Method and device for efficient object tracking and correspondingly designed motor vehicle

Abstract

The invention relates to a method (1) for object tracking and an object tracking device (13). The invention also relates to a correspondingly designed motor vehicle (13). In the method (1), a motion state of an object (17) detected by means of sensor data (3) of a sensor system (12) is determined, and the object is classified as a moving object or a stationary object (17) according to the motion state. Object data (2) is stored for the objects (17), said object data comprising a plurality of properties of the respective object (17) and the motion state determined for this purpose. The object data (2) of the objects (17) that have been detected in the previous measurement cycle is updated by means of the current sensor data (3). In this case, the properties of the objects stored in the object data (2) are updated independently of each other only for those objects (17) classified as stationary objects (17).

Description

Technical Field

[0001] The invention relates to a method for tracking an object based on sensor data, and also to a corresponding device for the method and a motor vehicle equipped with the device. Background Art

[0002] In various technical applications and fields, for example but not only in the field of vehicles and traffic, sensor detection and tracking, i.e. tracking objects in the corresponding environment, can be useful, for example for collision avoidance, for trajectory planning, for monitoring, etc. However, the sensors and measurement methods used here, especially in combination with automatic object recognition, are not always completely accurate and precise, which may be undesirable or even safety-related depending on the application. The adaptive cruise control (ACC) of a motor vehicle, for example, requires that the vehicle traveling ahead be identified as accurately as possible and its position be determined as accurately as possible. In particular, if the vehicle traveling ahead is currently stationary, the distance between the own vehicle and the vehicle traveling ahead should not be overestimated. The result of this overestimation may be an unintentional automatic restart of the own vehicle, which in turn may lead to a drop below the prescribed minimum distance or may lead to a collision.

[0003] For example, DE 102020 123 585A1 describes an extended object tracking method implemented by radar. The method therein comprises generating one or more clusters based on the received return points of the radar sensor of the vehicle. Then, an estimated position and speed are calculated for each of the clusters. In addition, it is determined whether the cluster is associated with an existing object track. Based on the one or more clusters associated with the existing object track, the existing object track is updated using at least the corresponding position of the one or more clusters.

[0004] For effective object tracking, the state of the corresponding object can usually be determined in some form. To this end, DE102019 109 332A1, for example, describes a method for obtaining the state of an object at a series of time points. Here, an occupancy grid is obtained based on sensor data, and the occupancy grid indicates evidence for each of a plurality of cells, namely that the corresponding cell is idle or occupied by an object at a time point. Based on this, a subset of cells belonging to the object at this time point is obtained. Finally, the object state at this time point is obtained based on the object state predicted for this time point and based on the obtained subset of the object's cells. Therefore, the object can be detected in an accurate and robust manner, and information related to the object can be obtained. Summary of the invention

[0005] The object of the present invention is to enable particularly efficient sensor-based object tracking.

[0006] This object is achieved by the subject matter of the independent claims. Further possible embodiments of the invention are disclosed in the dependent claims, the description and the drawings. The features, advantages and possible embodiments described within the scope of the description for one of the subject matter of the independent claims should be regarded at least similarly as the features, advantages and possible embodiments of the corresponding subject matter of the other independent claims, as well as any possible combination of an independent claim with the subject matter of one or more dependent claims, if necessary.

[0007] The method according to the invention can be used for object tracking, i.e. for object tracking in multiple measurement cycles of a sensor system for environment or object detection. In particular, the method can be applied in a motor vehicle, but is not limited to this application. In the method steps of the method according to the invention, objects are detected by means of sensor data detected in the measurement cycle of the sensor system. In other words, the current sensor data of the sensor system can be detected in each measurement cycle, and evaluated or processed for object detection or object recognition. The sensor system can include one or more individual sensors, wherein multiple such individual sensors of the sensor system can be different types in particular. Thus, for example, according to the measurement or recording frequency or the synchronization of multiple individual sensors, the sensor data of one sensor or multiple sensors can be detected in one measurement cycle. The sensor system or the individual sensors included therein can even detect and track objects internally. The corresponding objects detected and tracked, i.e. managed, inside the sensor can be referred to as sensor objects. Similarly, the sensor data provided by the sensor system can be processed, for example, by an object tracking or tracking device designed to perform or apply the method according to the invention or coupled to the sensor system, such as a correspondingly designed control unit or auxiliary system, etc., to detect objects outside the sensor system.

[0008] In another method step of the method according to the invention, the motion state of the object is determined for the detected object by means of sensor data. Here, each detected object is classified as a stationary object or a moving, i.e. self-moving object. Therefore, the motion state of the object can indicate whether the object is in motion or in stillness. For example, a stationary object can be a stationary object that can move in principle or an immobile object, which is fixed according to the natural position and / or at least never detected in the scope of the respective current use method. For example, the motion state of the detected object can be determined relative to a preset global coordinate system or relative to a sensor system or a device equipped with a sensor system (such as a motor vehicle, etc.). In order to determine the motion state, the sensor data or the data or parameters derived therefrom can be evaluated, for example, in view of at least one preset criterion. Here, for example, a speed threshold value or statistical data can be evaluated by multiple detections of the corresponding object in continuous measurement cycles and / or by multiple individual sensor detections assigned to the corresponding object, especially from the respective current measurement cycles. In the current sense, such individual sensor detections can be individual data points or measurement points, such as individual radar echoes or radar detections or laser radar detections or radar reflections, etc.

[0009] In addition, in the method according to the invention, object data related to each object detected by means of sensor data are stored, the object data comprising a plurality of characteristics of the corresponding object and its specific motion state. Therefore, the object data can be an object data set describing or characterizing the corresponding object, which is also referred to as an object trajectory. Therefore, the detected objects represented in the sensor data or by these sensor data can be respectively assigned to such object data sets or object trajectories, in which the corresponding characteristics of the objects are stored or managed. Therefore, the object data are respectively object-specific, so that the corresponding object trajectories can be provided or managed for each detected object. Therefore, objects or their characteristics can be tracked or managed in the object data within multiple measurement cycles. New object trajectories can be provided for the objects newly detected in the current measurement cycle, for which no object trajectory with characteristics determined by means of sensor data from at least one previous measurement cycle is provided or exists.

