Method for surveying the surroundings of a motor vehicle

Abstract

The invention relates to a method for surveying the surroundings of a motor vehicle (1), comprising: a) carrying out (S1) a plurality of measurements using ultrasonic sensors at different measurement positions (31-35), comprising emitting an ultrasonic pulse and receiving an ultrasonic signal profile (8); and creating an echo feature, which defines at least one locus curve (21-25) of possible reflection points in the surroundings, on the basis of the measurement position and an echo location (t3) of an echo (81) in the ultrasonic signal profile; b) forming (S2) all possible combinations of a plurality of echo features from the created echo features; c) creating (S3) an object feature candidate (41-44) for each combination by lateralising the locus curves defined by the echo features, wherein the respective object feature candidate comprises at least one position of an object obtained during lateralising; d) determining (S4) an evaluation (110) for each of the object feature candidates determined in step c) on the basis of the information comprised therein and / or in the echo features used for creating said object feature candidate; and e) adopting (S5), as a newly created object feature, the best evaluated of the object feature candidates determined in step d).

Description

[0001] 2023PF00270

[0002] 1

[0003] METHOD FOR SURVEYING THE ENVIRONMENT OF A MOTOR VEHICLE

[0004] The present invention relates to the field of ultrasonic sensor technology for motor vehicles and, more specifically, a method for measuring the environment of a motor vehicle, a corresponding computer program product and a corresponding control device for carrying out the proposed method, and a motor vehicle with the control device.

[0005] Motor vehicles can be equipped with ultrasonic sensors to measure their surroundings. A control device causes an ultrasonic sensor to transmit an ultrasonic pulse into the vehicle's environment and records the received ultrasonic signal using the same or a different ultrasonic sensor. Based on the time difference between the transmission of the ultrasonic pulse and the occurrence of an echo signal, which manifests as a peak in the received ultrasonic pulse, and using the known positions of the transmitting and receiving ultrasonic sensors, a locus of possible reflection points in the vehicle's environment can be determined.In the general case, this locus is an ellipse with the ultrasonic sensors at the foci; if the transmitting and receiving ultrasonic sensors are identical, the locus is a circle as a special case of an ellipse with identical semi-axes, whose radius is exactly half the signal path distance (speed of sound times signal travel time).

[0006] It is also known to construct object features based on ellipses using a lateration method, which specify the position and extent of obstacles in the vicinity of the vehicle. These object features can be used by an assistance system for semi- or fully autonomous driving or parking for navigation, or taken into account by a door opening assistant to decide whether and to what extent a vehicle door can be opened automatically. 2023PF00270

[0007] 2

[0008] Here, a point feature can be constructed at the intersection of several ellipses, describing a post-like obstacle. Similarly, a tangent can be drawn to several ellipses, and a line feature can be constructed that describes a wall-like obstacle. When dealing with a large number of ellipses, it is often difficult to automatically and correctly determine which ellipses can be combined to form an object feature of which type, in order to obtain a feature that corresponds to reality—that is, to distinguish between physically meaningful or realistic combinations of ellipses and those that are physically meaningless or unrealistic.

[0009] DE 10 2021 121 155 A1 discloses an ultrasonic sensor system that determines distance values ​​based on ultrasonic echoes detected by at least four ultrasonic sensors, determines solutions from these distance values ​​using a trilateration method, filters each of these solutions to filtered solutions using a Kalman filter method, and clusters the filtered solutions to acceptable solutions using a clustering method and discards unacceptable filtered solutions.

[0010] WO 2012 / 190712 A1 discloses a method for detecting at least one object in the vicinity of a motor vehicle, in which ultrasonic signals are emitted, echoes in reflected ultrasonic signals are received, distance values ​​are determined on the basis of the received echoes, and at least one object is detected based on a determined variance of the distance values ​​of a group of several temporally successive echoes.

[0011] US Patent 2023 / 0375702A1 discloses a method for measuring the lateral surroundings of a vehicle, in which a set of reflection points is formed from echo signals identified in measured received signal waveforms by trilateration, and then pairs of reflection points are formed based on a position-based criterion. Reflection points for which a pair can be formed are determined to be reflection points at a high object, while reflection points for which no pair can be formed are determined to be reflection points at a low object. 2023PF00270

[0012] 3

[0013] Against this background, one object of the present invention is to provide an improved method for determining object features in the environment of a motor vehicle.

[0014] According to one aspect, a method for measuring the environment of a motor vehicle equipped with a number of ultrasonic sensors is proposed. The proposed method comprises: a) performing several measurements with the number of ultrasonic sensors at different measurement positions, each measurement comprising: emitting an ultrasonic pulse with one of the number of ultrasonic sensors; determining an echo in an ultrasonic signal waveform received by one of the ultrasonic sensors in response to the emission of the ultrasonic pulse; and creating an echo feature formed from information describing at least one locus of possible reflection points in the environment of the motor vehicle, based on the measurement position and an echo position of the echo in the ultrasonic signal waveform; b) exhaustively forming combinations of several echo features from at least a subset of the echo features created in step a).c) Creating an object feature candidate for the respective combination by lateralizing the locus curves defined by the echo features, wherein the respective object feature candidate is formed from information that includes at least one presumptive position of an object in the vicinity of the motor vehicle obtained as a result of the lateralization; d) Determining a score for each of the object feature candidates determined in step c) based on the information contained in the object feature candidate and / or in the echo features used to create the object feature candidate; and e) Adopting the object feature candidate determined in step d) that has the best score, provided that the best score meets a predefined minimum criterion, as a newly created object feature.

