MONITORING AN UNDERFLOOR SURVEILLANCE AREA

Abstract

Disclosed is a computer-implemented method for monitoring a monitoring area located between a floor surface and an underbody of a vehicle parked in a parking position on the floor surface, the method comprising: Firstly, detecting first ground information characterising the ground surface with a first sensor system arranged on the vehicle before the vehicle reaches the parking position; Storing reference information based on the first soil information in a storage system; and In response to detecting that the vehicle is expected to leave the parking position: Second acquisition of second ground information characterizing the ground surface with a second sensor system arranged on the vehicle; and First classification, by a classifier trained with a data set based on the reference information and the second ground information, whether an object is present in the monitoring area.

Description

FIELD OF TECHNOLOGYThe invention relates to a computer-implemented method for monitoring a monitoring area lying between a floor surface and an underbody of a vehicle parked on the floor surface, and to a corresponding computer program product, computer system and vehicle.PRIOR ARTModern vehicles in the automotive field are equipped with an increasing number of sensors that monitor the environment of the vehicle. In the field of autonomous vehicles, sensor systems are indispensable for reliable, accident-free control of the vehicle. When the sensors detect objects, the autonomous vehicle control systems may provide automatic maneuver, stop, and steering functions. Examples of vehicle sensors are cameras, ultrasonic, radar, lidar and infrared sensors, and laser scanners.The environment monitoring of a vehicle also includes its subfloor region. Because a vehicle should only leave an unobstructed footprint, techniques have been developed that allow to determine whether an obstacle is located under a stationary vehicle. Ultrasonic sensors offer a comparatively cost-effective and interference-insensitive possibility for implementing underbody monitoring. For example, the underbody of a vehicle may be equipped with ultrasonic sensor pairs configured as transmitters and receivers to measure an ultrasonic pulse reflected on the ground below the vehicle at the stationary vehicle, after stopping and before driving. By analyzing the time profile of the reflected pulse and / or a classification, a difference between the sensor data measured after the stop and before the departure can be recognized, which could indicate an object located under the vehicle.It is an object of the invention to provide an improved method for monitoring a monitoring area lying between a floor surface and an underbody of a vehicle parked on the floor surface in a storage position, and a corresponding computer system, computer program product and vehicle. The objects underlying the invention are achieved by the features of the independent claims.SUMMARYIn one aspect, a computer-implemented method for monitoring a monitoring area lying between a floor surface and an underbody of a vehicle parked on the floor surface in a storage position is disclosed, the method comprising:first detecting first ground information characterizing the ground surface using a first sensor system arranged on the vehicle before the vehicle reaches the parking position;storing reference information based on the first ground information in a storage system; andIn response to determining that the vehicle is expected to leave the parking position:second capturing of second ground information characterizing the ground surface with a second sensor system arranged on the vehicle; andFirst classifying, by a classifier learned with a data set, based on the reference information and the second ground information, whether an object is present in the monitoring area.In a further aspect, a computer program product, in particular a computer-readable storage medium, is disclosed which stores computer-executable code, wherein the code, when executed by at least one processor of a computer device, causes the computer device to execute a method for monitoring a monitoring region lying between a floor surface and an underbody of a vehicle parked on the floor surface in a storage position, wherein the method comprises:first detecting first ground information characterizing the ground surface using a first sensor system arranged on the vehicle before the vehicle reaches the parking position;storing reference information based on the first ground information in a storage system; andIn response to determining that the vehicle is expected to leave the parking position:second capturing of second ground information characterizing the ground surface with a second sensor system arranged on the vehicle; andFirst classifying, by a classifier learned with a data set, based on the reference information and the second ground information, whether an object is present in the monitoring area.In another aspect, a computer system is disclosed, comprising a processor and a memory functionally connected to the processor, wherein the memory stores computer-executable code, wherein the code, when executed by the processor, causes the computer system to execute a method for monitoring a monitoring area lying between a floor surface and an underbody of a vehicle parked on the floor surface in a storage position, wherein the method comprises:first detecting first ground information characterizing the ground surface using a first sensor system arranged on the vehicle before the vehicle reaches the parking position;storing reference information based on the first ground information in a storage system; andIn response to determining that the vehicle is expected to leave the parking position:second capturing of second ground information characterizing the ground surface with a second sensor system arranged on the vehicle; andFirst classifying, by a classifier learned with a data set, based on the reference information and the second ground information, whether an object is present in the monitoring area.In another aspect, a vehicle is disclosed, comprising a computer system configured to control components of the vehicle, the computer system comprising a processor and a memory operatively connected to the processor, the memory storing computer-executable code, the code, when executed by the processor, causing the computer system to execute a method for monitoring a monitoring area lying between a ground surface and an underbody of the vehicle parked on the ground surface in a storage position, the method comprising:first detecting first ground information characterizing the ground surface using a first sensor system arranged on the vehicle before the vehicle reaches the parking position;storing reference information based on the first ground information in a storage system; andIn response to determining that the vehicle is expected to leave the parking position:second capturing of second ground information characterizing the ground surface with a second sensor system arranged on the vehicle; andFirst classifying, by a classifier learned with a data set, based on the reference information and the second ground information, whether an object is present in the monitoring area.It should be understood that one or more of the embodiments disclosed herein may be combined with each other as long as the embodiments are not mutually exclusive.BRIEF DESCRIPTION OF THE DRAWINGSExamples are explained in more detail below with reference to the drawings. The following are shown: FIG. 1 shows a parking process of a vehicle on a floor surface, FIG. 2 is a block diagram showing components of a vehicle; and FIG. 3 shows a flow diagram of a method for monitoring a monitoring region.DETAILED DESCRIPTIONAbove all, but not exclusively, in the field of autonomous driving, it is of interest, before driving off a stationary (e.g. parking, holding or waiting) vehicle, to ensure that no object (e.g. an object or a living being) is located under the vehicle in order to avoid damage to the object and / or the vehicle as a result of the driving off.The method disclosed herein is computer-implemented as described in more detail herein. It may be implemented, for example, by program instructions executable by one or more processors of a computer system configured to control components of the vehicle. Such a computer system could preferably be installed in the vehicle ("vehicle computer"), but could also be connected to a computer system installed in the vehicle via a communication network such as the Internet or a local area network ("remote computer", e.g. a server). A working-part execution of parts by the method by a vehicle computer and of further parts of the method by a remote computer is also conceivable.The vehicle is parked in a parking position on a floor surface defined on a floor. The parking position could be defined, for example, by two or more contact surfaces of wheels of the vehicle on the ground, or by two or more points on the ground, perpendicularly above which two reference points of the vehicle are located. The floor surface overlaps with at least a part of the vehicle. For example, the ground surface could be defined by a closed geometric shape, e.g. a circle, an ellipse or a polygon (e.g. a rectangle), which