[0010] The characteristics stored or described in the object data or object trajectory can be or describe the static characteristics and / or dynamic states or state data of the corresponding object. Static characteristics in this sense can be, for example, or describe the type or kind of the corresponding object. Dynamic states or state data can be or describe (for example, from measurement cycle to measurement cycle) at least possible data or parameters of the corresponding object, such as position and / or speed and / or acceleration and / or yaw angle and / or yaw rate, etc. Object trajectory can also be called internal trajectory, because the object trajectory can be managed, for example, inside the sensor system or inside the sensor or inside the object tracking device close to the sensor along the corresponding signal or data processing pipeline. The motion state stored in the sensor object or as part of the sensor object and / or in the corresponding object data can be derived directly from the motion speed of the object, and / or estimated by the sensor system or the corresponding separate sensor, and / or determined or estimated, for example, by the object tracking device according to the sensor data from the sensor system or the sensor object or the object data. Here, if available, the data of multiple sensors or the estimated motion state can be combined. The speed of movement of the object can be measured directly by means of a sensor, for example, or estimated by means of available sensor data, for example over a plurality of measurement cycles, for example by means of the change in the sensor data over a plurality of measurement cycles.

[0011] Stationary objects can be identified, for example, directly based on the corresponding motion state or object state, i.e., data or properties of the object, which have been determined, for example, by a sensor and / or with the aid of sensor data from one or more sensors. An object can be identified or classified, for example, as a stationary object when a speed below a preset speed threshold is indicated in the complete object data of the object. As such a speed threshold, for example, 0.5 m / s can be preset, but other values ​​are also possible depending on the application or requirements. In contrast to individual sensor detections or sensor raw data assigned to an object, complete object data can be understood here, for example. Object data can be or include data or properties of the corresponding object determined or derived from such sensor detections or sensor raw data.

[0012] Depending on the design of the sensor system, within the sensor or sensor system, by detecting one or more sensors of a specific object, the motion state of the object can be determined or classified for the detected object, and / or at least some of the static and / or dynamic properties can be determined or estimated. This can also be used as a basis for classifying the object as a stationary object or a moving object if necessary. If the sensor system includes a plurality of, in particular, different sensors, then the object can be classified, for example, as a stationary object or a moving object when the motion states determined internally by the plurality of sensors for the object are correspondingly consistent. This enables a particularly fast and at the same time particularly reliable classification. In addition, this classification can also be particularly flexible and robust, because, for example, the same sensor does not always have to be used. This can be particularly useful, for example, if the sensor cannot detect the object during this period, for example due to coverage, etc.

[0013] In a further method step of the method according to the invention, object data of objects that have already been detected in a previous measurement cycle are automatically updated with the aid of current sensor data, i.e., detected in the current measurement cycle, or at least the dynamic states contained in the object data are automatically updated. It can thus be ensured that current data or information about the respectively assigned object are respectively stored in the object data, so that the stored object data take into account the respective current measurement or sensor data. The updating of the object data provided here can be performed at least for existing objects, i.e., already detected in one or more previous measurement cycles. For objects that are newly contained in the current sensor data or newly detected, i.e., detected for the first time, a new object trajectory can be provided in each case, in which the current sensor data are also taken into account.

[0014] According to the invention, only for those objects that are classified as stationary objects, the characteristics of the objects stored in the object data are updated independently of each other. In other words, for stationary objects, their characteristics or at least their dynamic states or state data are decoupled from each other, that is, for example, updated without mutual correlation or mutual influence or correlation. This update process can also be called a stationary update because it is applied to stationary objects. If the stored characteristics include, for example, the position and speed of the object, then for stationary objects, the updated position changed relative to the previous object data has no effect on the specific or stored speed. Therefore, by updating the characteristics, such as position and speed, independently of each other, the new position changed compared to the position previously stored in the object data does not automatically lead to a changed or non-zero speed of the corresponding object according to the current sensor data. Therefore, for objects classified as stationary, the speed of the object can be automatically kept at zero, for example, unless, for example, an independent speed measurement indicates a speed of the corresponding object that is different from zero or higher than a preset threshold. Here, any corresponding dynamics can be prohibited during the update for stationary, i.e., stopped objects.

[0015] As a result, a correspondingly stable tracking of stationary, i.e., stationary objects with respect to dynamic, i.e., potentially variable properties is thus produced. This represents an advantage of the method according to the invention over conventional methods. Such conventional methods usually use data fusion methods for object tracking and updating the corresponding object data, which are applied to all detected objects. Therefore, in conventional methods, there are no separate logic or data processing methods for stationary objects and moving objects, but rather a comprehensive detection of the corresponding objects is always performed and a correspondingly comprehensive processing of the corresponding sensor or object data is performed without adaptation to the actual physical properties or physically possible properties of the corresponding physical real object. Therefore, as long as the vehicle is stationary, the stationary vehicle, for example, cannot actually change its orientation or does not establish a yaw rate. In addition, measurement uncertainties or changes in the sensor data, for example over a plurality of measurement cycles, do not indicate the actual speed of the object for an object that is stationary during the entire time.

[0016] For example, conventional methods apply a preset model to all measuring points or sensor data of an object, so as to finally determine, for example, a bounding box of the corresponding object and thus update the corresponding object data. If such a bounding box or bounding frame is newly determined for the object in each measuring cycle, then the bounding box or bounding frame is not always stable from measuring cycle to measuring cycle, i.e., not always constant, although the corresponding detected object is actually not moving. Therefore, in these cases, the (actually non-existent) movement of the corresponding object can still be detected incorrectly. For example, in correspondingly high traffic density or in an urban environment, etc., stationary measuring points may usually appear in the vicinity of the detected external vehicle, for example, due to sensor detection assigned to the external vehicle, which may actually come from the traffic infrastructure, such as the ground or curbstone, etc., or from the exhaust gas tail (Abgasfahne), etc. Such measuring points or sensor detections can be temporarily detected or detectable when the external vehicle, i.e., the corresponding detected object is not moving, and are incorrectly assigned to the corresponding object or integrated into the bounding box determined for the object. This may lead to the incorrect detection of speed or rotation or movement for stationary objects. Such an erroneous detection or erroneous estimation can in turn lead to a corresponding erroneous reaction of the automation system and thus to accidents or unstable behavior.

[0017] The invention solves this problem by operating sensor data or object data differently for objects classified as stationary and for moving objects, wherein the object data is updated for stationary objects so that erroneous recognition of moving or dynamically stationary objects that are not actually present is avoided or reduced.