[0015] Accordingly, a brute-force approach is proposed in which an exhaustive number of combinations of all or a significant subset of all created echo features are formed, an object feature candidate is created for each combination and evaluated. 2023PF00270

[0016] 4 and the object characteristic candidate with the best rating is adopted as the newly created object characteristic candidate.

[0017] It has been shown that such an essentially exhaustive brute-force approach advantageously produces more realistic object features than selecting and combining only individual echo features for lateration using mathematical filtering, clustering, or other ontological criteria.

[0018] The brute-force approach has proven particularly advantageous in a door-opening assistance scenario where the vehicle is stationary and the objective is to determine whether there is merely a point-like post at a specific position in front of the vehicle's side doors, or an extensive linear wall completely blocking the doors. In this scenario, the proposed brute-force approach, with its exhaustive evaluation of all possible combinations, can yield significantly improved results.

[0019] The term "exhaustive formation" can encompass forming all possible combinations of all created echo features, or forming all possible combinations of at least the subset of the created echo features, or at least forming a significant number among all possible combinations of the created echo features.

[0020] The term "subset" is not intended as a limitation of the exhaustive brute-force approach. However, the "subset" can be defined to keep the computational effort for the exhaustive search within a reasonable range. The term "subset" should be interpreted such that a multitude of different combinations of echo features can still be generated within the subset. That is, if combinations of at least n echo features are to be generated, the subset comprises at least n+1 echo features, but preferably n+m echo features, with m»n. Within the subset, an exhaustive evaluation of a multitude of [2023PF00270] then continues.

[0021] 5 possible combinations. The "subset" is selected, for example, based on a criterion that excludes echo feature combinations where a good rating cannot be expected a priori, and includes echo feature combinations where a good rating is potentially achievable.

[0022] The subset comprises at least four of the echo features created in step a), but preferably the subset comprises at least 10%, more preferably at least 50%, and even more preferably at least 70%, and most preferably all of the echo features created in step a).

[0023] When creating exhaustive combinations, a restriction may also apply regarding a minimum and a maximum number of echo features that can be combined. The minimum number is 2 or preferably 3, the maximum number is not specifically restricted and may, for example, be 5, preferably 8 or 10 or more.

[0024] The motor vehicle can be, for example, a passenger car or a commercial vehicle, such as a truck, a bus or a tractor.

[0025] The ultrasonic sensor can, for example, comprise an ultrasonic membrane, a piezoelectric element mounted internally on the ultrasonic membrane, and a control circuit for the piezoelectric element. The ultrasonic sensor can be arranged in a recess of an external body component of the vehicle, such as a vehicle door, bumper, or fender, such that the ultrasonic membrane is flush with an outer surface of the vehicle body. Alternatively, the ultrasonic sensor can be mounted internally on a body panel, such as the vehicle door or fender, with the ultrasonic membrane of the sensor being acoustically coupled to the body panel.

[0026] Controlled by a control device of the motor vehicle, the control circuit can

[0027] Piezoelectric element and, via the piezoelectric element, the ultrasound membrane is set into vibration - 2023PF00270

[0028] 6. and in this way emit an ultrasonic pulse into the vicinity of the vehicle. When a reflected signal component of such an emitted ultrasonic pulse reaches the ultrasonic membrane again from the surroundings, this reflected signal component excites the ultrasonic membrane and, via the ultrasonic membrane, the piezoelectric element to vibrations, which are converted by the piezoelectric element into an electrical signal, so that the control device can receive and detect an ultrasonic signal profile after the ultrasonic pulse has been emitted, using the control circuit.

[0029] An "ultrasound signal waveform" is understood here to be a processable and analyzable representation of the corresponding signal over a predefined time period. For example, the ultrasound signal waveform could consist of samples of the respective signal at a suitable sampling frequency, such as in the range of a few megahertz. If the ultrasound signal waveform includes all samples acquired at a frequency of a few megahertz, it is also referred to as a "raw ultrasound signal waveform." However, the ultrasound signal waveform need not be a raw ultrasound signal waveform; it could also be a derived ultrasound signal waveform.

[0030] The signal waveform derived from the ultrasound signal can be obtained by processing the raw ultrasound signal. To reduce the processing load, the signal waveform derived from the ultrasound signal can, for example, be a downsampling of the raw ultrasound signal within the specified time period, based on a resolution of, for example, approximately 50 to 200, preferably approximately 100 ps per sample. It is also conceivable to downmix the raw signal, for example, in an IQ mixer implemented as a DSP; in this case, the signal waveform derived from the raw ultrasound signal can, for example, include the downmixed i-signal waveform and the downmixed q-signal waveform within the specified time period.Downsampling or downmixing can be performed by the control device to further reduce the processing load of the control device; especially 2023PF00270.

[0031] 7 Preferably, however, the downsampling or downmixing can be performed by an ASIC or other circuit of the ultrasonic sensor, so that transmission bandwidth is also saved on a bus or other transmission link via which the ultrasonic sensor is connected to the control device.

[0032] The ultrasound signal can be present as a signal, for example as an analog or digital signal, i.e. as a signal in transit, or the ultrasound signal can be present as static data (a sequence of samples stored in working memory or the like).

[0033] The term "echo" here describes a signal superimposed on the ultrasound signal waveform, which can be interpreted as a reflection of the emitted ultrasound pulse and manifests itself particularly as a peak in the ultrasound signal waveform. A peak in the ultrasound signal waveform denotes a local maximum in the amplitude of the received ultrasound signal waveform and / or its envelope. That is, a necessary condition for the presence of a peak is, for example, that the amplitude of a sample identified as a peak is higher than the amplitudes of the preceding or subsequent samples of the ultrasound signal waveform. However, further criteria, such as a minimum height, a minimum width, a shape criterion, a minimum interval since the emission of the ultrasound signal, and the like, can be used to identify actual peaks and reduce noise.

