Event-based object recognition and tracking

Abstract

A method for identifying and / or tracking an object in a spatial region, the method comprising the steps of observing the region using at least one event-based sensor, wherein the event-based sensor has light-sensitive pixels, and wherein a relative change in the intensity of light incident on the pixel by at least a predetermined percentage causes the sensor to output an event assigned to the pixel; in response to the sensor outputting a new event, determining an evaluation for the event, the evaluation being a measure of how well the event matches a distribution of events already detected and / or how confident it is that the event originated from a detected object; in response to the determined evaluation satisfying a predetermined criterion, assigning the new event to the detected distribution or the detected object. A system for operating a vehicle and / or operating a monitoring system for monitoring a spatial region, comprising: an evaluation module, an interface module, and an operating module.

Description

Technical Field

[0001] The present invention relates to identifying and tracking objects in an area based on optical observation of the area. Background Art

[0002] The human driver controlling the vehicle receives the majority of the information relevant to the driving task with his eyes. Accordingly, information from optical observation of the vehicle's surroundings is also an important source of information for driver assistance systems and systems for at least partially automated driving.

[0003] To avoid accidents, quick reaction time is extremely important. Scientific studies of multiple accidents have proven that a high percentage of these accidents could have been avoided if just a little more time (mostly just one second) had been provided for reaction.

[0004] There are two factors that increase the response time when processing camera images. On the one hand, the camera only updates the image at a pre-set repetition rate per second. On the other hand, the image data stream requires a high bandwidth for transmission within the vehicle.

[0005] Therefore, US 2016 / 096 477 A1 proposes to expand a conventional image sensor with an additional event-based sensor element. Whenever the light intensity incident on such an element changes by more than a certain relative value, the sensor immediately outputs an event, regardless of the repetition rate of the camera. Summary of the invention

[0006] In the context of the present invention, a method for detecting and / or tracking objects in a spatial area is developed. The method can be used in particular for observing the surroundings of a vehicle for the purpose of solving driving tasks. However, the method can also be used, for example, for monitoring an area in a fixed position with respect to intruders or unauthorized activities.

[0007] In the method, at least one event-based sensor is used to observe the area. The event-based sensor has light-sensitive pixels. A relative change of at least a predetermined percentage in the intensity of light incident on a pixel causes the sensor to output an event assigned to the pixel.

[0008] In response to the sensor outputting a new event, the following evaluation is determined for the event: to what extent the event matches the distribution of events that have already been detected and / or to what extent it is plausible that the event originated from the detected object. In response to the determined evaluation satisfying a predetermined criterion, the new event is assigned to the detected distribution or the detected object.

[0009] The term "detection" does not limit the method and manner of obtaining knowledge of the distribution or the object. Thus, for example, the distribution can be formed from events provided by an event-based sensor, but can also be obtained from any external source. Objects can also be determined, for example, from images, but can also be informed of the object by other vehicles, for example via vehicle-to-vehicle (V2V) communication.

[0010] It has been recognized that in this way, events that were originally only assigned to a place and time can be given a semantic meaning, i.e., from which object the event originated. This is important information for downstream evaluations, which are intended, for example, to find out whether a specific object in the vehicle's surroundings is on a collision course with the vehicle itself and whether the vehicle's behavior should be changed accordingly. The significantly faster availability of events provided by event-based sensors compared to camera images therefore results in a significantly less loss of semantic information than previously possible.

[0011] In particular, it has been recognized that events originating from one and the same object often satisfy a distribution with certain characteristic properties or characteristic variables. Different objects thus impose, as it were, a "fingerprint" on the event stream originating from them. This makes it possible to separate events belonging to different objects from one another even if the corresponding "event clouds" are very close to one another in space and time or even overlap.

[0012] Such an overlap can occur, for example, if a street scene is observed as a two-dimensional projection using a two-dimensional event-based sensor. For example, if two vehicles are driving side by side in each driving direction on a road with two lanes and this is observed from the side, the "event clouds" from these two vehicles overlap even if the vehicles themselves do not touch. However, in particular in cases where two different types of vehicles are involved, the "event clouds" can also be described by different distributions. Therefore, the event clouds can be separated from each other.

[0013] This separation of the "event cloud" is possible in particular even without exact knowledge of the object type in question. Therefore, the assessment of which objects in the scene move in what way does not presuppose an initial semantic segmentation of the scene according to object type.