[0018] For the remaining, i.e. moving objects, the updating of sensor data can be performed in other ways or in a conventional way, for example, based on sensor objects that have been tracked by or in the sensor system and output by the sensor system, and the fusion based on filters with the corresponding existing, i.e. previous object data can be performed. Here, for example, a Kalman filter or the like can be used, and the correlation or cross-correlation between different characteristics of the corresponding object can be considered. However, other data filters or fusion mechanisms can also be applied, especially if it effectively acts as a low-pass filter. The current sensor data and the data or characteristics of the corresponding object that have been stored in the corresponding object data can be combined or fused with each other in a weighted manner through such a preset filter or fusion mechanism. This may especially contribute to the advantageously smooth and consistent tracking in the uniform movement of the detected object. For the detected object that is not classified as stationary, for example, the characteristics of the object or at least its dynamic state or state data can be updated in the case of the corresponding correlation between the corresponding current sensor data or the corresponding characteristics or parameters determined thereby and the previous object data. Therefore, it can be considered that these characteristics or dynamic states of the moving object can actually change dynamically in an unpredictable manner, but still change from measurement cycle to measurement cycle with limited dynamics or rate.

[0019] It may happen that the sensor system or at least one specific sensor, such as a camera or the like, which is at least part of the sensor system, does not provide detection or sensor data in the current measuring cycle for a specific object, in particular one that has already been detected in a previous measuring cycle. If the object has previously been classified as a stationary object, the described settings for updating stationary objects can also be carried out or applied for this object in the current measuring cycle.

[0020] The method according to the invention can be combined or extended with further measures or method steps. Before the object data are used as described, previous object data of an object that has been detected in advance can first be predicted or extrapolated to the current time point, for example, the recording time point of the sensor data of the current measurement cycle, for example, based on a predefined model.

[0021] In a possible embodiment of the present invention, at least for objects that have been classified as stationary in a previous measurement or update cycle, the corresponding motion state of the detected object is determined by means of radar detection, i.e., radar data or radar echoes. These radar detections can in particular be part of the sensor data detected respectively. Here, when the speed of the corresponding object determined based on radar, i.e., by means of the radar detection assigned to the corresponding object, is less than a preset speed threshold in absolute value, the object is classified as a stationary object. Such a speed threshold can, for example, follow the corresponding measurement uncertainty of the radar sensor used, i.e., take this measurement uncertainty into account. In other words, the radar detection, in particular the speed determined based on it, can be used for the corresponding stationary estimation.

[0022] This is based on the recognition that radar detection is particularly suitable for determining the state of motion due to the thus possible direct measurement of the double speed. In addition, a stable distance of the sensor system to the corresponding object can be estimated based on the radar detection within one or more measurement cycles using a predefined robust estimation method (which can be based on a median determination or a histogram, for example). This can avoid or reduce a potentially dangerous overestimation or underestimation of the distance, which can occur, for example, in camera-based distance estimation. Thus, for example, functions based on the position and / or speed and / or distance of the corresponding object, such as automatic distance keeping or adaptive cruise control, etc., can be implemented particularly safely and reliably.

[0023] If the absolute value of the speed of the corresponding object determined based on the radar is less than a preset speed threshold value, then, for example, a corresponding flag can be automatically set in the corresponding data processing, for example in the object data of the corresponding object, whereby the corresponding object is classified or defined as a stationary object. For stationary objects, the process noise in the corresponding signal or data processing can also be reduced. If, during the measurement cycle, the speed determined based on the radar for the corresponding object indicates a movement of the object, the process noise can, on the other hand, be automatically increased again. In this way, for example, by auxiliary systems or functions that use the object data as input, a faster reaction to the corresponding object starting to move can be achieved.

[0024] In another possible embodiment of the invention, an object is only classified as, in particular only as, a stationary object if at least one preset number of measurements and / or at least one preset share of a plurality of performed measurements results in or indicates a speed of the corresponding object that is below a preset speed threshold value mentioned in one or elsewhere. If a preset share of measurements is used as a criterion here, a preset minimum number of measurements, with the aid of which the share is determined or has been determined, can also be preset for this purpose before the share is used or evaluated as a criterion for classifying the state of motion of the corresponding object. In the embodiment of the invention proposed here, in each case, the state of motion of the corresponding object is determined not on the basis of a single speed measurement, but on the basis of a plurality of speed measurements. For example, the described dynamics in the object data can be inhibited when updating the object data for the object only if a plurality of Doppler measurements for the object in one or more measurement cycles indicate or confirm the complete stillness of the corresponding object.

[0025] As a corresponding criterion for classifying an object as a stationary object, it can be preset, for example, that at least five measurements (which have been performed continuously or within a preset time period) produce or indicate a speed of the corresponding object below a preset speed threshold. The speed threshold can be, for example, 0.1 m / s or 0.5 m / s, etc. However, other values ​​can also be preset according to the application and requirements. Regarding the proportion of the measurements performed, it can be preset as a criterion, for example, that among all measurements or sensor detections of the corresponding object within a preset time period of a preset length or within a sliding time window, a maximum of 10% or a maximum of 20% or a maximum of 30% can indicate a speed above the speed threshold, so that the object is still classified as a stationary object. Here, the corresponding given proportion can also be related to the corresponding application or the corresponding requirement or be sensor-specific. In general, the design scheme proposed here of the present invention can particularly reliably and robustly determine the corresponding motion state of the detected object and can accordingly robustly apply the method according to the present invention.

[0026] In another possible design of the present invention, for objects classified as stationary, the object data of the object is updated based on individual sensor detections, and conversely, for objects classified as moving, the object data of the object is updated based on sensor objects output by the sensor system. Such sensor objects can be output by the sensor system or the corresponding sensor, for example, in each measurement cycle, for example in the form of a list or the like. Here, the sensor object is an object that is tracked in multiple measurement cycles by the sensor system or the corresponding sensor itself, that is, detected internally. Therefore, the sensor object may have been formed or reconstructed by the sensor system or the corresponding sensor based on multiple sensor detections. Here, the sensor object may include or describe multiple characteristics of the corresponding basic real object. For example, the sensor object can describe the boundary frame and / or position and / or orientation and / or speed and / or yaw rate, etc. for the corresponding object as a whole (that is, not only for each sensor detection assigned to the object).

[0027] Based on such a sensor object, a corresponding update can be performed for objects classified as moving by applying a predefined filtering or fusion mechanism, which can also be referred to as a complete update here.