[0034] The term "echo position" describes, in particular, the location or time of occurrence of an echo within the ultrasound signal waveform. For example, the echo position can indicate the signal transit time between the emission of the ultrasound pulse and the occurrence of the echo within the ultrasound signal waveform. The distance traveled by the ultrasound signal can be determined from the signal transit time by multiplying the signal transit time by the speed of sound in air. In this respect, the echo position can also be a measure of the ultrasonic waveform.

[0035] The distance traveled by the sound signal is the signal path distance. Whether the echo position is processed internally by the control device as signal travel time or as signal path distance is irrelevant here; these are equivalent configurations of the feature "echo position".

[0036] Each of the measurement positions is defined by a position of the sending and a position of the receiving ultrasonic sensor.

[0037] If, in particular, the transmitting and receiving ultrasonic sensors are the same ultrasonic sensor, the measurement position is a position of the ultrasonic sensor in question. In this case, the locus of possible reflection points in the vicinity of the vehicle, defined by the echo characteristic, is a circle (a special case of an ellipse) around the measurement positions, the radius of which is half the signal path length specified by the echo position.

[0038] If, in particular, the transmitting and receiving ultrasonic sensors are different ultrasonic sensors, the measurement position is the center point of an ellipse defined by the set of all points for which the sum of the distances to the positions of the transmitting and receiving ultrasonic sensors located at the foci of the ellipse is equal to the signal path length specified by the echo position. In this case, the ellipse is the locus defined by the echo feature.

[0039] In this context, "latering" means any suitable method based on the basic idea of ​​determining at least one position of a reflection point in the vicinity of the motor vehicle based on at least two distance measurements and the associated measurement positions.

[0040] The terms "echo feature," "candidate object feature," and "object feature" refer specifically to data (measurement data) provided by the proposed method as an intermediate or final result. Here, a "candidate object feature" is a feature that has not yet been confirmed, while an "object feature" is a feature that has been confirmed to be, with a high degree of probability, an obstacle in the environment. 2023PF00270

[0041] 9 of the motor vehicle describes, and which can subsequently be evaluated by an assistance function of the motor vehicle, such as a parking assistant, driving assistant, door opening assistant or the like.

[0042] The respective echo feature includes at least information defining the locus of possible reflection points, such as the measurement position and signal path length. Optionally, the respective echo feature may contain further information obtained during step a), which will be discussed below with reference to preferred embodiments.

[0043] The respective object feature and object feature candidate each comprise at least information indicating the presumed position of an object in the vicinity of the motor vehicle. The respective object feature may optionally contain further information obtained during steps a) to c), which will be discussed below with reference to preferred embodiments.

[0044] The evaluation can relate to any aspect of the information contained in the candidate features. A precision measure obtained during laterality in the form of a squared error or the like is an example of a criterion suitable for evaluation. Further and / or alternative criteria are discussed below with reference to preferred embodiments. The evaluation can be presented as numerical information, which can be normalized to a predefined interval, and a convention can specify whether increasingly high numerical values ​​represent increasingly good or increasingly poor evaluations.

[0045] The predefined minimum criterion can therefore be a numerical threshold, and an object feature candidate can be adopted as an object feature only if the threshold is exceeded or only if it is not met, depending on the chosen convention. A suitable value for the threshold can be determined through simple experiments. This advantageously avoids the creation of implausible object features. 2023PF00270

[0046] 10

[0047] Object feature candidates that are not adopted in step e) can be discarded.

[0048] In this context, "number" means a number of N>1, that is, a number of one or more elements. In contrast, "plurality", "multiple", "several", and the like mean a plurality of N>1, that is, two or more elements.

[0049] According to one embodiment, the method further comprises: f) discarding the echo features used in step e) to form the object feature candidate adopted as the newly created object feature and repeating steps b) to f) until no further object feature candidate is adopted as the newly created object feature in step e).

[0050] Accordingly, further object features can be determined from specific echo features in an iterative process, which is important because there could be more than one object in the vicinity of the vehicle. However, it is avoided that an echo feature already used to create an initial object feature is also used to create further object features. This prevents results that contradict physical reality and allows the created object features to be even better adapted to reality.

[0051] According to one interpretation, the evaluation is based at least on a precision measure of the candidate object characteristic as a criterion.

[0052] A precision measure is obtained particularly when laterally determining the presumed position of the object in the vicinity of the motor vehicle and indicates, in particular, how precisely the lateration could be carried out, that is, how exactly the locus curves to be latered 2023PF00270

[0053] 11. Intersect and / or adapt to a tangent. The precision measure can, in particular, be an error measure obtained by iteratively adapting the latered position to the loci, or be derived from such an error measure.

[0054] Accordingly, particularly precise object feature candidates can be advantageously preferred over imprecise ones.

[0055] According to a further embodiment, in step c) for the respective combination: depending on at least one deviation of the loci defined by the echo features from a best possible intersection point of the loci and depending on one deviation of the loci from a tangent line best adapted to the loci, either a point-like object feature candidate at the position of the best possible intersection point or a line-like object feature candidate at the position of the best possible tangent line is determined, and the created object feature candidate further includes as information a specification of a type of object feature candidate as point-like or line-like and the precision measure, wherein the precision measure specifies the deviation of the loci from the best possible intersection point or from the best possible tangent line.

[0056] Accordingly, it is advantageous to automatically decide for each of the combinations, based on the precision measure, whether a point-like or a line-like object feature should be created, and preference can be given to the more precise of the two object features, whereby the precision measure obtained in this way is stored in the object feature and can subsequently be used for evaluation.