encloses the contact surfaces of the wheels of the vehicle, e.g. with minimal surface area. The floor surface could also be defined, for example, by a closed curve which lies perpendicularly below an outer contour of the vehicle. The floor surface could also include, for example, a surrounding area of the vehicle, for example defined by a closed line which lies perpendicularly below a line which has a predetermined distance (for example 1 meter (1 m)) around the outer contour of the vehicle. For geometric simplification, the floor surface could be defined by a closed geometric shape (as described above) that includes a floor region thus defined. The ground surface could also be defined by a drivable area with predefined properties, for example by the dimensions of a parking space or a traffic gap, or e.g. by the geometry of a traffic or parking space predefined by means of markings or by a traffic or parking guidance system. It is assumed that the vehicle has a sensor and a computer system in order to be able to automatically identify the ground surface or to define it by interaction with an external information source.The vehicle has an underbody. The volume located between the underbody and the floor surface is referred to herein as a monitoring area. As described herein, the method could provide a classification that provides an indication of whether or not the statement that the first and second ground information indicate the presence of an object within the surveillance area is true with a certain probability. Based on this classification, the computer system executing the method could generate a reaction, for example, an enable to leave the parking position in the absence of an object in the monitoring area, or a signal that prevents autonomous leaving of the parking position, and / or a signal that alerts a driver or passenger of the vehicle if an object is in the monitoring area.The vehicle may use sensor systems to acquire ground information describing measurable characteristics of the ground surface. More specifically, a first sensor system of the vehicle acquires first ground information and a second sensor system of the vehicle acquires second ground information. The first and second sensor systems could include various sensors. Each of the two sensor systems could have at least one sensor. Each of the two sensor systems could have at least one sensor that is not part of the second sensor system. Likewise, each of the two sensor systems could not have a sensor which is also a component of the respective other sensor system. The sensor systems could include multiple sensors, for example a sensor configured as a receiver and optionally, e.g. depending on the sensing technology implemented, a sensor configured as a transmitter or a non-sensor transmission unit or energy source such as a light source. A sensor system could comprise matched pairs of transmitters and receivers, or else sensors which are not configured in pairs, such as a transmitter having a plurality of receivers at different positions.The acquisition of the first ground information is referred to as first acquisition below and the acquisition of the second ground information is referred to as second acquisition below. The first capturing could take place before the second capturing, for example in a period of time that ends before the vehicle has reached the parking position. For example, the computer system executing the method could recognize based on a sensor, e.g. on the basis of roadway markings, traffic signs and signals, position and geometry of a spatial gap, a position and structure of a ground geometry and / or obstacles, etc., that a driver of the vehicle starts a parking maneuver or is in a traffic situation that requires a stop of the vehicle (e.g. approach to a red traffic light, to a non-free zebra strip or to a end of traffic). The computer system could also receive a control signal via surrounding network stations that identifies an intermediate or final stopping point of a route to be traveled by the vehicle. A detected or received indication of an imminent shutdown of the vehicle could trigger the first detection.The second detection could be performed in response to the computer system determining that the vehicle is expected to exit the parking position. This could be determined, for example, by receiving a signal corresponding to a desire of a person in the vehicle or waiting on the vehicle to leave the parking position, or corresponding to a plan of a unit managing the position of the vehicle that requires changing the position of the vehicle, or by sensor-based detection of an event in the environment of the vehicle that requires leaving the parking position (e.g., changing a red light to red-yellow).The first acquisition and / or the second acquisition could each comprise one measurement (e.g. passing through a measurement cycle) or a plurality of measurements. The measured, unprocessed raw data of the one or more measurement(s) and / or data derived from the raw data by conversion (e.g. scaling the measured values and / or arithmetic determination of a height profile of the ground surface) could be incorporated into the first or second ground information. In principle, it could be possible to repeat the first / second acquisition and to include a measurement series aggregated from the plurality of individual measurements in the first or second piece of ground information. Such aggregation of measurement data could provide more significant ground information, which is characterized, for example, by a higher signal-to-noise ratio and / or a higher confidence level.The first piece of ground information and / or information derived from the first piece of ground information, which could also go beyond arithmetic conversion, could be stored in a storage system as reference information for later use in the course of the first classification. For example, the first ground information could be provided to a preclassification that could provide an interpretation of the first ground information beyond recalculating as part of the reference information, such as a classification of a ground condition of the ground surface, as described herein. A storage system storing the reference information could be part of the computer system executing the method or another computer system on board the vehicle, for example, or be interchangeably connected to a computer system installed in the vehicle (e.g., in the form of a memory card), or the storage system could be accessible to the computer system executing the method via a network connection (e.g., cloud storage).The computer system executing the method could implement a classifier and / or be connected, for example, via a bus system on board the vehicle to a local computer system implementing a classifier and / or via a computer network to a remote computer system implementing a classifier. In a machine learning method, the classifier could be trained with a training data set which contains combinations of individual data sets of first training data corresponding to the reference information and of second training data corresponding to the second ground information, to which a classification specific to this combination and made by a human is assigned in each case for each combination of individual data sets of the first and the second training data, which classification states whether the first and / or the second training data of this individual combination corresponds to an absence or a presence of an object.The computer system could pass the reference information and the second ground information and / or information derived from the second ground information to the classifier as input. The computer system could use, for example, an optional preliminary stage which, analogously to the above-described preclassification of the first piece of ground information, could provide a preclassification of the second piece of ground information on its own. Based on the reference information, i.e. on the first ground information and / or optionally information derived from the first ground information, and on the second ground information and / or optionally information derived from the second ground information, the classifier could then find a classification corresponding to its training state by the first classification, which classification states states indicates whether the first and the second ground information in the light of the training dataset indicate the presence of an object within the monitoring area. This could be, for example, a binary classification, or a probability for the presence of an object within the monitoring area, which e.g. the computer system subsequently converts into a binary decision e.g. with the aid of a threshold value.The fact that the vehicle is in the parking position at the time of the second detection could require the second sensor system to be arranged on the underbody of the vehicle in order to detect the second ground information. Underbody sensors could be exposed to comparatively severe contamination due to their proximity to the floor. In addition, components of the vehicle which are arranged below the underbody (e.g. exhaust pipes, wheel axles, wheel suspension) and / or unevennesses of the underbody could protrude into the monitoring region and thus