[0028] In contrast, a simplified update can be performed for objects classified as stationary. In addition to at least one sensor detection assigned to the respective object, properties of the respective object stored in the associated object data from one or more previous measuring cycles can also be used or taken into account. In comparison with sensor objects that are only used to update object data for objects classified as moving, individual sensor detections can optionally be processed more simply and / or more quickly, i.e., with less effort overall, for updating the respective object data.

[0029] Here, the sensor detections associated with objects classified as stationary can be determined based on object data, i.e., based on corresponding object trajectories, which are stored in particular outside the corresponding sensor itself, in particular based on the dynamic state of the corresponding object stored therein or based on the dynamic state predicted at the corresponding measurement time point on the basis of this. This is therefore possible in particular because the position and orientation of stationary objects cannot change according to definition, and therefore the sensor detections associated with the corresponding object can be selected from all available sensor detections by means of a preset position- or location-based criterion. As a result, the sensor detections that are actually associated with the stationary object can be selected particularly reliably and robustly and used to update the corresponding object data. Since the assignment of sensor detections to stationary objects is not performed by the corresponding sensor itself, the probability of an incorrect assignment of sensor detections to the corresponding object or to the last incorrect position of the corresponding object, which results from abnormal values, measurement errors, faults and / or drifts of the corresponding sensor, can be reduced in particular.

[0030] If the sensor system comprises a plurality of different sensors, or in the current measuring cycle, the corresponding object sensor detection is detected by different sensors, then these sensor detections can be taken into account in full or only in part, i.e. for the corresponding update. This may depend, for example, on the detected deviations or inconsistencies between the sensor detections of the different sensors and the preset priority lists of the different sensors. Such a priority list can indicate which sensor is usually preferred for stationary objects, or for assigning sensor detections to stationary objects, or for correspondingly determining or updating specific characteristics of stationary objects by other sensors or their sensor data. For example, for stationary objects, lidar detections and radar detections can exist as sensor detections, and / or the dynamic state stored in the existing object data, in particular the position of the corresponding stationary object, can be based on an older lidar detection. Since the radar detection of a stationary object may be less suitable, i.e. may have a lower priority, compared to the lidar detection, for example due to the drift tendency of the radar sensor, the radar detection to be assigned to the stationary object can be determined or selected based on the position determined by the lidar or based on the bounding frame determined by the lidar for the corresponding stationary object. This can be achieved in particular if the lidar detection from the current measuring cycle is also consistent with the known stationary state of the corresponding object at the position stored in the relevant object data. Thus, in the case of drift or inconsistency of radar detections, at least a portion of the radar detections that match the current position of the respective stationary object can optionally also be used. As a result, the sensor detections associated with the respective stationary object can be determined more accurately and more reliably overall, and if necessary, more sensor detections (i.e., more data) can be used or taken into account to update the object data accordingly.

[0031] In another possible embodiment of the present invention, for objects classified as stationary, a stable end detection is determined or selected from all current sensor detections that are assigned to or currently assigned to the corresponding object, in order to determine or set the current position of the corresponding object to be updated as a characteristic in the object data. The end detection can be one of the sensor detections assigned to the corresponding object, which is used as the end facing the sensor system (or, for example, a motor vehicle equipped with a sensor system, etc.) or as the corresponding side of the corresponding object, that is, interpreted or regarded as this end or this side. Here, the sensor detection is assigned to the corresponding object according to a preset criterion. The preset criterion can be based on distance in particular with respect to the previous position of the corresponding stationary object, that is, already known or obtained from an earlier measurement cycle and stored or specified in the corresponding object data. In other words, the criterion can, for example, query whether a specific sensor detection is located within a preset distance from the position of the corresponding stationary object specified in the object data. Only if this is the case, the corresponding sensor detection is assigned to the corresponding stationary object. The specific or selected stable end detection can, for example, be stably regarded or used as the end of the corresponding object or as a reference for determining the position of the corresponding object in multiple measurement cycles.

[0032] For stationary objects, the same sensor detection, i.e., for example, corresponding sensor detections from the same location of the corresponding object, can be used, for example, as stable end detections within a plurality of measurement cycles or as long as the corresponding object is classified as stationary or remains stationary. With the aid of such stable end detections, the end or position of the corresponding object can be determined in an unchanging (i.e., stably) manner. This is particularly possible if, for example, other current sensor detections appear to indicate other positions of the corresponding object.

[0033] Likewise, a stable end detection can be determined, for example, as the median or average value of the sensor detections of the sensor system respectively closest to the object over a plurality of measuring cycles. Thus, a stable end detection and thus the position of the corresponding stationary object can also remain stable or be determined if not all sensor detections associated with the corresponding object are detected constantly, in particular not in a positionally stable manner, over a plurality of measuring cycles.

[0034] The embodiment of the invention proposed here therefore makes it possible to track stationary objects particularly reliably, accurately, robustly and stably, even in the case of consistent or fluctuating sensor detections, and thus to react appropriately and robustly thereto.

[0035] In a possible development of the invention, when the sensor detection associated with the respective stationary object comprises a radar detection, a stable end detection is determined or selected taking into account the respective radar cross section (RCS). This can therefore be applied if the sensor system is or comprises a radar sensor. In particular, the radar detection closest to the sensor system can be determined as an end detection, whose radar cross section is greater than a preset threshold value. For this purpose, in particular over a plurality of measuring cycles, for example, a bar graph can be input in which the radar cross section is plotted as a function of the distance of the radar detection from the sensor system.

[0036] By taking into account the radar cross section or a threshold value preset therefor, the probability of taking into account scatter or clutter detections in the area of ​​the respective object for determining the position of the object, which scatter or clutter detections could distort the position and orientation of the bounding frame of the respective object, can be reduced.

[0037] The threshold value of the radar cross section can be preset according to the object type. Therefore, different threshold values ​​of the radar cross section can be considered for different types or categories of objects. As a result, the method can be robustly and reliably applied to different types or categories of objects.

[0038] For further improvement, it can be provided that in each measuring cycle, a plurality or all of the radar detections of the radar detections assigned to the respective object are taken into account, i.e., an evaluation is performed with respect to the radar cross section of the radar detections to determine or select the end detections. As a result, the position of the respective object can finally be determined particularly reliably and robustly. Since the radar detections can also inherently represent or include measurements or descriptions of the Doppler velocities, the speed of the respective object can also be determined or estimated based on the radar detections assigned to the object on the basis of these Doppler velocities. The speed thus estimated can then also be taken into account when updating the object data, for example, be entered or incorporated into the object data.

