) receiver 66 is provided to receive GPS signals to determine the position of the vehicle and for mapping and for other purposes. A transceiver 68 is provided for remote two-way wireless communication with other remote devices.

[0029] Figure 1 The vehicle system 20 in one embodiment includes an electronic planning unit 70 for providing driver assistance and, in another embodiment, includes autonomous control of the vehicle. The electronic planning unit 70 includes an electronic processor 74, a memory 76, and an input / output (I / O) interface 78 connected to a communication bus 40. The memory 76 includes at least one or both of a random access memory (RAM) and a read-only memory (ROM). The electronic processor 74 is configured to execute programs for planning the control of an autonomous vehicle or for planning driver assistance for a driver-controlled vehicle.

[0030] Figure 1 The vehicle system 20 includes a steering control unit 80 for controlling the steering direction of the vehicle and an accelerator control unit 84 for controlling the acceleration of the vehicle. The vehicle system 20 includes a brake control unit 86 for selectively controlling the braking of the vehicle. The vehicle system 20 further includes a steering angle sensor 90 for determining the steering angle position of the vehicle and a yaw rate sensor 94 for determining the yaw rate of the vehicle. In one embodiment, the vehicle system 20 includes a data event recorder 96 for recording vehicle data for future use. In an autonomous vehicle control embodiment, the GPS receiver 66 receives signals to map the position of the vehicle, and the electronic processor 74 of the electronic planning unit 70 determines where to control the steering control unit 80 and how to control the accelerator control unit 84 and / or the brake control unit 86 based on vehicle mapping, vectors from the electronic tracking unit 30, and other data to guide the vehicle to a predetermined stored destination.

[0031] Operation

[0032] A method involves an electronic processor 34 of an electronic tracking unit 30 for calculating a separate object presence probability p_ex for each sensor modality based on: a) the detection probability p_D of an object; b) the probability of an incorrect measurement result; c) the measurement result likelihood, that is, how well the associated measurement result matches the target estimate; and d) the presence probability of the object in the previous cycle. The calculation of the stated reliability representation is performed by the electronic processor 34 in each cycle independently of which type of sensor is used to perform the measurement to update the object. The presence probability value p_ex is between 0 and 1, where a value of 0 means the object is not detected. In another embodiment, p_D is calculated in sensor preprocessing rather than by the electronic tracking unit 30.

[0033] In this method, the electronic tracking unit 30 calculates separate presence probabilities for one or more sensor modalities. For example, in the case of a sensor set including a radar sensor 54, a Lidar sensor 60, and a video imaging sensor 50, the presence probability p_ex,R (for radar), the presence probability p_ex,L (for Lidar), and the presence probability p_ex,V (for video) are calculated. This is an advantageous embodiment for dynamic objects representing other road users because these objects can be identified with all sensor modalities. In other embodiments, some sensor modalities do not identify these objects.

[0034] The method can be applied to attributes of one or more related objects that can only be identified with a specific sensor modality but are in turn identified by multiple instances of that sensor. An example is the identification of traffic lights. The traffic light state (red, yellow, green,...) can only be measured by the video imaging sensor 50. In some embodiments, the transceiver 68 is a Car2X transceiver for receiving the traffic light state. If multiple cameras are used to determine the color of the traffic light, it is advantageous to calculate a separate presence probability for each camera, that is, for example, p_ex,V1 (the first camera), p_ex,V2 (the second camera), p_ex,V3, etc. There is no limitation to three values here. In other words, in a generalized way, a vector of presence probabilities is calculated with N values p_ex,i. These N values here reflect the type of redundancy that is intended to be modeled (that is, the redundancy of object measurements by different sensor modalities, the redundancy of traffic light state measurements by different video imaging sensors 50).

[0035] Only the measurement results of the corresponding sensor type are used to update the corresponding presence probability p_ex,i, for example, p_ex,R is updated only when radar measurement results are integrated.