partially obstruct the second detection. Due to the close space conditions, reflections could reduce the signal quality of the second detection. In a second detection in the visual spectral range, the shading of the monitored region by the vehicle could also make the second detection of a high-contrast image of the ground surface more difficult.In contrast, the first sensor system performing the first detection could differ from the second sensor system. This difference could allow the ground information to be acquired from another mounting location on the vehicle. The first sensor system could thus have the advantage of a more free field of view onto the ground surface, i.e. a more free and / or larger volume could be available for the propagation of signals serving the first detection. In particular, the first sensor system could be arranged in the front or rear region of the vehicle. It could be possible to use existing sensors as the first sensor system, so that dedicated sensors for the first sensing would not have to be added to the vehicle.As a consequence of a more free view of the ground surface, the first ground information could include a richer, more complete and / or more true-to-truth reproduction of the ground surface. This could provide the advantage of increasing the confidence level of the classification based on the first ground information and thus providing a more accurate statement about the presence of an object within the monitored zone. In particular, the basis of the classification on ground information generated by two different sensor systems could necessitate training the classifier with a training data set which contains both first training data sets corresponding to the first ground information and second training data sets corresponding to the second ground information. This could also contribute to an increase in the confidence level of the classification.A difference between the two sensor systems could also be found in the selection of the sensor types used. The first sensor system could advantageously have one or more sensors of a sensor type, the use of which in the second sensor system is disadvantageous. For example, the first sensor system could use a camera operating in the visual spectral range for the first detection. Due to the naturally poor lighting conditions under the underbody of the vehicle, the second sensor system could more reasonably use one or more sensors for the second detection that do not operate in the visual spectral range, such as ultrasonic sensors. The use of different sensor systems could thus provide flexibility in the selection of the technical means for implementing underbody monitoring. In addition, the use of different physical measured variables could reduce the effect of systematic error sources on the classification and thus also contribute to an increase in the confidence level of the classification.According to one example, the first sensing is triggered when the vehicle begins autonomous parking on the ground surface. This could be, for example, a point in time at which the vehicle begins to move from a starting position in a parking manner to the parking position. This does not necessarily mean that the vehicle needs to stand in the starting position when the first capturing begins. Rather, the start position could correspond to a start time at which the parking process is logically started by a control module, regardless of whether the vehicle is stationary or driving at the start time. In particular, the first capturing could be completed before the vehicle travels on the part of the floor surface covered by the vehicle in the parking position. This could be the case, for example, when recording a camera image. However, the first detection could also be continued during the approach to the parking position, for example when the floor surface is measured with a scanner. Triggering the first detection at the beginning of an autonomous parking maneuver could enable a vehicle with autonomous parking and unparking functionality to detect an object under the vehicle with high confidence based on a high data quality of the first ground information.According to one example, detecting a parking onto the ground surface initiated by a driver of the vehicle triggers the first detection. For example, the computer system executing the method could recognize that a driver of the vehicle starts a parking maneuver based on roadway markings, traffic signs and signals, position and geometry of a spatial gap, position and structure of a ground geometry and / or obstacles, by engaging a reverse gear, moving to a free ground surface with blocked possibilities for continuing, etc. Similar to the autonomously parking vehicle, the start of a parking maneuver could be, for example, a time at which the vehicle starts to move from a starting position in a parking manner to the parking position. This does not necessarily mean that the vehicle needs to stand in the starting position when the first capturing begins. Rather, the starting position could correspond to a starting time at which it is detected that the driver is likely just in a parking maneuver, regardless of whether the vehicle is stationary or driving at the starting time. In particular, the first capturing could be completed before the vehicle travels on the part of the floor surface covered by the vehicle in the parking position. This could be the case, for example, when recording a camera image. However, the first detection could also be continued during the approach to the parking position, for example when the floor surface is measured with a scanner. Triggering the first detection at the beginning of a manual parking maneuver could enable a vehicle with an absence or deactivated autonomous parking and unparking function to detect an object under the vehicle with high confidence based on a high data quality of the first ground information.According to one example, the determination comprises a deactivation and renewed activation of the vehicle and / or a detection of a predefined traffic situation. Deactivation and renewed activation of the vehicle could be recognizable with high certainty by the computer system and have a high correlation with imminent departure from the parking position, e.g. upon departure from a parking space. However, the method could also be used in traffic situations which do not comprise parking the vehicle in and out, such as, for example, stopping in front of a red traffic light, a zebra strip crossed by a pedestrian, in a traffic jam, etc., followed by a subsequent restart when the traffic light changes from red to green, when the zebra strip is cleared, when the vehicle continues to drive through the traffic jam or at its end, etc. The computer system could be configured to detect a predefined set of such traffic situations corresponding to an imminent restart of the vehicle by means of sensors integrated into the vehicle, which could also comprise the first and / or the second sensor system or parts thereof, and then to trigger the second detection and the first classification. In this way, the risk potential could be increased in a greater number of traffic situations that provide for starting of the vehicle.According to one example, if the first classification reveals that an object is present in the monitored zone, the method additionally comprises a second classification as to whether the object interferes with leaving the parking position. This could likewise be done by the classifier trained with a correspondingly expanded training data set as described below, or a second classifier trained with the expanded training data set could be used. Such an expanded training data set could contain combinations of individual data sets from the first training data corresponding to the reference information and from the second training data corresponding to the second ground information, to which a classification carried out by a human being specific to this combination is respectively assigned for each combination of individual data sets of the first and the second training data, which classification states whether the first and / or the second training data of this individual combination correspond to an absence of an object or correspond to a presence of an object which could impair the leaving of the storage position (e.g. a living being, glass chip, a nail) or correspond to a presence of an object which does not impair the leaving of the storage position (e.g. leaf leaves, a ball, waste). In this way, the computer system could permit the departure from the parking position in the event of the detection of a non-interfering object, as a result of which waiting times could be saved for a passenger of the vehicle by means of unnecessary delays.According to one example, the method additionally comprises evaluating the first ground information in order to determine a first ground condition and / or evaluating the second ground information in order to determine a second ground condition, wherein the reference information comprises the first ground condition, wherein the first classification is based on the first and the second ground condition, wherein the first and the second ground condition comprise:a first and a second height profile of the floor surface; and / ora first and a second