[0039] In another possible embodiment of the present invention, at least for objects classified as stationary, at least a portion of the characteristics stored or storable in the corresponding object data is determined based on a lidar detection. Here, the sensor system can therefore be a lidar sensor or include such a lidar sensor. Therefore, a lidar detection is a sensor detection of such a lidar sensor here. The lidar detection from the corresponding current measurement cycle is assigned to the boundary frame of the corresponding object here. For this assignment or association, for example, a preset position- or location-based criterion can be evaluated. Therefore, for example, only such lidar detections within a preset area near the position of the corresponding object described in the object data can be assigned to the boundary frame. For this purpose, for example, corresponding distances and / or angles and distance ranges, etc. can be preset. The distance of the lidar detection can also be determined and evaluated by the sensor system, for example, similarly to the description in conjunction with the radar detection elsewhere.

[0040] In addition, a stable end detection is determined here by means of the distribution of the lidar detections assigned to the boundary frame of the corresponding object, in particular a distance histogram, which serves as the end or the corresponding side of the corresponding object facing the sensor system (or, for example, facing a motor vehicle equipped with the sensor system, etc.), i.e. is interpreted or regarded as this end or this side. Therefore, based on the lidar, in particular the position of the object is determined as a property of the corresponding object. Therefore, the end detection or the position of the end of the corresponding object defined by the end detection is used here to determine the position of the object as a corresponding property of the object in the assigned object data. In particular, within a plurality of measuring cycles, the lidar detections can be entered into the distance histogram in a classified manner according to their distance, for example, to the sensor system or other predetermined reference points.

[0041] As an end detection, a laser radar detection can be determined or selected, which is located in a significant distance bin (Abstandsbin) with or for a minimum distance or distance range. When a preset threshold is exceeded by a distance bin or is exceeded in a distance bin, the distance bin of the distance histogram may be significant in the current sense. For example, such a threshold can be preset for the number of laser radar detections in a distance bin or for the intensity, i.e., the corresponding reflectivity value of one or more laser radar detections in a distance bin. Through the design scheme proposed here of the present invention, the position of a stationary object can be determined and tracked in a particularly stable, reliable and robust manner. Therefore, the function constructed accordingly can be applied particularly reliably and intuitively, i.e., when unstable fluctuations, etc., do not occur. In particular, when, for example, the same laser radar detection is not always detected for a specific stationary object in multiple measurement cycles or the laser radar detection from the same location of the object is not always detected, this is also possible through the design scheme proposed here of the present invention. In this case, the use of lidar detection proposed here can be particularly useful and effective especially for determining the position of stationary objects, since lidar sensors for stationary objects are not prone to drift, or are at least less prone to drift than radar sensors.

[0042] In another embodiment of the present invention, for objects classified as stationary, in order to update the object data of the object, the correlations between different characteristics of the corresponding object stored or described in the corresponding object data are eliminated respectively. In other words, the mutual correlations or couplings between different characteristics of the corresponding object can be deleted or removed, or reset to zero, for example. This may include, for example, correlations between the position and / or speed and / or acceleration and / or yaw angle and / or yaw rate of the corresponding object. By eliminating these correlations, these characteristics are decoupled from each other. Therefore, for stationary objects, a change in the position of the corresponding object from the last measurement cycle or the last position update to the current measurement cycle does not automatically lead to, for example, the corresponding speed or acceleration or the changed yaw angle or yaw rate being caused or stored in the object data within the scope of the update.

[0043] Additionally or alternatively, for objects classified as stationary, in order to model or manage the corresponding objects in the relevant object data, switch to a diagonal or diagonalized P matrix or use such a matrix. The P matrix can, for example, illustrate the uncertainty of the object model or the deviation from the object model or its prediction of the characteristics of the corresponding object. What can be prevented by using a diagonal or diagonalized P matrix is ​​that the corresponding correlation or cross-correlation with one or more other characteristics of the object is constructed due to the update characteristics. This can be seen differently from a moving object, for example, in a moving object, compared with the previous object data, the changed position or a specific or changing speed or a specific or changing acceleration is respectively accompanied by the change of the corresponding other values. Therefore, for example, a P matrix with auxiliary diagonal elements different from zero can be used for moving objects. Through the design scheme proposed here of the present invention, it is particularly simple, effective and reliable to realize or implement the independent updating of the characteristics of stationary objects.

[0044] The present invention also relates to an object tracking device, which can be designed or arranged in particular for a motor vehicle. The object tracking device can therefore be designed, for example, as an assistance system of a motor vehicle or as part of such an assistance system. The object tracking device according to the present invention has an interface for detecting sensor data and a data processing device for processing the detected sensor data. To this end, the object tracking device, in particular its data processing device, can, for example, include a processing device, such as a microchip, a microprocessor or a microcontroller, etc., and a computer-readable data memory coupled to the processing device. In the data memory, a corresponding operation or computer program can be stored, which encodes or implements the method steps, measures or processes described in conjunction with the method according to the present invention or corresponding control instructions. Then, the operation or computer program can be executed by the processing device to realize the execution of the corresponding method. The sensor data can be or include raw data, sensor detection, sensor object, etc. The object tracking device according to the present invention can be or correspond to the object tracking device or tracking device mentioned in conjunction with the method according to the present invention. Therefore, the object tracking device according to the present invention can be designed in particular for managing object data, i.e. object trajectories, of objects detected by the sensor system or represented in the detected sensor data. For example, these object data can be stored in the above-mentioned data memory or managed. The object tracking device can provide the object data, for example via the mentioned or other interfaces, or can make the object data retrievable or output the object data, for example for other assistance systems.

[0045] The invention also relates to a motor vehicle having a sensor system for detecting an environment and an object tracking device according to the invention coupled thereto. The motor vehicle according to the invention can in particular be or correspond to a motor vehicle mentioned in conjunction with the method according to the invention and / or the object tracking device according to the invention. The motor vehicle according to the invention can therefore have some or all of the properties or features mentioned in this case. For example, the motor vehicle according to the invention can have an assistance system for preset driving assistance functions or for auxiliary guidance or at least semi-automatic guidance of the motor vehicle. The assistance system can then be designed to perform its functions based on the object data managed by the object tracking device according to the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Further features of the invention can be derived from the claims, the drawings and the description of the drawings. Without departing from the scope of the invention, the features and feature combinations mentioned above in the description and the features and feature combinations described below and / or shown individually in the drawings can be used not only in the respectively specified combination but also in other combinations or alone.