[0057] A point feature describes, in particular, an essentially point-like object, such as a post. A line feature describes, in particular, an essentially line-like object, such as a wall. 2023PF00270

[0058] 12

[0059] "Best possible" in this context means that, generally, more than two loci do not intersect perfectly at a single point, and / or a perfect tangent line cannot be drawn between more than two loci. In this case, the intersection point and / or the tangent line can be determined approximately using an analytical or numerical iterative method, such as a gradient-based method or another optimization technique. The numerical iterative method attempts to minimize the deviation of the loci from a current candidate for the intersection point or the tangent line. The resulting residual deviation is then used as a precision measure for the accuracy of the approximately determined intersection point or tangent line.

[0060] The information may also include details about the possible extent of the hypothesized object. In the case of a point-like object feature candidate, this information can be derived from the precision measure. In the case of a line-like object feature candidate, this information can be obtained from the distance between the points of tangency (or approximate points of tangency) of the tangent line with the outermost loci.

[0061] It should be noted that in the present embodiment, the number of individual ratings can further include an assessment of the plausibility of the occurrence of a particular object feature type at the presumed position. For example, line features may be preferred (assigned a better rating than point features) at short distances from the vehicle, and point features may be preferred (assigned a better rating than line features) at greater distances from the vehicle.

[0062] According to a further embodiment, the echo feature created in step a) also includes the following information: the quality of the echo determination; and / or the type of the determined echo. 2023PF00270

[0063] 13

[0064] A measure of the quality of the echo determination can be determined, for example, by means of a signal-to-noise ratio of the associated peak, a signal shape of the associated peak, or the like.

[0065] Accordingly, the quality of the echo characteristics used to create a respective object characteristic candidate can advantageously be incorporated into the evaluation of the object characteristic candidates.

[0066] A specific echo type can, for example, assume one of the values ​​"single echo" / "low obstacle" or "double echo" / "high obstacle". A "double echo" describes a case in which, in the subsonic signal waveform, a further echo occurs temporally after the (first) echo, the peak of which is preferably in a predetermined amplitude ratio to the peak of the first echo. In this case, it is assumed that multiple signal propagation paths exist: a first signal propagation path that led directly from the transmitting ultrasonic sensor to the object and back to the receiving ultrasonic sensor, resulting in the first echo, and a second signal propagation path that led from the transmitting ultrasonic sensor to the object, from there to the ground, and from there back to the receiving ultrasonic sensor, resulting in the second echo.In this case, the echo position is determined based on the signal travel time of the first echo, and the type of the determined echo is defined as "double echo" and / or "high obstacle".

[0067] Accordingly, object feature candidates created using echo features of the same type can advantageously be evaluated better than object feature candidates created using echo features of different types.

[0068] According to one embodiment, step a) is repeated in several successive measurement cycles, the echo features created in each step a) are retained over several measurement cycles, and one of the echo features created in a first of the measurement cycles is updated in a subsequent second measurement cycle, 2023PF00270

[0069] 14 if another echo corresponding to the echo characteristics created in the first measurement cycle is determined.

[0070] In particular, echo characteristics are updated if the echo was determined at the same measurement position in the subsequent measurement cycle and in the first of the measurement cycles.

[0071] In particular, an echo feature for which no corresponding echo has been determined over a predetermined number of measurement cycles can be discarded. Accordingly, the stored set of echo features can be purged of outdated echo features, ensuring that only relatively current echo features remain.

[0072] Accordingly, it is advantageous to maintain a large number of echo features over a certain period of time, and echoes determined in different measurement cycles can be associated with one another. In this way, even if not every obstacle is "seen" in every measurement cycle, all the necessary information can be obtained by temporally summarizing echo features from several measurement cycles.

[0073] Association, that is, determining that an echo in a subsequent measurement cycle corresponds to an echo feature from a previous measurement cycle, can be done by comparing similarities regarding measurement position, echo location, echo velocity, signal shape, predictions, and the like.

[0074] According to a further embodiment, in step b) the respective subset comprises at least one echo feature for which a corresponding echo has been determined in a current measurement cycle.

[0075] This means that in the exhaustive brute-force approach, only combinations from a subset of the specified echo features are formed, in which at least one current echo feature is contained. Accordingly, the computational effort for the 2023PF00270 can be advantageously reduced.

[0076] 15

[0077] The execution of the brute-force approach can be reduced by excluding object feature candidates that would be based on combinations of exclusively outdated echo features and would therefore be rated low anyway, from the brute-force test and not creating them in the first place.

[0078] According to a further embodiment, the created or updated echo feature also includes the following information: an indication of a measurement cycle in which the echo feature was last created or updated; and / or an indication of a number of measurement cycles in which an echo corresponding to the echo feature was determined; and / or a speed at which the respective echo corresponding to the echo feature has moved over several measurement cycles; and / or a prediction accuracy that indicates a degree of agreement between a locus defined by the updated echo feature and a predicted locus predicted based on the speed at the time the echo corresponding to the echo feature was last determined.

[0079] All the aforementioned information is available when echo features, or the locus curves defined by them, are laterated as described, and echo features from different measurement cycles are associated with one another. Advantageously, a criterion for evaluation can be formulated for each of these pieces of information, allowing for a more informed assessment.