floor material of the floor surface; and / ora first and a second time-variable ground state, respectively.This could make it possible to distinguish whether a difference recognizable in the first and the second ground information indicates the presence of an object in the monitored area or is due to an actual or apparent change in the ground condition. In this way, the accuracy of the classification for detecting the presence of an object in the monitored zone could be increased.The first or second soil condition could be determined in each case by an arithmetic and / or classification evaluation of the first and / or the second soil information. In this case, a material property of the floor surface which is considered to be invariable for the duration of the shutdown could be determined as the floor material. For example, the first and / or the second ground information could have specific signals, signal sequences, signal components, signal distributions and the like, which are specific for the material composition of the ground surface and thus permit a determination of this composition. For example, an arithmetic and / or classification evaluation of the first and / or the second ground information could enable a recognition of specific differences for asphalt, headstone pavement, gravel, sand, gravel, mud, grass, snow, etc. A change in the ground material between the first and second sensing could indicate the presence of an object in the surveillance area.In contrast, a material property of the ground surface that is considered variable for the duration of the shutdown could be determined as the ground state. For example, the first and / or the second ground information could have specific signals, signal sequences, signal components, signal distributions and the like, which are specific for the ground state and thus permit a determination of this state. For example, an arithmetic and / or classification evaluation of the first and / or the second piece of ground information could enable a detection of specific differences for a wet, moist or dry state of the ground surface or a covering of the ground surface with snow, leaves, dirt particles and the like. In this way, a detected change in the second ground information versus the first ground information due to a change in the ground state could be more easily distinguishable from a change due to the presence of an object in the surveillance area.A detection of a height profile of the ground surface could be obtained, for example, by comparing a measured signal shape (e.g. time of flight or brightness differences depending on the measurement method) with a known or expected signal shape of a planar ground. A further possibility could be the emission of a pulse having a known pulse shape and the deconvolution of the measured response signal having the known pulse shape. These or other non-contact measurement methods could make it possible to determine the presence of an object in the monitoring area by a difference in the height profiles obtained from the first and second ground information.The determination of a ground condition of the ground surface does not necessarily have to be carried out on the basis of both the first and the second ground information. If, for example, one of the two sensor systems is suitable for providing ground information which enables determination of the ground condition with a higher reliability than the ground information provided by the other sensor system, the complexity of the first classification could be reduced by omitting the determination of the ground condition from the ground information which is less meaningful for determining the ground condition.The following case examples illustrate the use of the first and / or second soil condition in the first classification. In one example case, the vehicle stops on a paved road. One or both of the first and second ground information provide a detection of the ground material "paving stones", and the second ground condition additionally includes a second height profile of the ground surface. Based on the reference information, the second ground information, the ground material "paving stones", and the second height profile, the classifier finds a structure in the second height profile that does not match its state learned for paving stones from the training dataset. As a result of the classification, the classifier returns a presence of an object in the monitoring area.In a further example of the case, the vehicle is parked on a field path covered with grass. Pre-evaluations of the first ground information and the second ground information result in a classification of the first ground material "sand" and the second ground material "grass". The classifier additionally uses the first and the second ground information as input during the first classification. The two pieces of ground information include waveforms and distributions indicating asperities of the ground surface. These asperities lie in a range of variation for grass learned from the training data. The classifier then returns the absence of an object in the monitored zone.In another example of the case, a trolley travels under the parked vehicle. The first and second soil conditions do not include a height profile, but the first soil material "asphalt" and the second soil materials "asphalt" and "coat". The state learned from the training data causes the classifier in all cases containing the second ground material "coat" to return that an object is present in the monitoring area. In this way, for example, a trolley located under the vehicle can be identified on the basis of a material difference.In a further example of the case, the vehicle parks on a grass surface for several days during rainy weather. During this time, the grass dries and aligns under the vehicle. The first and second soil conditions include the soil material "grass", the first soil condition "wet", the second soil condition "dry", and different height profiles. The classifier does not recognize features atypical for grass based on the state learned from its training data set and returns that no object is present in the monitoring area.In a further example of the case, the vehicle is parked on the side of the road at a winter morning. As the first ground state and the first ground material, "snow" is determined. Vehicles passing throughout the day distribute snowmats among the vehicle. Again, "snow" is recognized as the second ground state, with the second ground materials "snow" and "sludge". The accumulated snowmat is recognized as a change in the second height profile with respect to the first height profile, but does not correspond to a case of an object in the monitoring area learned from the training data. The classifier then returns an absence of an object in the monitored zone.According to an example, the reference information comprises the first ground information. This could allow performing the first classification directly based on the first ground information and thus reduce a computational effort to prepare the first classification, for example, by deriving additional information from the first ground information.According to one example, the classifier comprises an artificial neural network, in particular a convolutional neural network (CNN) or a support vector machine (SVM). These artificial neural networks could be particularly well suited for performing the first classification with high confidence and robustness without excessive data processing complexity, with good environmental suppression and good suitability for binary classifications.According to one example, the first capturing begins before the underbody overlaps with the ground surface. This could ensure that at least at the beginning of the first capturing a free field of view onto the ground surface is present, i.e. a freer and / or larger volume could be available for the propagation of signals serving the first capturing. In the example of a first detection using a camera, the entire floor surface could thus be detected in a single recording. In the example of a scanning first detection, it could be ensured that an obstacle-free scanning of the floor surface is possible during the entire duration of the first detection, whereby a detection of the entire floor surface could become possible. In this way, it could be ensured that the first ground information is complete and has the best possible data quality. This could counteract a reduction in the confidence of the first classification as a result of a low data quality of the first piece of ground information.According to one example, the first piece of ground information is aggregated from a plurality of individual measurements of the first sensor system carried out during a drive of the vehicle to the parking position. This could on the one hand enable scanning of the floor surface with a scanning device such as a laser scanner or a line scanner operating in the visual or infrared spectral range with high resolution. On the other hand, by superimposing data points corresponding to identical ground surface coordinates, an increase in the signal-to-noise ratio in the first ground information could be achieved. Both could contribute to a higher data quality of the first piece of ground information and thus to a first classification