[0036] In addition to the vector with the existence probability, the electronic tracking unit 30 also determines a vector of the same size with a detection probability p_D,i. This vector represents the sensor modality for which the object is visible (or the video imaging sensor 50 in the case of a traffic light sign). The detection probability is made available as information from the sensor based on sensor data, and in some cases based on the current environmental model, and in some cases by using map data in each measurement cycle for each object. In this case, factors such as concealment, sensor visibility range, object category, object attributes, etc. are considered, but sensor failures, sensor blindness, etc. are also considered. In addition, it is possible to consider specific electrical hardware errors on the signal path in the calculation of p_D. For example, when the demosaicing in a video imaging sensor 50, such as a camera, is defective, the p_D of that camera decreases. Then, the measurement result does not necessarily need to be discarded. The higher the probability that the sensor can measure the object, the closer the corresponding value of p_D is to 1. Each entry of the vector with the detection probability typically represents multiple sensor instances, such as all Lidar sensors 60. Therefore, in each processing step, the electronic tracking unit 30 forms the maximum value of all the detection probabilities p_D belonging to the vector entry. If no measurement result is received in the processing step and thus no p_D,i of the sensor modality is received, the corresponding entry from the previous cycle is used and the corresponding entry is decreased by a value that depends on the time difference relative to the last measurement value. As an example, the value of p_D,i for cycle k can then be calculated by the electronic tracking unit 30 as follows:

[0037] p_D,i(k) = p_D,i(k - 1) – ΔT * constant.

[0038] In this way, each object contains information about in which measurement principle the object can currently be seen and about how well the corresponding sensor measurement results match the object estimate.

[0039] Example of no video sensor operation

[0040] In one operation, the electronic processor 34 of the electronic tracking unit 30 receives sensor data from sensors 50, 54, 60 and determines that the vehicle object contains or corresponds to the following values:

[0041] p_D,Radar = 0.9, p_D,Video = 0.1, p_D,Lidar = 0.8

[0042] p_ex,Radar = 0.1, p_ex,Video = 0, p_ex,Lidar = 0.99

[0043] At this moment, the object can actually only be measured by the radar sensor 54 and the Lidar sensor 60 (e.g., this is because the video imaging sensor 50 is soiled). However, the radar sensor 54 measures the object only very unreliably (p_ex,Radar is very low), while the Lidar sensor 60 measures the object very reliably. Therefore, in the calculations performed by the electronic tracking unit 30, the above values mainly depend on the Lidar sensor 60.

[0044] Four examples of video imaging sensors

[0045] The vehicle has been equipped with four video imaging sensors 50 such as cameras for identifying the status of traffic lights. The electronic tracking unit 30 determines the following values for the traffic light object for cameras 1 - 4:

[0046] p_D,1 = 0.1, p_D,2 = 0.3, p_D,3 = 0.9, p_D,4 = 0.9

[0047] p_ex,1 = 0, p_ex,2 = 0.1, p_ex,3 = 0.8, p_ex,4 = 0.7

[0048] In this instance of the four cameras, only cameras #3 and #4 can reliably see the traffic light(s) (e.g., due to the smaller range / distance of cameras #1 and #2). The electronic tracking unit 30 determines that the measurements from the third and fourth cameras match the traffic light estimate very well and that the latter's measurements are consistent. The traffic light is first identified based on how many pixels represent it, and the color of the light is determined.

[0049] In a further calculation step of the method performed by the electronic tracking unit 30, the sensor - specific parts are extracted in order to be able to determine a vector for the general object interface for the electronic planning unit 70.

[0050] For this purpose, first the electronic tracking unit 30 forms a subset of all p_ex,i for which p_D,i exceeds a threshold. The electronic tracking unit 30 selects the sensor modalities that can actually measure the object at the current time point. In one embodiment, the threshold p_D,th is selected to be 0.3. After that, the remaining p_ex,i, the maximum value p_ex,max, the minimum value p_ex,min, and the median value p_ex,med are calculated. These three values are made available as redundancy information for the vector and for the object interface. Thus, the electronic tracking unit 30 is configured to sense the presence of the traffic light and its color.

[0051] Radar / video / Lidar example

[0052] Another set of examples for multiple different sensor modalities is as follows. Three sensor modalities (radar, video, Lidar) consistently measure an object; the object is visible to all sensor modalities. In this case, p_ex,max, p_ex,med, and p_ex,min are all very close to 1. Thus, the object has complete redundancy, and if necessary, ASIL D maneuvers (e.g., emergency braking) can be performed for the object. All three values being very high is normal for an object in the immediate vicinity of the SDC, e.g., for a vehicle ahead. Thus, the electronic tracking unit 30 is configured to provide a set of existence probabilities defining the vectors of each sensed object based on the existence probability (p_ex) for each object for each of the plurality of sensors.