[0080] According to a further embodiment, in step d) the evaluation is further carried out on the basis of one or more of the following criteria: a degree of recency of the object feature candidate based on the measurement cycles in which the respective echo feature used to create the object feature was last updated; and / or a degree of reliability of the object feature candidate based on the number of measurement cycles in which an echo corresponding to the respective echo feature used to create the object feature candidate was determined; and / or a degree of agreement of the speeds of the objects used to create the object feature candidate 2023PF00270

[0081] 16 echo features used; and / or the predictive accuracy of the echo features used to create the object feature candidate.

[0082] Thus, it is advantageous to award higher scores to object feature candidates based on one or more echo features updated in the current measurement cycle; that is, the use of current echo features can be rewarded. Likewise, object feature candidates based on one or more echo features that have already been updated in several measurement cycles can be awarded higher scores; that is, the use of echo features that can be considered reliable because they have already been seen, for example, 2, 3, 4, or 5 times or more, can be rewarded. Penalties can be imposed if echo features with different speeds have been combined, for example, an echo that has moved towards the vehicle over several measurement cycles and an echo that has moved away from the vehicle over several measurement cycles.Furthermore, the use of echo features whose association is questionable because a predicted echo position does not correspond to the actually determined echo position can be punished.

[0083] According to a further embodiment, in step d) the evaluation is an overall evaluation which is determined on the basis of a plurality of individual evaluations of a respective criterion.

[0084] The plurality of individual evaluations may include an evaluation of a precision measure of the object feature candidate, an evaluation of the quality of the determination of the echo on which the respective echo feature used to create the object feature candidate is based, and / or an evaluation of a degree of agreement between the qualities and / or types of echoes of the echo features used to create the object feature candidate. Furthermore, the plurality of individual evaluations may include a separate evaluation of each of the criteria discussed for the preceding embodiments and / or each 2023PF00270

[0085] 17 of the information discussed for the preceding embodiments, included in the object feature candidates and / or the echo features used to create them.

[0086] By determining an evaluation based on a plurality of individual evaluations, these individual evaluations can mutually compensate for or complement each other, thus resulting in a more balanced overall evaluation. In other words, unlike a known approach that uses fixed rules to decide whether or not to create or adopt a candidate object attribute, this approach is flexible. A poor individual evaluation according to one criterion can be compensated for by a good individual evaluation according to another criterion, or conversely, a good individual evaluation according to one criterion can be invalidated by a poor individual evaluation according to another criterion.In this way, within the framework of the exhaustive evaluation approach, a number of object feature candidates that actually best match physical reality can be determined more reliably than with rule-based or filter-based approaches.

[0087] According to another embodiment, the overall rating is formed by weighting and summing the majority of individual ratings.

[0088] The individual ratings can also be normalized before being weighted and summed.

[0089] Standardization can involve standardizing to a uniform scale, such as an integer scale from 0 to 100 or a flow-chamber scale from 0 to 1. Standardization can also involve inverting individual scales so that a uniform convention is established for all individual ratings, where high or low values ​​always mean the same thing, e.g., always good or always bad ratings. This advantageously makes the various individual ratings comparable. 2023PF00270

[0090] 18

[0091] Weighting can involve multiplying each individual assessment by a predefined weighting factor to give certain criteria more weight than others. By appropriately selecting the weighting factors, which can be determined by a person skilled in the art through routine experiments for the specific application, the object characteristics generated by the proposed method can be optimally adapted to physical reality.

[0092] The overall rating can then be calculated as the sum of the weighted and normalized individual ratings. Accordingly, an advantageous method for the automated, objective calculation of the overall rating from the individual ratings is presented.

[0093] According to a further embodiment, the proposed method further comprises the following step, preferably performed in advance at the factory: determining weights of the individual ratings using an iterative procedure based on numerical optimization or machine learning or a neural network, wherein the determination of the weights is carried out under controlled conditions, and wherein a deviation of object characteristics obtained with respective weights chosen within the iterative procedure from known actual objects in the environment of the motor vehicle is used as feedback for the method used to determine the weights.

[0094] Controlled conditions refer in particular to factory conditions or conditions in a workshop where the positions of objects in the vicinity of the vehicle are precisely known.

[0095] By using a computerized iterative procedure to "try out" a large number of possible weightings of the individual ratings, it is advantageously possible to quickly and automatically find optimal weightings that better align the object characteristics provided by the proposed procedure with reality. 2023PF00270

[0096] 19

[0097] Furthermore, under a second aspect, a computer program product is proposed which includes commands that, when the program is executed by a control device of a motor vehicle with a number of ultrasonic sensors, cause the control device to execute the procedure described above.

[0098] The computer program product, such as a computer program tool, can be provided or delivered, for example, as a storage medium such as a memory card, USB stick, CD-ROM, DVD, EPROM, EEPROM, Flash-ROM, SD card, or as a downloadable file from a server on a network. The download can be carried out, for example, over a wired or wireless communication network by transmitting the corresponding file along with the computer program product or tool.

[0099] According to a third aspect, a control device for a motor vehicle equipped with a number of ultrasonic sensors is proposed. The control device is designed to execute the proposed method.

[0100] The control device can comprise a number of functional units that correspond to the steps of the proposed procedure.

[0101] The respective functional unit can be implemented in hardware and / or software. In a hardware implementation, the unit can be, for example, a computer or a microprocessor. In a software implementation, the unit can be a computer program, a function, a routine, an algorithm, part of program code, or an executable object. Furthermore, each of the aforementioned units can also be part of a higher-level vehicle control system, such as a central electronic control unit and / or an engine control unit (ECU). 2023PF00270

[0102] 20

[0103] According to a fourth aspect, a motor vehicle is proposed with a number of ultrasonic sensors and the proposed control device.

[0104] The embodiments, advantages and features described for the proposed method of the first aspect apply accordingly to the proposed computer program product of the second aspect, the proposed control device of the third aspect and the proposed motor vehicle of the fourth aspect.