with a confidence level that is as high as possible.According to an example, the first sensor system detects the first ground information with a sensor implementing a sensor type selected from:a sensor type operating in the visual or infrared spectral range, in particular a camera;a sensor type operating according to an echo method, in particular an ultrasonic or radar sensor, and / or in particular configured to acquire the first ground information with a multi-frequency measurement; anda sensor type operating according to a scanning method, in particular a lidar sensor or a laser scanner.A sensor operating in the visual or infrared spectral range could enable a detection of the first piece of ground information with a high sensitivity, a high spatial resolution and / or a high dynamic range. In particular, a sensor present or provided in the vehicle, such as a camera, in particular a front and / or a rear camera, could already be used for this purpose. The vehicle preferably has a light source which is designed to illuminate the floor surface under dark lighting conditions in order to enable reliable detection of an object in the monitored region even in these cases.A sensor operating according to an echo method could provide the energy for generating a signal that allows the first ground information to be acquired by decoupling in the direction of the ground surface and thus allow a first acquisition that is more independent of external factors such as the given lighting conditions. A sensor operating according to an echo method could, for example, capture the first ground data on the basis of a back scattering of the transmitted signal from the ground surface and / or on a reflection of the signal deflected at a reflecting surface in the direction of the ground surface at the ground surface. In this case, a measurement of the ground surface using a multi-frequency signal (multi-frequency measurement) could facilitate a plausibility check of the first ground information and / or could further increase the confidence of the first classification on the basis of the provision of an independent additional measurement.A scanning sensor could provide the energy for a signal that allows the first ground information to be acquired by interaction with the ground surface at high spatial resolution. This could achieve a high data quality of the first piece of ground information, which enables reliable classification with high confidence. For example, the sensor could capture the first piece of ground information while the vehicle is moving towards the parking position, e.g. by one-dimensional scanning at a constant distance from the captured area of the ground surface. In another example, a scanning sensor could be configured for two-dimensional scanning of the floor surface, such that the first detection could also be carried out in the stationary state of the vehicle.According to one example, the second sensor system includes an underbody sensor and optionally a vehicle environment sensor. An underbody sensor could be a short distance from the ground surface and thus allow for accurate and complete sensing of the second ground information. The underbody sensor could, for example, implement a sensor type that is insensitive to the lighting situation and / or the narrow space conditions in the monitored region, for example one or more ultrasonic sensors. A vehicle environment sensor could enrich the second ground information with additional information corresponding to outer regions of the ground surface and / or ground regions adjacent to the ground surfaces, which can be detected by the underbody sensor, e.g., not at all or only with high uncertainty. A vehicle environment sensor could include, for example, a scanner or a camera. For example, the vehicle environment sensor could be configured to sense the second ground information within a limited area around the vehicle, e.g., one or two meters.A vehicle environment sensor could have the same type of sensor as the underbody. In this case, it could be possible to combine the second piece of ground information provided by the underbody sensor and the second piece of ground information provided by the vehicle environment sensor to form a combined data record of second piece of ground information, optionally after a calibration of the underbody sensor and / or of the vehicle environment sensor and / or after a conversion of the respective second piece of ground information on the basis of a known calibration of the underbody sensor and / or of the vehicle environment sensor. However, the vehicle environment sensor could also implement a sensor type different from the underbody sensor. In this way, second piece of ground information generated during the second detection, which corresponds to different regions of the ground surface and / or an adjacent environment of the ground surface, could be based on different physical measured variables and / or different measurement methods. This could reduce a possible influence of systematic error sources of the second detection on the first classification and further increase the confidence of the first classification by multidimensionality of the database.According to an example, the first sensor system comprises a sensor implementing a sensor type that does not implement a sensor of the second sensor system. Thus, the first ground information and the second ground information could contain information components characterizing the ground surface based on different physical measured variables and / or different measurement methods. This could reduce a possible influence of systematic error sources of a single sensor type on the first classification and further increase the confidence of the first classification by multidimensionality of the database.Reference will now be made to the drawings, wherein elements similar to each other are denoted by the same reference numerals.FIG. 1 shows a schematic view of a vehicle 100 approaching (arrow) an area with optional floor marking 154. The surface serves for parking a vehicle. In the scenario of FIG. 1, the vehicle 100 comes to a standstill on the surface and subsequently leaves the surface again, as is the case, for example, when parking or when stopping due to a traffic situation. The ground marking 154 could, for example, have a parking area in a parking garage or in a parking lot or be a stop marking on a road, e.g. in front of a traffic light or at an intersection or junction. The vehicle 100 is to ensure that no object (e.g., a ball, an animal, a piece of waste) is located beneath the vehicle prior to leaving the area on which it is placed, to ensure that leaving the area does not cause damage to the vehicle 100 and / or an object that may be located beneath the vehicle.Using a first sensor system 120 and / or further sensor systems installed on board the vehicle 100, for example a front or rear camera and / or a scanning device such as a laser scanner, the vehicle 100 could acquire data which, for example, represent a two-dimensional geometry of the floor marking 154. In this example, processing of this data by a computer system 110 installed on board the vehicle 100 and / or an external computer system connected to the vehicle 100 via a computer network could internally define a ground surface 150 derived from the two-dimensional geometry of the ground marker 154, e.g., enclosed by the ground marker 154. It is understood that the scenario illustrated in FIG. 1 of an internal definition of the floor surface 150 by analysis of a floor marking 154 is purely exemplary and serves only for illustration. However, the ground surface 150 could be defined based on other structural indications detected using sensors of the vehicle 100. In particular, a definition of the floor surface 150 could be made in the absence of a floor marking 154. Alternatively, the ground surface 150 could be specified by an external computer system (e.g. a server of a parking garage or traffic guidance system) and communicated, e.g. using a corresponding set of coordinates, via a computer network to a computer system 110 of the vehicle 100.The approach of the vehicle 100 to the ground surface 150 could end in a parking position 152 on the ground surface 150, which is illustrated in the example of FIG. 1 by a dashed contour of the vehicle 100 and on which the vehicle 100 could come to a standstill. For example, by appropriate programming of the computer system 110, the vehicle 100 could be configured to acquire first ground information characterizing the ground surface 150 by means of the first sensor system 120 before reaching the parking position 152. For example, by detecting a difference between the first ground information and second ground information detected after parking the vehicle 100 on the ground surface 150 before leaving the parking position 152 (see below), the computer system 110 could detect whether an object is located in the space between an underbody of the vehicle 100 and the ground surface 150 (surveillance area).The data acquired in the example described above, which represent the two-dimensional geometry of the ground marking 154, could be used further for this purpose as part of the first ground information, and / or first ground information, which characterizes the ground surface 150, could