[0047] Figure 1An exemplary scheme for illustrating a sensor-based object tracking method is shown;

[0048] Figure 2 A schematic diagram for illustrating radar-based object state estimation for object tracking is shown;

[0049] Figure 3 shows an exemplary bar graph for further illustrating object state estimation; and

[0050] Figure 4 A schematic diagram for illustrating lidar-based object state estimation is shown.

[0051] In the figures, identical elements and elements having the same function are provided with the same reference symbols. DETAILED DESCRIPTION

[0052] Figure 1 A method scheme 1 for an object tracking method is shown by way of example, which can be used, for example, in a vehicle for tracking, i.e., for tracking objects in a corresponding environment. The properties of the tracked objects are managed in object trajectories 2, wherein each object trajectory is respectively assigned to the tracked object. Sensor data 3 are respectively detected during a measurement cycle or operation of the method. First, the object trajectory 2 of the tracked, i.e., already known object or the data or properties stored in the object trajectory can be predicted 4 at the current time point or the recording time point of the respectively detected sensor data 3. The object trajectory 2 or the corresponding object and the object detected in the current sensor data 3 or described in the sensor data 3 (which can also be referred to as sensor object here) are then processed in an object association 5. Here, the sensor objects from the current measurement cycle are respectively assigned to one of the object trajectories 2 as far as possible. This can be achieved, for example, based on a possibly abstract preset distance metric or the like. If necessary, a new object trajectory 2 can be provided for each sensor object that cannot be assigned.

[0053] Environmental objects classified as moving by means of object trajectories 2 and / or properties stored or specified in the associated sensor objects can be further processed in a complete update 6. In this case, the current sensor data 3 or sensor objects can be incorporated into the respectively associated object trajectories 2, in particular when using a predefined filter mechanism, such as a Kalman filter or the like. In this case, the object trajectories 2 are updated, whereby a correspondingly updated object trajectory 2 is generated as the starting point for the next measurement cycle or run of the method.

[0054] The object trajectory 2 is respectively a real object detected by the sensor, whose properties or states are estimated recursively over time in or by a recursive tracking algorithm. The object trajectory 2 can also be called an internal trajectory and is stored and updated in a plurality of measurement cycles. In each such measurement cycle or frame, i.e., in the operation of the method, sensor data can be detected by one or more sensors, which describe the measurement data of the sensor object that has been tracked over time in the corresponding sensor. In this case, not only directly measured properties such as position and orientation can be estimated over time, but also dynamic states such as speed or yaw rate can be estimated over time.

[0055] Each object trajectory 2 can also include an entry or a date or a data field in which it is stored or specified whether the corresponding object is moving or stationary, ie is stationary or motionless.

[0056] The object trajectories 2 for stationary objects are described, managed or updated separately in a different way than for moving objects. For this purpose, in the corresponding detection associations 7, not the sensor objects but the individual sensor detections 18 (see Figure 2 , Figure 4 ) is assigned to the object track 2 or an object managed in the object track. For further explanation, Figure 2 For this purpose, a schematic overview is shown with a motor vehicle 11 in which the method described here is used. For this purpose, the motor vehicle 11 has a sensor system 12, which can in particular include a radar sensor and / or a lidar sensor. To carry out the method, the motor vehicle 11 also has an object tracking device 13, which is coupled to the sensor system 12 via an interface 14. The sensor data 3 provided by the sensor system 12 can be processed by the object tracking device 13, for example by means of a processor 15 and a data memory 16, which are schematically shown here.

[0057] Motor vehicle 11 can scan its surroundings with the aid of sensor system 12 and detect objects therein. Here, an external vehicle 17 or a corresponding boundary frame is schematically shown as such an object. In addition, a plurality of sensor detections 18 of sensor system 12 are shown here. These sensor detections 18 include object detections 19 that actually come from external vehicles 17 and clutter detections 20 that do not come from external vehicles 17 and / or are caused by measurement errors or faults. For detection association 7, all sensor detections 18 that are located in search area 21 that are assigned to the corresponding object, here to external vehicle 17, can now be determined based on current sensor data 3. Then, for the assignment or association with external vehicle 17 or the corresponding boundary frame, only sensor detections 18 that are located in search area 21 are taken into account.

[0058] Figure 3A corresponding diagram is shown for illustrating a possible distribution of sensor detection 18 when using a lidar sensor. Figure 2 Compared to the radar detections shown, a greater number of stable object detections 19 may generally occur at the rear of the external vehicle 17 or at the end facing the sensor system 12. However, additional clutter detections 20 may occur. These clutter detections may, for example, come from the exhaust tail of the external vehicle 17 and / or occur inside the external vehicle 17 or the boundary frame of the external vehicle 17, for example due to reflections from windows or parts of the bottom of the external vehicle 17, etc. Here, by way of example, other shapes of the search area 21 are shown for the use of a lidar sensor.

[0059] Depending on the sensors used or available or depending on the type of sensor detection 18, the assignment of sensor detections 18 to external vehicles 17 can be carried out, for example, based on a predefined distance criterion and / or a predefined clutter criterion and / or a predefined similarity criterion. For each radar detection, for example, the associated double speed can be determined and compared with the speed stored in the object track 2 for the external vehicle 17. Only such radar detections or sensor detections 18 whose Doppler speed deviates from the speed stored in the associated object track 2 by at most a predefined absolute value can then be assigned to the external vehicle 17. If, for example, a speed of 0 m / s is stored in the associated object track 2 for stationary objects and a deviation threshold of, for example, a maximum of 5 m / s is predefined, radar detections or sensor detections 18 whose Doppler speed is correspondingly, for example, greater than 5 m / s can be discarded so that they are not assigned to the external vehicle 17.