[0105] Other possible implementations of the invention also include combinations of features or embodiments described previously or subsequently with regard to the exemplary embodiments, even if not explicitly mentioned. In such cases, the person skilled in the art will also add individual aspects as improvements or additions to the respective basic form of the invention.

[0106] Further advantageous embodiments and aspects of the invention are the subject of the dependent claims and the exemplary embodiments of the invention described below. The invention will be explained in more detail below with reference to preferred embodiments and the accompanying figures.

[0107] Fig. 1 shows a motor vehicle with several ultrasonic sensors according to an exemplary embodiment from the side;

[0108] Fig. 2 illustrates steps of a method according to the exemplary embodiment;

[0109] Fig. 3 shows functional units of a control device according to the exemplary embodiment;

[0110] Fig. 4 shows a plot of a received ultrasound signal waveform; 2023PF00270

[0111] 21

[0112] Fig. 5 shows exemplary loci of possible reflection points in a top-down view;

[0113] Fig. 6 shows exemplary loci of possible reflection points and two point-like

[0114] Object feature candidates;

[0115] Fig. 7 shows exemplary loci of possible reflection points and a point-like object feature candidate;

[0116] Fig. 8 shows exemplary loci of possible reflection points and a linear object feature candidate;

[0117] Fig. 9 shows an enlarged view of Fig. 7;

[0118] Fig. 10 shows envelopes of several ultrasound signal waveforms received in successive measurement cycles; and

[0119] Fig. 11 shows a table with individual ratings and an overall rating for several object characteristic candidates.

[0120] In the figures, identical or functionally equivalent elements have been given the same reference symbols, unless otherwise indicated.

[0121] Fig. 1 shows a motor vehicle 1 with several ultrasonic sensors 11, 12, 13, 14 according to an exemplary embodiment. The motor vehicle 1 has several external body components 2-6, in particular a front fender 2, a front side door 3, a rear side door 4, a rear fender 5, and a side sill 6. In the present example, the several ultrasonic sensors 11-14 are mounted on the front side door 3 (ultrasonic sensors 13, 14) and on the rear side door 4 (ultrasonic sensors 11, 12). The ultrasonic sensors 11-14 can be installed openly, i.e., in respective recesses in the outer sheet metal of the 2023PF00270

[0122] 22

[0123] The ultrasonic sensors 11-14 are mounted either on the side doors 3 and 4, or concealed, meaning attached to the inside of the respective outer panels and acoustically coupled to them. Each of the ultrasonic sensors 11-14 is configured to emit an ultrasonic pulse into the lateral surroundings of the motor vehicle 1 and to receive ultrasound from the lateral surroundings of the motor vehicle 1. The motor vehicle 1 also has a control device 7, also referred to as an ECU, mounted in an engine compartment or at another suitable location in the motor vehicle 1, which is communicatively connected to each of the ultrasonic sensors 11-14 via a wired bus, a wired network, or wirelessly and can cause each of the ultrasonic sensors 11-14 to emit an ultrasonic pulse or to receive ultrasound.

[0124] In the present example, the control device 7 is also to implement a door opening assistance which, upon request from a driver or passenger, automatically opens one of the vehicle doors 3, 4, but first wants to make sure, by means of distance measurements carried out with the ultrasonic sensors 11-14, that there is no obstacle in a swing area of ​​the vehicle door 3, 4 to be opened, or where an obstacle may be, in order to determine whether and how far the respective vehicle door 3, 4 can be opened.

[0125] Fig. 2 illustrates steps S1-S5 of a method according to the exemplary embodiment, and Fig. 3 shows functional units 71-75 of the control device 7 according to the exemplary embodiment. The functional units 71-75 can be configured, for example, by loading a computer program product into the control device 7 and executing it, thereby causing the control device 7 to configure the functional units 71-75 and to carry out the method illustrated in Fig. 2 in steps S1-S5.

[0126] With reference to Figures 1 to 3, a method for measuring the area around a motor vehicle 1 is now described. 2023PF00270

[0127] 23

[0128] In step S1, the first unit 71 of the control device 7 with the multiple ultrasonic sensors 1 1 -14 performs several ultrasonic distance measurements. During each measurement, one of the ultrasonic sensors 1 1 -14 emits an ultrasonic pulse into the lateral surroundings of the vehicle 1 and is then switched to a measurement mode in which ultrasound received from the lateral surroundings of the vehicle 1 is measured.

[0129] Fig. 4 shows a plot of a raw ultrasonic signal waveform 8 received during measurement as a plot of the amplitude A of vibrations of an ultrasonic membrane of the ultrasonic sensor 11-14 and / or a signal derived therefrom against the signal propagation time t since time to of the emission of the ultrasonic pulse. Reference is made to Fig. 1-4. The ultrasonic signal waveform 8 initially shows high amplitudes during a transient phase lasting from to to ti, and then decreases. When the emitted ultrasonic pulse encounters an obstacle in the vicinity of the vehicle 1 and is reflected back to the ultrasonic sensor 11-14, the reflected signal component, also called an echo, is recorded as peaks 81, 82 in an envelope of the ultrasonic signal waveform 8.

[0130] In the present example, a first such echo 81 occurs in the period from t2 to t4, and the position of its maximum at tg is recognized as the echo position of echo 81.

[0131] A further increase in the ultrasound signal curve 8 can be observed between times ts and te; however, the amplitude does not reach the predefined amplitude threshold A. t h. Accordingly, the increase in the ultrasound signal curve 8 between t% and te is considered a disturbance signal and is not taken into account further.