additionally be acquired using the first sensor system 120. The acquisition of the first ground information by means of the first sensor system 120 is referred to herein as first acquisition. The first piece of ground information could be acquired, for example, at a point in time when the vehicle 100 is in the position shown in FIG. 1, and / or during the approach to the ground surface 150 shown by the arrow. From the first ground information, additional information characterizing the ground surface 150 at the time of the first detection could additionally be derived arithmetically and / or classifying, such as a ground condition of the ground surface 150 (e.g. a height profile, a ground material and / or a time-variable ground state). The sensorially acquired first ground information and / or the information possibly derived therefrom could be stored on board the vehicle 100 and / or in an external computer system for later use.The vehicle 100 could be further configured to recognize a situation in which departure from the parking position 152 is expected. This could be done, for example, by receiving data and / or another corresponding signal (e.g., requesting the vehicle 100 to autonomously return to a place where waiting passengers can access; operation by a driver of the vehicle, e.g., disabling and reactivation of the vehicle 100, entering, buckling, enabling onboard electronics or a drive of the vehicle 100; detection of a traffic light changing from red to red-yellow; clearing a traffic gap, etc.) The computer system 110 could then use a second sensor system 130 (e.g., ultrasonic and / or radar sensors) of the vehicle 100 to acquire (second acquire) the aforementioned second ground information characterizing the ground surface 150 at the time of the second acquisition. From the second ground information, additional information characterizing the ground surface 150 at the time of the second capturing could additionally be derived arithmetically and / or classifying, such as a ground condition of the ground surface 150 as described above.The ground information acquired with the sensor systems 120, 130 and / or the ground information possibly derived from these could finally be used by a classifier learned with a suitable dataset to determine a classification indicating whether an object is likely to be present in the monitoring region between the ground surface 150 and the underbody of the vehicle 100. The departure from the parking area 152 by the vehicle 100 and / or further reactions initiated by the computer system 110 (e.g. a warning that causes a passenger of the vehicle 100 or a person located in a surrounding area of the vehicle 100 to check the monitored area and optionally to remove the object from the monitored area) could be based on the classification thus obtained. The classification could allow the computer system 110 to assess whether sensory differences between the first and second ground information are due to the presence of an object or due to changes in the ground surface 150 not caused by an object (e.g., wet, snowmat, movements of plants).FIG. 2 shows a block diagram of a selection of certain components of an example vehicle 100 suitable for explaining the component disclosed herein. The vehicle 100 could be equipped with a computer system 110 configured to control a first sensor system 120, a second sensor system 130, and optionally further vehicle components 140. For this purpose, the computer system 110 could have an interface, not shown, which is connected to corresponding interfaces, likewise not shown, of the sensor systems 120, 130 and of the further vehicle components 140 via a bus 102 or another in-vehicle communication system.The computer system 110 could further include a processor 112 and a memory 114 operatively connected to the processor 112. The processor 112 shown in the drawing could represent a totality of a plurality of processors and / or a plurality of processor cores. Likewise, the memory 114 shown in the drawing could represent an overall arrangement of various storage systems and storage media such as volatile memory, in particular random access memory (RAM), nonvolatile memory, in particular hard disk(s) and / or solid state memory, and / or removable data carriers.Said configuration of the computer system 110 for controlling the first sensor system 120, the second sensor system 130, and optionally further vehicle components 140 could be realized by appropriate programming of the memory 114 with software 116 which is executed by the processor 112. The software 116 could include a variety of software components, including but not limited to one or more operating systems, firmware, programs, program modules, plug-ins, libraries, virtual machines, etc. As an example, the drawing extracts from the entirety of the software 116 a program 118 including code executable by the processor 112 that, when executed by the processor 112, causes the computer system 110 to perform the method disclosed herein for monitoring a monitoring area located between the ground surface 150 and the sub-floor of the vehicle 100 parked on the ground surface 150 in the storage position 152.In the purely illustrative example of FIG. 2, the first sensor system 120 could include a sensor 122 that the program 118 uses to acquire the first ground information. The second sensor system 130 could also include, by way of illustration only, two sensors 132, 134 that the program 118 uses to acquire the second ground information. It should be appreciated that the sensor system 120 and the sensor system 130 could include the same and / or different sensors and / or sensor types. In particular, the sensor systems 120, 130 could share one or more sensors onboard the vehicle 100. However, it should be noted that the two sensor systems 120, 130 are different, that is to say that at least one of the sensor systems 120, 130 has at least one sensor which differs from the sensors of the respective other sensor system 130, 120. Each sensor system 120, 130 may, taken alone, include sensors of different sensor types and / or include sensors implementing sensor types not included in the other sensor system 130, 120, respectively. Preferably, the sensor systems 120, 130 could differ in their respective installation position within the vehicle 100 or comprise at least one sensor which is not installed at the same installation location of the respective other sensor system 130, 120. In an example application, the first sensor system 120 includes a front camera and a rear camera, each operating in the visual spectral range, and the second sensor system 130 includes one or more pairs of ultrasonic sensors integrated into or mounted to the underbody of the vehicle 100.The software 116 could include other programs configured to control other components 140 and / or functions of the vehicle 100. Without limitation, the other vehicle components 140 could include, for example, a braking system operable, for example, by a brake assist included in the software 116; a steering system operable, for example, by a lane keeping assist included in the software 116; a drive operable, along with the steering system and the braking system, by an autonomous parking and unparking module included in the software 116; an onboard computer operable by a user interface function included in the software 116; another sensor system providing additional measurement data and information to the software 116 programs; etc.FIG. 3 shows a flow chart illustrating an example flow 300 of the computer-implemented method disclosed herein for monitoring a monitoring area lying between a ground surface 150 and an underbody of a vehicle 100 parked on the ground surface 150 in a parking position 152 (e.g., parking or holding position). The sequence 300 could be implemented, for example, by the program code 118.The process 300 includes acquiring 302 first ground information, which represents measurable properties of the ground surface 150, with one or more sensors 122 of the first sensor system 120. The first piece of ground information could comprise first measurement data measured with the sensor 122 in unprocessed or preprocessed (e.g. scaled, standardized) form and / or data derived from the unprocessed or preprocessed first measurement data, such as, for example, arithmetically (e.g. height profile) and / or classifierally (e.g. ground material, ground state) determined characteristics of the ground surface 150. The first ground information could be stored 304, for example, in the memory 114 of the computer system 110 executing the method and / or in another storage medium located onboard the vehicle 100, or a storage system (e.g., cloud storage) connected to the computer system 110 via a computer network, as reference information.The program 118 could be further configured to determine 306 whether the vehicle 100 is expected to leave the parking position 152. This could be detected, for example, with the aid of the first sensor system 120, the second sensor system 130, further sensor systems contained in the further vehicle components 140, and / or on the basis of an event received, for example, from a computer network or generated by a user (e.g. driver) of the vehicle 100, on the basis of a correspondence with one or more predefined patterns or instructions. If there is no