[0060] In the case where the sensor system 12 comprises a radar sensor or the sensor data 3 comprises radar detections, a stationary estimation 8 can be performed based thereon for the corresponding object, here the external vehicle 17 . Figure 2, the double speed of the sensor detection 18 is shown by a corresponding arrow, the length of which is associated with the absolute value of the corresponding Doppler speed. Although the sensor detection 18 located in the area of ​​the boundary frame for the external vehicle 17, whose Doppler speed is however much greater than the Doppler speed of the remaining object detections 19, in particular than this or a predetermined speed threshold value, can be classified as a clutter detection 20 on this basis. If the Doppler speed of the object detection 19 or, for example, its average value or median value is less than a predetermined speed threshold value in absolute value (which can take into account or describe the measurement uncertainty of the radar sensor), it can be determined from this that the external vehicle 17 is stationary. Therefore, for example, a corresponding flag can be set to indicate this stationary state. If a plurality of radar detections, i.e., for example, a predetermined minimum number or a predetermined minimum portion of the Doppler speeds of the radar detections or object detections 19 assigned to the corresponding object, coincide with the stationary state of the corresponding object, then the stationary flag can be set in particular. If this is not the case, i.e., if, for example, the corresponding predetermined stationary criterion is not met, then the flag cannot be set, reset or deleted.

[0061] Furthermore, state estimation 9 may be performed based on object detection 19 assigned to external vehicle 17. For example, the corresponding radar detections may be evaluated in a histogram. Figure 4 Such a histogram is shown as an example. There, the distance d is plotted on the x-axis, and the radar cross section RCS for the radar detection is plotted on the y-axis. In particular, the radar detections occurring in a plurality of measurement cycles can be cumulatively input into the histogram. In addition, a significance threshold 22 is also shown here. The significance threshold can be preset, for example, as a fixed value or a share or a percentage value of the maximum value of the radar cross section RCS in the histogram. For example, a radar cross section RCS of 75% of the maximum radar cross section RCS contained in the histogram can be determined as the significance threshold 22. Then, the minimum distance d exceeding the significance threshold 22 or the first distance d or distance bin from the distance d=0 can be determined. This is represented here as the first significance bin 23. Then, the radar detection assigned to the external vehicle 17 classified there can be determined as a stable end detection according to the first significance bin 23, or a virtual, that is, an end detection that does not necessarily correspond to the real sensor detection 18 can be formed or generated according to the radar detection classified as the first significance bin 23, for example, by forming an average value or a median value.

[0062] In a similar manner, a histogram may also be constructed for the lidar detections assigned to external vehicle 17 , in which, for example, intensity or reflectivity values ​​of the lidar detections may be used instead of the radar cross section RCS.

[0063] The stable end detection determined in this way based on the object detection 19 can be used as a holding point for the corresponding external vehicle 17. Therefore, this end detection or this holding point can represent the external vehicle 17 or be used to determine its current properties or object data, in particular the position or the distance d from the sensor system 12b or the motor vehicle 11.

[0064] The corresponding object trajectories 2 classified as stationary objects or with an associated current object detection 19 and / or properties determined therefrom (e.g., distance d or a corresponding position or a speed determined based on radar, etc.) can be supplied to the update 10 or further processed within the scope of the update 10. The update 10 represents a method step for updating the object trajectories 2 based on the current sensor data 3 of the stationary objects. The update 10 for the stationary objects can therefore be performed separately from the complete update 6 for the moving objects and can also be different from the complete update.

[0065] For the update 10, all dependencies or cross-correlations between different data or characteristics (e.g. position, velocity, acceleration, yaw angle, yaw rate, etc.) can be eliminated in the corresponding object trajectory 2. As a result, different characteristics or states of the corresponding object can be decoupled from one another. For example, in the update 10, a change in the position of the corresponding object in the x direction, i.e., for example, along the longitudinal direction of the vehicle 11, does not result in the introduction of a velocity or acceleration of the object. For example, a change in the position of the corresponding object in the y direction, i.e., for example, along or parallel to the transverse direction of the vehicle 11, does not result in the introduction of a yaw rate or a changed yaw angle of the corresponding object. As long as the corresponding object is classified as stationary, i.e., for example, a corresponding flag is set, no movement or dynamics of the object are introduced in the update 10 of the object trajectory 2. In the update 10, current characteristics, in particular dynamic characteristics, determined by means of the corresponding current object detection 19 can be written into the corresponding object trajectory 2, for example. For example, these characteristics can be determined or have been determined by means of a stable end detection, i.e., a holding point, determined for the corresponding object.

[0066] Even when the corresponding object is in motion, the radar detection still corresponds to the object being stationary, and the position can still be updated accurately and reliably by independent updating of the position. In this case, for example, this also enables a certain distance d between the motor vehicle 11 and the corresponding object (here, the external vehicle 17) to be stably and accurately adhered to for the corresponding assistance system.

[0067] If, however, the current radar detection indicates a speed of the external vehicle 17 which is different from zero and no longer consistent with being stationary, this speed can be written into the corresponding object trajectory 2 within the scope of the update 10. Thus, in the next measuring cycle or the next run of the method, the corresponding object is no longer classified as a stationary object.

[0068] In the context of the update 10 or state estimation 9 for stationary objects, in particular only individual properties of the respective object can be determined and, if necessary, updated, without, for example, having to estimate or determine a new bounding frame for the respective stationary object. This can achieve an overall improvement in the stability of object tracking for stationary objects, for example, compared to handling or managing stationary objects in the same way as for moving objects with a completely newly determined corresponding bounding frame.

[0069] If, for example, in a measuring cycle the sensor system 12 or at least one sensor for an object that has already been detected in advance does not provide a sensor detection 18 or an object detection 19, the detection association 7 may therefore not be able to be performed or meaningfully performed for this object in the current measuring cycle if necessary. It may also happen that in the current case the detection association 7 may be too complex for a specific sensor detection (e.g. of a lidar sensor). However, starting from the object trajectory 2 that already exists for the object, an object association 5 may still be performed if necessary. If the object is an object that is classified as stationary, an update 10 may also be performed for this object. For this purpose, for example, the motion state of the object can be obtained from the corresponding object trajectory 2 or the object data stored in the object trajectory and / or the motion state determined for the object by at least one sensor that detects the object within the sensor can be used.