[0053] Sensor contamination example

[0054] Three sensor modalities measure an object, but the measurement result of one of these sensors only poorly matches the overall object estimate, or the object is only sporadically measured (the reason may be, for example, undetected contamination of the sensor). All sensor modalities have a high p_D, that is, these sensors are all capable of measuring the object. In this case, p_ex,max and p_ex,med are close to 1, but p_ex,min is low (e.g., at 0.4).

[0055] Example of two out of three sensor modalities

[0056] Only two of the three available sensor modalities are able to measure the object (e.g., this is because the visibility range of one of the sensor principles is less than that of the other sensor principles, and the object is correspondingly far away); these two consistently and reliably measure the object. In this case, p_ex,max, p_ex,med, and p_ex,min are all very close to 1. This is the same redundancy level as in the earlier example and shows that the method is able to encapsulate knowledge about the sensor setup used, such as the individual visibility ranges of the sensors, e.g., at the interface to the electronic planning unit 70.

[0057] Phantom object example

[0058] Only one sensor consistently measures and confirms the object; all other sensors do not confirm the object, even if it is within visibility range and unoccluded. In this case, p_ex,max is close to 1, but p_ex,med and p_ex,min are 0 (or close to 0). There is probably a phantom object involved here, in response to which, in some cases, a severe intervention should not be triggered. However, to minimize any risks, for example at intersections, waiting in a safe stop state will still continue until the object with a low redundancy level has driven through. Such objects will also be considered when planning evasive trajectories around other objects, for example.

[0059] In an alternative embodiment, the method can be generalized by outputting a variable-length vector instead of the minimum, maximum, and median of the presence probabilities. The vector can contain, for example, all the presence probabilities of all contributing sensor modalities. For three contributing sensor modalities (e.g., video, radar, Lidar), the vector is then the same as the described method that uses the maximum of the presence probabilities (p_ex,max), the minimum of the presence probabilities (p_ex,min), and the median of the presence probabilities (p_ex,med) for each object of the sensor modality.

[0060] Object interface

[0061] Figure 2 and Figure 3 shows the results of an implementation using three sensor modalities (video imaging sensor 50, radar sensor 54, Lidar sensor 60). The host vehicle is labeled as host object 102. An object sensed as moving with high redundancy is labeled as 104 (p_ex,min, p_ex,max, and p_ex,med are all close to 1), an object with low redundancy is labeled as phantom object 110 (only p_ex,max is close to 1), and an object measured by only two of the three sensors is labeled as object 114 (p_ex,min is close to 0, p_ex,med and p_ex,max are close to 1). Static objects are labeled as static object 120. The objects of the object interface 100 all correspond to different vectors. Thus, these vectors correspond to the object interface, where each of these vectors includes all the presence probabilities for the corresponding object for all contributing sensors among the multiple sensors. Of course, for one sensor, different presence probabilities can correspond to different objects, depending on their distance from the sensor or their position in the field of view from sensors 50, 54, 60. Therefore, when determining the presence probabilities of the vector corresponding to an object, different sensors are ignored.

[0062] Figure 2 A phantom object 110 at the front right relative to the main object 102 is shown. None of the sensors 50, 54, 60 identifies the vehicle ahead as the phantom object 110.

[0063] All real objects are identified with complete redundancy. Multiple objects each correspond to a separate vector to be displayed determined by the electronic tracking unit 30. However, Figure 3 A phantom object 110 at the front left of the main object 102 is shown, which is generated due to an error in one of the sensors 50, 54, 60 but not confirmed by other sensor modalities.

[0064] This embodiment is directly visible whenever the object interface is externally visible and can thus be demonstrated on the object interfaces 100, 150. The object interfaces 100, 150 are externally visible when: a) when delivered to one or more third parties for developing an electronic planning unit or for other purposes (an original equipment manufacturer (OEM) accessing the supplier's object interface 100); b) when transmitted between different electronic control units in the vehicle, such as between the electronic tracking unit 30 and the electronic planning unit 70; c) when recorded in the data event recorder 96 as a relevant interface; or d) when transmitted via the transceiver 68 to a remote operation location. In another embodiment, the object interface is provided from the electronic tracking unit 30 to at least one of the items from the group including: the electronic planning unit 70 in the vehicle; the data event recorder 96; and wirelessly transmitted by the transceiver 68 to a remote remote operation location.

[0065] In one embodiment, the control of the vehicle represents at least one selected from the group including: accelerating the vehicle, decelerating the vehicle, and steering the vehicle.

[0066] Various features, advantages, and embodiments are set forth in the following claims.