[0014] The assignment of new events to detected objects does not necessarily presuppose the described path via distribution, but can also advantageously be carried out directly. For example, the following information can be extracted from one or more images of the observed area: which objects there were exactly where at the time of the image capture. Therefore, from one or more such momentary photographs, a spatial-temporal "expected channel" can be described, in which events originating from the objects are expected to be located. Events located outside this "expected channel" are most likely not originated from the corresponding object. For example, in order to determine the "expected channel", in particular, object types that are classified or otherwise identified according to the image can also be considered. The regularities on which different types of objects move are partially significantly different. Therefore, for example, different types of vehicles have turning circles of different sizes, which determine the flexibility with respect to changes in direction.

[0015] It has further been recognized that, for the final application of identifying or tracking objects, precisely those events which cannot be assigned to the detected objects and / or cannot be assigned to the detected distributions may be particularly important. These events can be directly identified by the evaluation provided within the scope of the method and can be prioritized accordingly during further processing in order to shorten the reaction time even further.

[0016] For example, if a vehicle is driven through road traffic, the vast majority of events provided by event-based sensors are relatively unsurprising. For example, if the vehicle is driven through a tree-lined avenue, a large number of events are constantly generated due to the constant alternation between trees on the one hand and gaps on the other hand. Many events that are traced back to the movements of other road users are also predictable from the background of the traffic situation and are not unexpected. Accordingly, such events do not require a quick reaction either. It is therefore relatively unsurprising that a vehicle that has started a turning maneuver with its right lights flashing continues with the turning maneuver.

[0017] On the contrary, a quick reaction to unexpected events may be very important. For example, a pedestrian intending to enter the roadway may suddenly appear at the edge of the viewing area, or a ball that suddenly flies into the viewing area and onto the roadway may warn that a child is about to run into the roadway. Such an event may justify full braking, and in such situations it is very beneficial that processing power is not occupied by predictable, routine events.

[0018] In the context of fixed-position monitoring of an area with regard to intruders or unauthorized activities, the distinction between events originating from known objects on the one hand and events originating from unknown objects on the other hand can alleviate the conflict of objectives between alarm sensitivity and the risk of false alarms. Thus, for example, in the event of strong winds, the treetops in the monitored area can generate many events that could be misinterpreted as the movement of an intruder and could trigger a false alarm. If only the sensitivity of the monitoring system with regard to movement were now reduced, this could result in an intruder being able to sneak through the monitored area and remain below the alarm threshold. If, on the other hand, only events originating from known treetops were blocked, the full sensitivity would still be available for the identification of intruders as before.

[0019] In a particularly advantageous embodiment, in response to the assignment of a new event to a detected object and / or the distribution of events belonging to a detected object, an updated position, an updated speed and / or an updated direction of movement of the object is determined taking into account the new event. In this way, the position, speed and / or direction of movement can be updated with particularly short reaction times and sudden changes can be detected quickly.

[0020] Thus, for example, the vehicle ahead of you, which has already started a right turn, is obliged to wait for the road user who wants to continue straight ahead. A dangerous situation can then arise if the following vehicle does not see the road user for whom the turning vehicle must wait. From the perspective of the following vehicle, the stopping of the turning vehicle is completely unexpected and requires rapid braking to avoid a rear-end collision.

[0021] In another particularly advantageous embodiment, in response to determining that the set of events does not match the distribution of already detected events, but forms a new distribution according to a predefined criterion, the new distribution is detected. In this way, it can at least be determined that an object has newly appeared in the observed scene or that an object has begun to move. This is primarily the most important information, especially in avoiding collisions in road traffic; what is less important for the time being is exactly what object is involved.

[0022] In another particularly advantageous embodiment, in response to the detection of a new distribution and / or a set of events that do not match the detected distribution meeting other predetermined criteria, at least one image of an object present in the observed area is determined. The predetermined criteria can, for example, specify that a minimum number of events that cannot be assigned must exist before initiating the determination of an image. As explained above, the image can be used in particular to assign an "event cloud" to a specific object in the observed scene. In addition, one or more images can be used to set an "expected channel" for new events provided by the event-based sensor.

[0023] For example, the image can be detected using an image sensor which is subdivided into light-sensitive pixels and which outputs, for each pixel, a measure of the intensity of light incident on the pixel. By calling up images from the image sensor only when needed, rather than constantly, the capacity available for data transmission can be kept free for most of the time for events that require a quick reaction.