[0132] In the present example, a second echo 82 also occurs in the period from t? to tg, and the position of its maximum at ts is recognized as the echo position of echo 82. 2023PF00270

[0133] 24

[0134] Based on processing the ultrasonic signal waveform 8, the first unit 71 of the control device 7, when at least one echo 81, 82 has been determined in the ultrasonic signal waveform 8, creates an echo feature based on a measurement position defined by the known installation positions of the transmitting ultrasonic sensor 11-14 and the receiving ultrasonic sensor 14 and based on the echo position ts of the first of the echoes 81, 82. This echo feature includes information that defines at least one locus of possible reflection points in the vicinity of the vehicle 1. The locus is determined here starting from the echo position ts of the first echo 81, since it is assumed that the subsequent echo 82 could be an echo of a multiple reflection, so that its echo position ts is meaningless about the distance to the first of several reflection points.

[0135] Fig. 5 shows exemplary loci 21-25 of possible reflection points in the vicinity of the motor vehicle 1 in a top view.

[0136] This means that if the transmitting and receiving ultrasonic sensors 11-14 are identical, i.e., if ultrasonic sensor 12 both transmits and receives, the corresponding locus 23 is a circle with ultrasonic sensor 12 at its center. One radius of the circle is given by half the signal path length, which can be obtained by multiplying the echo position ts, given in this example as the signal propagation time, by the ultrasonic velocity in air. In this example, the measurement position 32 is identical to the installation position of ultrasonic sensor 12. Thus, the generated echo feature can, for example, include half the signal path length and the measurement position 32; this information is sufficient to reconstruct the circular locus 23.

[0137] If, on the other hand, the receiving ultrasonic sensor 11-14 receives an echo of an ultrasonic pulse emitted by another ultrasonic sensor 11-14, i.e., if, for example, ultrasonic sensor 11 receives an ultrasonic pulse and ultrasonic sensor 12 receives a 2023PF00270

[0138] 25

[0139] When the ultrasonic signal path 8 is recorded, the corresponding locus 21 is an ellipse, at whose foci the involved ultrasonic sensors 11 and 12 are located. The ellipse 21 is defined by the signal path distance resulting from the echo position ts as described above. The sum of the distances from each point of the ellipse 21 to each of the foci at the ultrasonic sensors 11 and 12 is the aforementioned signal path distance. That is, in the present example, the measurement position 31 is defined by both installation positions, i.e., the installation position of ultrasonic sensor 11 and the installation position of ultrasonic sensor 12. For the sake of clarity, however, the measurement position 31 associated with the ellipse 21 is not shown in Fig. 5 as two sensor positions at 11 and 12, but as a single position at the center of the ellipse 21. The measuring position 31 shown in Fig. 5 can also be referred to as virtual measuring position 31.The echo feature created in this case can, for example, include the signal path and the positions of the involved ultrasonic sensors 11, 12 (collectively also referred to as measurement position 31); this information is sufficient to reconstruct the elliptical locus 21.

[0140] Reference is made to Fig. 2 and Fig. 5. In step S1, the first unit 71 of the control device 71 performs several measurements with the ultrasonic sensors 11-14 at different (real or virtual) measurement positions 31, 32, 33, 34, 35 in an analogous manner to that described above, and creates a corresponding echo feature for each measurement performed, provided an echo was determined in that measurement, which defines a corresponding locus curve 21-25.

[0141] In the example shown in Fig. 5, the elliptical locus 21 is determined by an echo feature generated during a measurement at the virtual measurement position 31 using the ultrasonic sensors 11 and 12. The locus 21 can be understood as a graphical representation of the echo feature. Accordingly, the circular locus 22 is based on a measurement at the actual measurement position 32 of the ultrasonic sensor 12, the elliptical measurement curve 23 is based on a measurement at the virtual measurement position 33 using the ultrasonic sensors 12 and 13, and the circular locus 24 is based on a measurement at 2023PF00270

[0142] 26 of the actual measurement position 34 of the ultrasonic sensor 13 back, the elliptical measurement curve

[0143] 25 refers to a measurement at the virtual measuring position 35 with the ultrasonic sensors 13 and 14.

[0144] Since each echo feature defines at least one locus 21-25, in the following, for the purpose of illustration, locus 21-25 can also be referred to when the underlying echo features and / or the information contained therein, such as the measurement positions 31-35 and echo positions ts, are meant, and vice versa.

[0145] It should be noted that the generated echo feature only needs to include information defining the respective locus curve 21-25. According to preferred embodiments, however, the generated echo feature can include additional information. Referring to Fig. 4, preferred embodiments state, for example, that information about the accuracy of the echo 81 determination can be obtained from a waveform, signal-to-noise ratio, or the like of the echo 81 and incorporated into the generated echo feature. Furthermore, according to preferred embodiments, in the situation shown in Fig. 4, information about the type of the determined echo 81 can be obtained and incorporated into the generated echo feature: In Fig. 8, a so-called double echo is present, in which another echo 82 follows the first echo 81 within a certain time interval, exhibiting a lower amplitude A than the first echo 81.Such a double echo is considered indicative of the presence of a high obstacle in the vicinity of the motor vehicle 1, where a first, direct signal path leads from the ultrasonic sensor 11-14 to the obstacle and back to the ultrasonic sensor 11-14, and a further, indirect signal path with multiple reflections leads from the ultrasonic sensor 11-14 to the obstacle, from there to a road surface, and from there back to the ultrasonic sensor. Accordingly, in the case shown in Fig. 4, information such as "double echo", "high obstacle", or the like can be included as type information in the generated echo feature. However, if no further subsequent echo 82 follows the first echo 81, information such as "single echo", "low obstacle", or the like can be included as type information. 2023PF00270.