Claims

1. A system for use in driver assistance or automated driving of a vehicle for determining the reliability of a sensed object, the system comprising: a plurality of sensors for providing sensor data for a sensed object, the plurality of sensors including one or more sensor modalities; and an electronic tracking unit for receiving the sensor data, the electronic tracking unit being configured to process the sensor data to: determine a detection probability for each object for each of the plurality of sensors, determine a presence probability for each object for each of the plurality of sensors, and provide a set of presence probabilities defining a vector of each sensed object based on the presence probability for each object for each of the plurality of sensors, wherein the vector includes all presence probabilities for all contributing sensors of the plurality of sensors for each object, and wherein the vector is a sensor-independent representation.

2. The system according to claim 1, wherein the vector is provided by the electronic tracking unit for display as an object interface on a display device.

3. The system according to claim 1, comprising an electronic planning unit, wherein the electronic planning unit is configured to receive the vector and control the vehicle.

4. The system according to claim 3, wherein the electronic planning unit controls at least one selected from the group consisting of: accelerating the vehicle, decelerating the vehicle, and steering the vehicle.

5. The system according to claim 1, wherein the plurality of sensors includes Lidar sensors, radar sensors, and video imaging sensors.

6. The system according to claim 5, wherein the electronic tracking unit is configured to determine a presence probability for each sensor modality to provide a presence probability of the Lidar sensor, a presence probability of the radar sensor, and a presence probability of the video imaging sensor for each sensed object.

7. The system according to claim 6, wherein the presence probability of the Lidar sensor, the presence probability of the radar sensor, and the presence probability of the video imaging sensor all have values between 0 and 1, where a value of 0 means the object is not detected.

8. The system according to claim 1, wherein the electronic tracking unit is configured to sense the presence and color of a traffic light, wherein the plurality of sensors includes four video imaging sensors, wherein the four video imaging sensors have a detection probability for sensing the traffic light, and the electronic tracking unit is configured to provide a presence probability to a planning unit for control of the vehicle, and wherein the vehicle is an autonomous vehicle.

9. The system according to claim 2, wherein the object interface is provided to a third party for the development of an independent electronic planning unit.

10. The system according to claim 2, wherein the object interface is provided from the electronic tracking unit to at least one from the group including: an electronic planning unit in the vehicle; a data event recorder; and transmitted to a remote teleoperation location.

11. A system for a driver assistance arrangement or an autonomous vehicle for determining the reliability of a sensed object, the system comprising: a plurality of sensors for providing sensor data for an object, the plurality of sensors including one or more sensor modalities; an electronic tracking unit for receiving the sensor data, the electronic tracking unit being configured to process the sensor data to: determine a detection probability for each object for each of the plurality of sensors, determine a presence probability for each object for each of the plurality of sensors, and provide a vector for each object based on the presence probabilities for each object for all contributing sensors among the plurality of sensors, wherein the vector is a sensor-independent representation; and a display device for displaying the vector as an object interface.

12. The system according to claim 11, wherein the object interface is provided to a third party for the development of an electronic planning unit.

13. The system according to claim 11, wherein the presence probability comprises: a presence probability maximum, a presence probability minimum, and a presence probability median, which are provided as redundancy information for the object interface.

14. The system according to claim 13, wherein the plurality of sensors includes Lidar sensors, radar sensors, and video imaging sensors, all of which have different sensor modalities.

15. A system for driver assistance or automated driving of a vehicle by determining the reliability of a sensed object, the system comprising: a plurality of sensors for providing sensor data for an object, the plurality of sensors including one or more sensor modalities; and an electronic tracking unit for receiving the sensor data, the electronic tracking unit being configured to process the sensor data to: determine a detection probability for each object for each of the plurality of sensors, determine a presence probability maximum for each object for each of the plurality of sensors, determine a presence probability minimum for each object for each of the plurality of sensors, and determine a presence probability median for each object for each of the plurality of sensors; wherein the electronic tracking unit is configured to: provide a vector for each object based on the presence probability maximum, presence probability minimum, and presence probability median for each object for each of the plurality of sensors, wherein each of the vectors includes all the presence probabilities for the corresponding object for all contributing sensors among the plurality of sensors, and wherein the vector is a sensor-independent representation.

16. The system according to claim 15, wherein the vector corresponds to an object interface, and the object interface is provided to at least one from the group including: an electronic planning unit in a vehicle, the electronic planning unit being configured to evaluate the redundancy / reliability of each object; a data event recorder; and wireless transmission to a remote teleoperation location.

Images