[0024] However, alternatively or in combination therewith, the image can also be determined, for example, taking into account detected events. Ultimately, each event recorded by an event-based sensor can be used as an update of the instantaneous state represented by the previously recorded image. This means that an image can be generated based on the previous image and the events that occurred during this time, which image represents the instantaneous state at a later point in time. The later point in time can be freely selected and is not constrained in particular by a grid that is predetermined by a fixed repetition rate.

[0025] It is also possible to construct an image solely from the detected events. For example, all events occurring in a certain spatial and temporal window can be summed up for this purpose. Even if this only results in a rough outline of the objects, it is already possible to identify these objects from the images obtained in this way.

[0026] Furthermore, it is particularly simple to selectively determine the image for a sub-region of the observed area taking into account events that occurred during this period. In particular, this can be a sub-region from which the event that led to the determination of the image originated. As explained above, in particular in road traffic and in the case of fixed monitoring of an area for safety purposes, a large part of the scene consists of routine information. By querying the image only for sub-regions that are expected to contain new and unexpected information, transmission bandwidth and computing power are saved.

[0027] The image can be used in particular, for example, to assign new distributions and / or event sets that do not match the detected distribution to at least one object. Thus, as the object is present or moves, an increasing portion of the total events generated can be described incrementally.

[0028] In another particularly advantageous embodiment, the position, speed and / or direction of movement of at least one object on which these events are based is determined from a set of events that form a new distribution and / or from a set of events that do not match the detected distribution and that meet other predetermined criteria. It is not necessary to know what type of object it is. As explained above, the object can therefore initially be considered as an abstraction, from which only where the object is and / or how it moves is initially known. As explained above, the object can be concretized at a later point in time, for example based on an image.

[0029] As explained above, in many technical applications, the identification and / or tracking of objects is an intermediate product. The effect that is ultimately achieved in the application is that the corresponding technical system is controlled in a manner suitable for the respective situation using this intermediate product. Therefore, in another particularly advantageous embodiment, the control signal is determined based on at least one assignment of an event to an object and / or to a distribution and / or based on at least one position, movement direction and / or speed of at least one object determined based on the event. The control signal is used to control a vehicle and / or a monitoring system for monitoring the spatial area.

[0030] For example, in response to determining that a predicted trajectory of any type of object in space and time intersects with the current or planned trajectory of the vehicle, the trajectory of the vehicle can be modified by intervening in the vehicle's driving dynamics system so that the trajectory of the vehicle no longer intersects with the predicted trajectory of the object.

[0031] For example, in response to determining that an intruder is moving in an area monitored for security purposes or that other unauthorized activity is occurring in the area, a perceptible alarm and / or a silent alarm may be triggered, and / or other technical measures may be triggered to prevent the unauthorized activity. For example, if an intruder or unauthorized activity is determined on the premises of a car shop, the anti-theft locks of all vehicles parked there may be locked, for example, so that the vehicles can no longer be started with keys stored in the car shop.

[0032] As explained above, the distinction between events that should be assigned to known objects or distributions on the one hand and unexpected events that should not be assigned to known objects or distributions on the other hand can provide meaningful event prioritization for applications in which rapid reactions are involved. Therefore, the present invention also relates to a system for operating a vehicle and / or for operating a monitoring system for monitoring a spatial area, which system is designed for such prioritization.

[0033] The system includes at least one event-based sensor and at least one evaluation module. The evaluation module is configured to determine the following evaluation for an event output by the event-based sensor: to what extent the event matches the distribution of detected events; and / or to what extent it is reliable that the event originates from the detected object.

[0034] At least one control module is also provided. The control module is configured to form a control signal based on at least one assignment of an event to an object and / or to a distribution and / or based on at least one position, movement direction and / or speed of at least one object determined based on the event, and to control the vehicle or the monitoring system using the control signal.

[0035] An interface module is also provided, which is designed to transmit events output by the event-based sensor to the control module. The interface module is designed to prioritize the event higher when it is transmitted to the control module, the less the event matches the detected distribution or the detected object according to the evaluation by the evaluation module.

[0036] This means that the more "unknown" the event is with respect to detected (=known) events and distribution, the higher the priority with which the event is passed to the control module. As explained above, this has the effect that the control module can then react particularly quickly to unexpected events.

[0037] A module is understood to be any self-contained functional unit which is designed to perform a specific task (e.g. determine an evaluation or control signal or transmit the event) with the aid of a specific type of input (e.g. an event). Such a module can be implemented in the form of hardware (e.g. in the form of a computing unit located on a chip), in the form of a software routine or by any desired interaction of hardware and software.