such match or instruction, then the determination 306 could be repeated.If, on the other hand, the presence of a situation is detected 306 that requires the vehicle 100 to leave the parking position, then the sequence 300 could continue to record 308 second ground information that reflects measurable properties of the ground surface 150, using one or more sensors 132, 134 of the second sensor system 130. The second ground information could comprise second measurement data measured with one or more of the sensors 132, 134 in unprocessed or preprocessed (e.g. scaled, normalized) form and / or data derived from the unprocessed or preprocessed second measurement data, such as, for example, arithmetically (e.g. height profile) and / or classifierally (e.g. ground material, ground state) determined characteristics of the ground surface 150.The reference information and the second ground information could then be fed as input to a classifier, which is implemented by the program 118, for example. The classifier could be, for example, an artificial neural network, in particular a convolutional neural network (CNN), or a support vector machine (SVM), and trained with a training data set that contains combinations of individual data sets from first training data corresponding to the reference information and from second training data corresponding to the second ground information, together with an assignment of a classification result corresponding to each combination, which specifies whether the training data sets of the respective combination contain features that indicate the presence of an object in the monitoring area of a vehicle.Based on its state corresponding to the training data, the classifier could achieve a classification result (i.e., a classification) that indicates whether the reference data received as input and the second ground information include features that indicate the presence of an object in the monitoring area of the vehicle 100. Based thereon, the program 118 could cause the computer system 110 to control the vehicle 100 in the sense of a reaction corresponding to the classification (e.g. leaving the parking position 152 if there is probably no object within the monitoring area; remaining on the parking position 152 if there is probably an object within the monitoring area, for example also with generation of a warning notice perceptible to a user / driver of the vehicle 100, communication of a message to an external computer system achievable via a computer network, in particular a mobile computer system of a passenger located in or on the vehicle 100 or waiting on the vehicle 100, etc.).Although the invention has been illustrated and described in detail in the drawings and the foregoing description, this illustration and description is to be considered as exemplary and not restrictive; the invention is not limited to the disclosed embodiments.Those skilled in the art will understand that aspects of the present invention may be embodied as an apparatus, method or computer program or computer program product. Accordingly, aspects of the present invention may take the form of a hardware-only embodiment, a software-only embodiment (including firmware, software in memory, microcode, etc.), or a software and hardware aspect combining embodiment, all of which may be generally referred to herein as a "circuit," "module," or "system.". Furthermore, aspects of the present invention may take the form of a computer program product carried by one or more computer readable media in the form of computer executable code. A computer program also comprises the computer-executable code. "Computer-executable code" can also be referred to as "computer program instructions".Any combination of one or more computer readable medium(s) may be used. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A "computer readable storage medium" as used herein includes a tangible storage medium that can store instructions executable by a processor of a computing device. The computer readable storage medium may be referred to as a computer readable non-transitory storage medium. The computer readable storage medium may also be referred to as a tangible computer readable medium. In some embodiments, a computer readable storage medium may also be capable of storing data that enables it to be accessed by the processor of the computing device. Examples of computer readable storage media include, but are not limited to, a floppy disk, a magnetic hard disk, a solid state hard disk, flash memory, a USB stick, random access memory (RAM), read only memory (ROM), an optical disk, a magneto-optical disk, and the register file of the processor. Examples of optical disks include compact disks (CD) and digital versatile disks (DVD), for example, CD-ROM, CD-RW, CD-R, DVD-ROM, DVD-RW, or DVD-R disks. The term computer readable storage medium also refers to various types of recording media suitable for being retrieved from the computing device via a network or communication link. For example, data may be retrieved via a modem, via the Internet, or via a local area network. Computer-executable code executing on a computer-readable medium may be transmitted via any suitable medium, including, but not limited to, wireless, wired, optical fiber, RF, etc., or any suitable combination of the foregoing.A computer readable signal medium may include a propagated data signal containing the computer readable program code, for example, in a base signal (baseband) or as part of a carrier signal (carrier wave). Such a propagation signal may be formed in any form including, but not limited to, an electromagnetic form, an optical form, or any suitable combination thereof. A computer readable signal medium may be any computer readable medium that is not a computer readable storage medium and that can transmit, propagate, or transport a program for use by or in connection with a system, apparatus, or apparatus for executing instructions."Computer memory" or "memory" is an example of a computer readable storage medium. Computer memory is any memory that is directly accessible to a processor."Computer data storage" or "data storage" is another example of a computer readable storage medium. Computer data storage is any non-transitory computer readable storage medium. In some embodiments, computer memory may also be computer data storage, or vice versa.A "processor" as used herein includes an electronic component capable of executing a program- or machine-executable instruction or code. Reference to the computing device comprising a "processor" should be interpreted as possibly comprising more than one processor or processing cores. The processor may be, for example, a multi-core processor. A processor may also refer to a collection of processors within a single computer system or distributed among multiple computer systems. The term computing device or computer shall also be interpreted to potentially indicate a collection or network of computing devices or computers, each comprising a processor or processors. The computer executable code may be executed by multiple processors, which may be distributed within the same computing device or even across multiple computers.Computer-executable code may include machine-executable instructions or a program that causes a processor to perform an aspect of the present invention. Computer-executable code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional method oriented programming languages such as the "C" programming language or similar programming languages, and translated into machine-executable instructions. In some cases, the computer-executable code may be in the form of a high-level programming language or in a pre-translated form, and used in conjunction with an interpreter that generates the machine-executable instructions.The computer executable code may execute entirely on a user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter case, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, via the Internet using an Internet Service Provider).The computer program instructions may be executed on one processor or on multiple processors. In the case of multiple processors, these may be distributed among multiple different entities (e.g., clients, servers). Each processor could execute a portion of the instructions provided for the respective entity. Thus, when reference is made to a system or method comprising a plurality of entities, the computer program instructions would be understood to be adapted to be executed by a processor associated with the respective entity.Aspects of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It is noted that each block or part of the blocks of the flowcharts, representations and / or the block diagrams may be executed by computer program instructions, optionally in the form of computer executable code. It is further noted that combinations of blocks in different flowcharts, diagrams, and / or block diagrams may be combined unless they are mutually exclusive. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer or other programmable data processing apparatus to produce an apparatus such that the instructions executed via the processor of the computer or other programmable data processing apparatus produce means for executing the functions / steps specified in the block or blocks of the flowcharts and / or the block diagrams.These computer program instructions may also be stored on a computer readable medium that can control a computer or other programmable data processing devices or other devices to function in a particular manner such that the instructions stored on the computer readable medium produce an article of manufacture including instructions that implement the / n function / step specified in the flowchart and / or block diagram block or blocks.The computer program instructions may also be stored on a computer, other programmable data processing devices, or other devices to cause a series of process steps to be performed on the computer, other programmable data processing devices, or other devices to produce a process performed on a computer, such that the instructions performed on the computer or other programmable devices produce methods for implementing the functions / steps specified in the block or blocks of the flowcharts and / or the block diagrams.LIST OF REFERENCE NUMERALS100 Vehicle 102 Bus 110 Computer system 112 Processor 114 Memory 116 Software 118 Program code 120 Sensor system 122 Sensor 130 Sensor system 132 Sensor 134 Sensor 140 Vehicle components 150 Floor surface 152 Parking position 154 Floor marking 300 Method for monitoring a monitoring area 302 First items of floor information are detected 304 First items of floor information are stored 306 If the vehicle is expected to leave the parking position? 308 Second items of floor information are detected 310 Classification of whether an object is present in the monitoring area

Claims

A computer-implemented method (300) for monitoring a monitoring area lying between a ground surface and an underbody of a vehicle (100) parked on the ground surface in a storage position, the method comprising: first acquiring (302), with a first sensor system (120) arranged on the vehicle (100), first ground information characterizing the ground surface before the vehicle (100) reaches the storage position; storing (304), in a storage system, reference information based on the first ground information; and responsive to a determination (306) that departure from the storage position is expected by the vehicle (100): second acquiring (308), with a second sensor system (130) arranged on the vehicle (100), second ground information characterizing the ground surface; and first classifying (310), by a classifier learned with a data set, on the basis of the reference information and the second ground information, whether an object is present in the monitoring area.The method (300) of claim 1, wherein the first detecting (302) is triggered when the vehicle (100) begins autonomous parking on the ground surface, and / or wherein detecting parking initiated by a driver of the vehicle on the ground surface triggers the first detecting (302).Method (300) according to one of the preceding claims, wherein the determination (306) comprises a deactivation and renewed activation of the vehicle (100) and / or a detection of a predefined traffic situation.Method (300) according to one of the preceding claims, additionally comprising, if the first classification (310) reveals that an object is present in the monitored region, a second classification whether the object interferes with leaving the parking position.The method (300) according to any one of the preceding claims, additionally comprising evaluating the first ground information to determine a first ground condition and / or evaluating the second ground information to determine a second ground condition, wherein the reference information comprises the first ground condition, wherein the first classification (310) is based on the first and the second ground condition, wherein the first and the second ground condition comprise: a first and a second height profile of the ground surface, respectively; and / or a first and a second ground material, respectively, of the ground surface; and / or a first and a second time-variable ground condition, respectively.The method (300) of any preceding claim, wherein the reference information comprises the first ground information.Method (300) according to one of the preceding claims, wherein the classifier comprises an artificial neural network, in particular a convolutional neural network, or a support vector machine.Method (300) according to one of the preceding claims, wherein the first capturing (302) begins before the underbody overlaps with the ground surface, and / or wherein the first ground information is aggregated from a plurality of individual measurements of the first sensor system (120) carried out during a drive of the vehicle (100) to the parking position.Method (300) according to one of the preceding claims, wherein the first sensor system (120) acquires the first ground information with a sensor (122) which implements a sensor type selected from: a sensor type operating in the visual or infrared spectral range, in particular a camera; a sensor type operating according to an echo method, in particular an ultrasonic or radar sensor, and / or in particular configured to acquire the first ground information with a multi-frequency measurement; and a sensor type operating according to a scan method, in particular a lidar sensor or a laser scanner, and / or wherein the second sensor system (130) has an underbody sensor and optionally a vehicle environment sensor.The method (300) of any preceding claim, wherein the first sensor system (120) comprises a sensor (122) implementing a type of sensor that is not implemented by a sensor (132, 134) of the second sensor system (130).A computer program product, in particular a computer-readable storage medium, storing computer-executable code (118), wherein the code (118), when executed by at least one processor (112) of a computer device (110), causes the computer device (110) to execute a method (300) for monitoring a monitoring area lying between a floor surface and an underbody of a vehicle (100) parked on the floor surface in a storage position, wherein the method (300) comprises: first acquiring (302) first ground information characterizing the floor surface with a first sensor system (120) arranged on the vehicle (100) before the vehicle (100) reaches the storage position; storing (304) reference information based on the first ground information in a storage system (114); and in response to a determination (306) that the vehicle (100) is expected to leave the parking position: second acquisition (308) of second ground information characterizing the ground surface using a second sensor system (130) arranged on the vehicle (100); and first classification (310), by a classifier learned using a dataset, based on the reference information and the second ground information, whether an object is present in the monitoring region.A computer system (110) comprising a processor (112) and a memory (114) operatively connected to the processor (112), the memory (114) storing computer executable code (118), the code (118), when executed by the processor (112), causing the computer system (110) to execute a method (300) of monitoring a monitoring area lying between a ground surface and an underbody of a vehicle (100) parked on the ground surface in a storage position, the method (300) comprising: first acquiring (302) first ground information characterizing the ground surface with a first sensor system (120) disposed on the vehicle (100) before the vehicle (100) reaches the storage position; storing (304) reference information based on the first ground information in a storage system; and in response to a determination (306) that the vehicle (100) is expected to leave the parking position: second acquisition (308) of second ground information characterizing the ground surface using a second sensor system (130) arranged on the vehicle (100); and first classification (310), by a classifier learned using a dataset, based on the reference information and the second ground information, whether an object is present in the monitoring region.A vehicle (100) comprising a computer system (110) configured to control components (140) of the vehicle (100), the computer system (110) comprising a processor (112) and a memory (114) operatively connected to the processor (112), the memory (114) storing computer-executable code (118), the code (118), when executed by the processor (112), causing the computer system (110) to execute a method (300) of monitoring a monitoring area lying between a ground surface and an underbody of the vehicle (100) parked on the ground surface in a storage position, the method (300) comprising: first acquiring (302), with a first sensor system (120) arranged on the vehicle (100), first information characterizing the ground surface before the vehicle (100) reaches the storage position; storing (304) reference information based on the first ground information in a storage system; and in response to a determination (306) that the vehicle (100) is expected to leave the parking position: second capturing (308) second ground information characterizing the ground surface using a second sensor system (130) arranged on the vehicle (100); and first classifying (310), by a classifier learned with a dataset, based on the reference information and the second ground information, whether an object is present in the monitoring area.

Images