[0070] The process described here for a stationary object, for example, for the stationary estimation 8 and / or the updating 10, can in principle be applied to any stationary object. However, it may be particularly useful to use the described method at least or only for an object directly in front, that is, for example, the front vehicle closest to the motor vehicle 11 in the direction of travel of the motor vehicle. As the front vehicle, for example, a vehicle located in front of the motor vehicle 11, that is, on the front side, can be used or considered, which has a similar orientation to the motor vehicle 11 and is located in the same lane as the motor vehicle 11. Such a front vehicle can be obtained, for example, with the help of sensor data 3 and / or with the help of road or lane recognition data and / or map data and / or vehicle data received via a Car2Car or Car2X data connection. The precise knowledge and updating or tracking of the position and speed of such a front vehicle can be important, for example, for the safety-related compliance with the preset target distance between the motor vehicle 11 and the corresponding front vehicle as accurately as possible. Compared with relying on the accurate tracking of other objects that may be located at a greater distance and / or on different lanes, adaptive cruise control, etc., for example, may rely to a greater extent on the corresponding accurate tracking of the front vehicle.

[0071] In summary, what can be automatically estimated in this method is whether or when the detected object is almost or completely stationary. In this identified stationary state, all characteristics or states of the object are estimated separately. For example, there is no influence of position on speed. If available, radar detection can be used in particular to identify stationary state. Due to the direct measurement of Doppler velocity, this radar detection can be particularly well suited for this. If multiple such Doppler measurements confirm the complete stationary state of the corresponding object, for example, in one or more measurement cycles, any dynamics, that is, any correlation between different characteristics or states, can be prohibited in the relevant object data. It is also possible to estimate the stable distance d from the corresponding object by means of a preset robust estimation method, for example, based on a histogram or a median, in one or more measurement cycles by means of radar detection and / or by means of laser radar detection. Therefore, it is also possible to achieve particularly stable and robust tracking of stationary objects. The corresponding data or characteristics can be directly written or incorporated into the corresponding object data. In this way, for example, an overestimation or underestimation of the distance d from the corresponding object can be prevented or reduced compared to conventional solutions. The method described here also makes it possible to achieve particularly stable tracking of stationary objects, prevent the apparent occurrence or detection of movements or rotations due to measurement uncertainties, and enable precise and stable distance estimation or distance monitoring even for objects that were previously stationary but have just been in motion. This in turn enables particularly reliable, robust and safe implementation of position- or distance-based assistance functions.

[0072] Overall, the described examples thus show how an improved object state estimation for stationary or immobile objects, in particular vehicles, can be achieved.

[0073] Reference numerals list

[0074] 1 Methods

[0075] 2 Object Trajectory

[0076] 3 Sensor Data

[0077] 4 Prediction

[0078] 5 Object Association

[0079] 6Fully updated

[0080] 7 Detecting associations

[0081] 8Stationary Estimation

[0082] 9 State Estimation

[0083] 10 Update

[0084] 11Motor Vehicles

[0085] 12 sensor system

[0086] 13 Object Tracking Device

[0087] 14Interfaces

[0088] 15 processors

[0089] 16 Data memory

[0090] 17 External Vehicles

[0091] 18 sensor detection

[0092] 19 Object Detection

[0093] 20Clutter detection

[0094] 21Search Area

[0095] 22 Significance Threshold

[0096] 23 First significance bin

[0097] RCS radar cross section

[0098] d distance

Claims

1. A method (1) for object tracking over a plurality of measurement cycles of a sensor system (12), wherein: - detecting an object (17) by means of sensor data acquired during a measuring cycle of the sensor system (12), - for a detected object (17), determining a movement state of the object with the aid of the sensor data (3), wherein: Each object (17) is classified as a stationary object (17) or a moving object, - storing object data (2) for each detected object (17), said object data comprising a plurality of properties of the respective object (17) and a specific movement state of the respective object, - updating the object data (2) of an object (17) which has already been detected in a previous measuring cycle with the aid of the current sensor data (3), - only for those objects (17) which are classified as stationary objects (17), the properties of these objects stored in the object data (2) are updated independently of one another.

2. The method (1) according to claim 1, characterized in that: The movement state is determined by means of radar detection (18, 19), wherein the object (17) is classified as a stationary object (17) if the speed of the object (17) determined by means of the radar is less than a predefined speed threshold value in absolute value.

3. The method (1) according to any one of the preceding claims, characterized in that The object (17) is only classified as a stationary object (17) if at least one predefined number of measurements and / or at least one predefined proportion of a plurality of performed measurements yields a speed of the object (17) that lies below a predefined speed threshold value.

4. The method (1) according to any one of the preceding claims, characterized in that For objects (17) classified as stationary, the object data (2) of the stationary objects are updated based on the respective sensor detections (18), and for objects classified as moving, the object data (2) of the moving objects are updated based on the sensor data output by the sensor system (12).

5. The method (1) according to any one of the preceding claims, characterized in that For an object (17) classified as stationary, a stable end detection is determined from all current sensor detections (18, 19) to determine the current position of the corresponding object (17) to be updated as a characteristic in the object data (2), wherein the current sensor detection is assigned to the corresponding object (17) based on a preset criterion, in particular a distance-based criterion with respect to a previous position of the corresponding object (17), and the end detection serves as the end of the corresponding object (17) facing the sensor system (12).

6. The method (1) according to claim 5, characterized in that: In the case where the sensor detection (19) assigned to the corresponding object (17) includes a radar detection, a stable end detection is determined taking into account a radar cross section (RCS) of the radar detection, in particular as a radar detection closest to the sensor system (12), the radar cross section (RCS) of the radar detection being greater than a preset threshold value (22).

7. The method (1) according to any one of the preceding claims, characterized in that For objects (17) classified as stationary, at least a part of the characteristics of the corresponding object (17), in particular the position, is determined based on a lidar detection, wherein the lidar detection from the current measurement cycle is assigned to a boundary frame (17) for the corresponding object (17), and a stable end detection is determined with the aid of a distribution of the lidar detection, in particular a distance histogram, which serves as the end of the corresponding object (17) facing the sensor system (12).

8. Method (1) according to any one of the preceding claims, characterized in that For objects (17) classified as stationary, in order to update (10) the relevant object data (2) of the stationary objects, the correlations between different characteristics of the corresponding objects (17) stored in the corresponding object data (2) are eliminated respectively, and / or in order to model the corresponding objects in the relevant object data (2), a switch is made to a diagonal P matrix.

9. An object tracking device (13), in particular for a motor vehicle (11), having an interface (14) for detecting sensor data (3) and a data processing device (15, 16) for processing the detected sensor data (3), wherein: The object tracking device (13) is designed to carry out the method (1) according to any of the preceding claims.

10. A motor vehicle (11) having a sensor system (12) for detecting the environment and having an object tracking device (13) according to claim 9 coupled to the sensor system.

Images