[0038] In a particularly advantageous embodiment, the control module is additionally designed to communicate the detected distributions and objects to the evaluation module. The evaluation module is then constantly up to date on which events are still considered unexpected with regard to the respective technical application and which events have become routine during the operation of the system.

[0039] In another particularly advantageous embodiment, the interface module is connected to the control module via a CAN bus of the vehicle. Since the CAN bus typically extends over the entire vehicle, a plurality of event-based sensors which detect different sub-areas of the vehicle environment can be connected particularly easily in this way. This has the effect that the available transmission bandwidth is mainly used to transmit unexpected events. Even a high-speed CAN bus has a bandwidth of only 1 Mbit / s, and the described system must share this bandwidth with the large number of systems that are already present in the vehicle in any case for event-based control of the vehicle.

[0040] In particular, the method can be implemented in whole or in part by a computer. Therefore, the invention also relates to a computer program with machine-readable instructions, which, when executed on one or more computers, cause the one or more computers to perform one of the described methods. In this sense, control devices for vehicles and embedded systems for technical devices that are also capable of executing machine-readable instructions should also be regarded as computers.

[0041] Likewise, the invention relates to a machine-readable data carrier and / or a download product with a computer program. A download product is a digital product that can be transmitted via a data network, i.e. can be downloaded by a user of the data network, and can be sold for immediate downloading, for example in an online store.

[0042] Furthermore, a computer can be equipped with the computer program, the machine-readable data carrier or the download product. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Further measures for improving the invention are shown in more detail below together with the description of preferred exemplary embodiments of the invention based on the drawings.

[0044] Figure 1 An embodiment of a method 100 is shown;

[0045] Figure 2 An exemplary distribution 2 , 2 ′ of events 1 b originating from two different objects 3 , 3 ′ is shown;

[0046] Figure 3 An embodiment of a system 10 for operating a vehicle 50 and / or operating a monitoring system 60 is shown. DETAILED DESCRIPTION

[0047] Figure 1is a flow chart of an embodiment of the method 100. In step 110, a spatial region 40 is observed using at least one event-based sensor 1. The event-based sensor 1 outputs an event 1b. In step 120, an evaluation 1c is provided for each new event 1b. The evaluation 1c is a measure of how well the event 1b matches the detected distribution 2 of events 1b and / or how confident it is that the event 1b originated from a detected object 3.

[0048] In response to the determined evaluation 1c satisfying the predetermined criterion 1c, a new event 1b is assigned to the detected distribution 2 or the detected object 3. Then, in step 140, an updated position 3a, an updated speed 3b and / or an updated movement direction 3c of the object 3 are determined taking into account the new event 1b.

[0049] Not all new events 1b output by an event-based sensor 1 have to be assigned to already detected objects 2 or 3. For example, completely new objects 3 may suddenly enter the detection area of ​​the sensor 1 at any time.

[0050] If necessary, it can be determined that the set of events 1 b does not match the detected distribution 2 of events 1 b , but forms a new distribution (truth value 1 ) in accordance with the predefined criterion 150 . This new distribution 2 is then detected in step 160 .

[0051] This or the determination that a set of events 1b that do not match the detected distribution 2 meets a further predetermined criterion 170 (truth value 1) can lead to the determination of at least one image 4 of an object 3 present in the observed area 40 in step 180. For example, if there is a certain number of events 1b that cannot be assigned, this can trigger the need to clarify which object 3 these events 1b originate from.

[0052] The image 4 may be obtained from any source. For example, the image 4 may be detected according to block 181 using an image sensor which is subdivided into light-sensitive pixels and outputs for each pixel a measure of the intensity of light incident on the pixel. Alternatively or in combination therewith, the image 4 may be determined according to block 182 taking into account the detected event 1 b, for example as an update of a previously captured image 4.

[0053] In particular, the image 4 can be determined selectively, for example according to block 183 , for a subregion of the observed region 40 from which the event 1 b resulting in the determination of the image 4 originates.

[0054] Based on the image 4 , a new distribution 2 and / or a set of events 1 b which do not match the already detected distribution 2 can be assigned to the at least one object 3 in a step 190 .

[0055] As long as the initially unassignable events 1b form a new distribution 2 and / or satisfy other predetermined criteria 200 (truth value 1), the position 3a, speed 3b and / or movement direction 3c of at least one object 3 on which these events 1b are based can be determined from these events 1b in step 210. As explained above, this object 3 can be regarded as an abstract thing, about which nothing is known except its position 3a and its kinematics 3b, 3c.

[0056] Any information obtained during the method 100, such as the assignment of events 1b to distributions 2 or to objects 3 as well as the position 3a, speed 3b and / or direction of movement 3c of objects 3, can be used in step 220 to determine the control signal 5. The control signal 5 can be obtained in particular in such a way that, when the control signal is used to control a vehicle 50 or another system 60 for monitoring a spatial region 40, the vehicle 50 or the system 60 reacts appropriately to the scene detected in the region 40 by means of the event-based sensor 1. In step 230, the control signal 5 can be used to control the vehicle 50 or the system 60.

[0057] Figure 2 It is shown by way of example how an event 1 b can be assigned to a distribution 2 , 2 ′ and thus ultimately to an object 3 , and how this assignment is facilitated by recording an image 4 .

[0058] Figure 2 The exemplary scene depicted in comprises a moving vehicle as object 3 and a similarly moving ball as object 3'. Figure 2 The images 4 and 4' shown in the figure represent instantaneous photography of the scene at different time points. At the time of taking the first image 4, the vehicle 3 and the ball 3' are respectively located at the position 3a. At the time of taking the second image 4', the vehicle 3 and the ball 3' are respectively located at the new position 3a'.

[0059] During the movement from position 3a to position 3a', objects 3 and 3' cause the event-based sensor 1 to output a large number of events 1b, respectively. Here, the events 1b originating from the vehicle 3 form a first distribution 2, which is characterized by a certain property. The events originating from the ball 3' form a second distribution 2', which is characterized by other properties. Therefore, when new events 1b are recorded, it is possible to check, for example using any statistical test, whether these events 1b match the first distribution 2, the second distribution 2' or neither of these two distributions 2 and 2'.

[0060] Independently of this, the positions 3a and 3a' at which the objects 3, 3' are located in the images 4 and 4', respectively, define in association with the points in time at which the images 4 and 4' were recorded, respectively, a spatial and temporal "expected channel" in which events originating from these objects 3, 3' should lie.

[0061] Figure 3 An embodiment of a system 10 for operating a vehicle 50 and / or operating a monitoring system 60 for monitoring a spatial area 40 is shown.

[0062] The system 10 comprises an event-based sensor 1 for observing an area 40. The sensor 1 is subdivided into light-sensitive pixels 1a and, as explained above, outputs an event 1b respectively assigned to the pixel 1a when the intensity of light incident on the pixel 1a changes.

[0063] The evaluation module 11 determines an evaluation 1c for each new event 1b as follows: to what extent these events 1b match the detected distribution 2 and / or objects 3. The event 1b is forwarded with its evaluation 1c via the interface module 12 to the control module 13. The control module 13 forms the control signal 5 and thus controls the vehicle 50 or the monitoring system 60.

[0064] The interface module 12 prioritizes the forwarding of the event 1b to the control module 13 based on the evaluation 1c. The more unexpected and unexpected the event 1b is, in view of the detected distribution 2 or object 3, the greater the probability that a quick reaction must be made to the event, and the higher the priority when forwarding it to the control module 13. This is particularly advantageous when the interface module 12 communicates with the control module 13 via a connection with limited bandwidth, for example via the CAN bus 51 of the vehicle 50 or via a radio connection. Thus, for example, for certain short-range radio connections in the 868 MHz frequency band, there are restrictions with regard to the proportion of the transmission time (duty cycle) per time unit (for example, per hour). If, at the point in time when a new unexpected object 3 appears in an unexpected event 1b, the transmission time quota has been exhausted for the event 1b originating from a known object 3, the unexpected event 1b must be retained until a new transmission time quota begins.

[0065] Information about which distributions 2 or objects 3 are known is fed back from the control module 13 to the evaluation module 11. Thus, the proportion of unexpected events 1b during operation tends to be reduced. For example, in particular the CAN bus 51 of the vehicle 50 can also be used for the feedback.

[0066] exist Figure 3 In the example shown, there are also further event-based sensors 1' with associated evaluation modules 11' and interface modules 12'. The further sensors 1' and further modules 11' and 12' function as described above, but observe further areas 40' (eg, further sub-areas of the vehicle environment).

Claims

1. A method (100) for identifying and / or tracking an object (3) in a spatial region (40), the method comprising the following steps: observing (110) the area (40) using at least one event-based sensor (1), wherein the event-based sensor (1) has light-sensitive pixels (1a), and wherein a relative change of at least a predetermined percentage in the intensity of light incident on a pixel (1a) causes the sensor to output an event (1b) assigned to the pixel (1a); In response to the sensor (1) outputting a new event (1b), determining (120) an evaluation (1c) for the event (1b), the evaluation being a measure of how well the event (1b) matches a distribution (2) of already detected events (1b) and / or how confident it is that the event (1b) originated from an already detected object (3); In response to the determined evaluation ( 1c ) satisfying a predetermined criterion ( 125 ), the new event ( 1b ) is assigned ( 130 ) to the detected distribution ( 2 ) or the detected object ( 3 ).

2. The method (100) according to claim 1, wherein: In response to assigning (130) a new event (1b) to a detected object (3) and / or a distribution (2) of events (1b) belonging to the detected object (3), an updated position (3a), an updated speed (3b) and / or an updated direction of movement (3c) of the object (3) are determined (140) taking into account the new event (1b).

3. The method (100) according to any one of claims 1 to 2, wherein: In response to determining that the set of events (1b) does not match the distribution (2) of already detected events (1b) but forms a new distribution (2) according to a predetermined criterion (150), the new distribution (2) is detected (160).

4. The method (100) according to any one of claims 1 to 3, wherein: In response to detecting (160) a new distribution (2) and / or a set of events (1b) that do not match an already detected distribution (2) meeting other predetermined criteria (170), at least one image (4) of an object (3) present in the observed area (40) is determined (180).

5. The method (100) according to claim 4, wherein: The image (4) is detected (181) using an image sensor subdivided into light-sensitive pixels, the image sensor outputting, for each pixel, a measure of the intensity of light incident on that pixel.

6. The method (100) according to any one of claims 4 to 5, wherein: The image (4) is determined (182) taking into account the detected event (1b).

7. The method (100) according to any one of claims 4 to 6, wherein: The image (4) is selectively determined (183) for a sub-region of the observed region (40) in which the event (1b) leading to the determination of the image (4) originates.

8. The method (100) according to any one of claims 4 to 7, wherein: The new distribution (2) and / or a set of events (1b) that do not match the detected distribution (2) are assigned (190) to at least one object (3) based on the image (4).

9. The method (100) according to any one of claims 1 to 8, wherein: The position (3a), speed (3b) and / or movement direction (3c) of at least one object (3) on which the event (1b) is based is determined (210) based on a set of events (1b) that form a new distribution (2) and / or based on a set of events (1b) that do not match the detected distribution (2) and meet other predetermined criteria (200).

10. The method (100) according to any one of claims 1 to 9, wherein: A control signal (5) is determined (220) based on at least one assignment of an event (1b) to an object (3) and / or to a distribution (2) and / or based on at least one position (3a), movement direction (3b) and / or speed (3c) of at least one object (3) determined based on the event (1b), and wherein the control signal (5) is used to control (230) a vehicle (50) and / or a monitoring system (60) for monitoring the spatial area (40).

11. A system (10) for operating a vehicle (50) and / or operating a monitoring system (60) for monitoring a spatial area (40), comprising: at least one event-based sensor (1), at least one evaluation module (11) which is designed to determine, for an event (1b) output by an event-based sensor (1), the following evaluation (1c): to what extent the event (1b) matches the distribution (2) of already detected events (1b); and / or to what extent it is plausible that the event (1b) originates from a detected object (3), at least one control module (13) which is designed to generate a control signal (5) as a function of at least one assignment of an event (1b) to an object (3) and / or to a distribution (2) and / or as a function of at least one position (3a), movement direction (3b) and / or speed (3c) of at least one object (3) determined on the basis of the event (1b), and to control the vehicle (50) or the monitoring system (60) using the control signal (5), and at least one interface module (12) which is designed to transmit an event (1b) output by the event-based sensor (1) to the control module (13), The interface module (12) is designed so that the less an event (1b) matches a detected distribution (2) or a detected object (3) according to the evaluation (1c) performed by the evaluation module (11), the higher the priority of the event (1b) is determined when it is transmitted to the control module (13).

12. The system (10) according to claim 11, wherein: The control module (13) is additionally designed to communicate the detected distribution (2) and the object (3) to the evaluation module (11).

13. The system (10) according to any one of claims 11 to 12, wherein: The interface module (12) is connected to the control module (13) via the CAN bus (51) of the vehicle (50).

14. A computer program product comprising machine-readable instructions which, when executed on one or more computers, cause the one or more computers to perform the method (100) according to any one of claims 1 to 10.

15. A machine-readable storage medium having stored thereon a computer program product according to claim 14.

16. A computer provided with the computer program product according to claim 14 and / or the machine-readable storage medium according to claim 15.

Images