[0146] 27. The ultrasound pulse can be incorporated into the echo feature. This can happen, for example, when the ultrasound pulse is reflected off a curb, where the direct signal path essentially coincides with the indirect signal path.

[0147] The problem to be solved by the proposed method is now described: In the further course of the proposed method, object features are to be created that at least indicate the position of obstacles in the vicinity of the motor vehicle 1.

[0148] However, if one considers Fig. 5, in which all the loci 21-25 determined in the manner described above are plotted, it becomes apparent that it is not easy to determine where an actual reflection point, and thus an actual obstacle, is located on the loci 21-25. The basic principle of lateration states that essentially point-like obstacles (negligibly small obstacles compared to the vehicle or the extent from the first ultrasonic sensor 11 to the last ultrasonic sensor 14) are to be found at the intersections of the determined loci 21-25. In contrast, a probable location for extended obstacles is tangent lines to two or more of the loci 21-25.

[0149] As shown in Fig. 6, a potential object feature 41 (hereinafter referred to as "object feature candidate") could be constructed at the intersection of the loci 21, 22, and 23. Likewise, another object feature candidate 42 could be constructed at the intersection of the loci 23, 24, and 25. If a rule is applied according to which an echo feature cannot be used to form multiple object feature candidates 4, only one of the object feature candidates 41 and 42 would be constructed. However, since, according to preferred embodiments, object feature candidates 41 and 42 are maintained over several measurement cycles, both object feature candidates 41 and 42 would nevertheless be formed after a few measurement cycles. 2023PF00270

[0150] 28

[0151] However, it is also conceivable, as shown in Fig. 7, to construct a point-shaped object feature candidate at an approximate intersection point of the locus curves 21 , 22, 24 and 25.

[0152] On the other hand, it is also conceivable, as shown in Fig. 8, to construct a linear object feature candidate 44 as a tangent line to the loci 21, 23, and 25. Since the loci 22 and 24 are disregarded here, the object feature candidate 43 shown in Fig. 7 could also be constructed in the situation shown in Fig. 87.

[0153] The proposed procedure aims to make a decision in such a situation as to which of the possible location feature candidates 41-44 (Fig. 6-8) is the object feature candidate that comes closest to reality.

[0154] With reference to Figures 1-3 and 5-8, the description of the proposed method according to the exemplary embodiment will now continue. After all echo features constructible from the multiple measurements (represented by the loci 21-25) have been generated in step S1, the second unit 72 of the control device 7 exhaustively generates all possible combinations of all the loci 21-25 formed from at least three loci 21-25 in step S2.

[0155] In the example shown in Fig. 5, this would lead to the formation of a five-locomotive combination 21, 22, 23, 24, 25; to the formation of five four-locomotive combinations 21, 22, 23, 24; 21, 22, 23, 25; 21, 23, 24, 25; 21, 22, 24, 15; and 22, 23, 24, 25; and to the formation of seven three-locomotive combinations 21, 22, 23; 21, 22, 24; 21, 22, 25; 22, 23, 24; 22, 23, 25; 22, 24, 25; 23, 24, 25.

[0156] Then, in step S3, the third unit 73 of the control device 7 creates an object feature candidate 41-44 for each of the combinations formed in step S2 by laterallying the loci 21-25 of the respective combination. At least those specified in 2023PF00270 are used.

[0157] 29

[0158] The object feature candidates 41, 42, 43, and 44 shown in Figures 6 to 8 are generated. Further object feature candidates (not shown) are also generated. It is also conceivable that for some of the combinations, no object feature candidate can be generated due to overdetermination. For other combinations, such as the combination of loci 21, 22, and 25, a point-like object feature candidate can be generated, for example, by an approximation method. This candidate lies at a position between the intersection of loci 21, 22 and the intersection of loci 21, 25 and the intersection of loci 22, 25, but would have a relatively high error rate. The object feature candidates 41–44 generated in this way for all possible combinations include, as information, at least one presumed position of an object in the vicinity of the motor vehicle 1 (i.e., the one shown in Figures 6 to 8).5-8 for drawing the object feature candidates 41-44 (position shown). Furthermore, the created object feature candidate 41-44 may include additional information in preferred further developments. In a preferred further development in which both point-like object feature candidates 41-43 and line-like object feature candidates 44 are created, the respective object feature candidate 41-44 also includes information about a type (point-like or line-like) of the object feature candidate.

[0159] In a particularly preferred embodiment, each object feature candidate 41-44 includes a precision measure of the object feature candidate. Reference is made to Fig. 9. Fig. 9 shows an enlarged section of Fig. 7. If the object feature candidate 43 is to be created based on the combination of the loci 21, 22, 24, 25, an exact intersection point of the loci 21, 22, 24, 25 cannot be determined. However, the intersection points of each pair of loci 21, 22, 24, 25 are sufficiently close to each other to justify the construction of the object feature candidate 43. In this case, the position of the object feature candidate 43 is determined such that the deviation of the respective locus curves 21, 22, 24, 25 from the position of the object feature candidate 43 (the position of a best possible approximation of the intersection point of all the locus curves 21, 22, 24, 25) is minimized. More precisely, the position of the object feature candidate 43 is determined as follows:the best possible intersection point is determined such that the sum of 2023PF00270.

[0160] 30

[0161] The squares of the distances 51, 52, 54, 55 of the respective locus curves 21, 22, 24, 25 to the best possible intersection point are minimized. To minimize the distances 51, 52, 54, 55, an iterative numerical method, such as a gradient-based method, or an analytical, computational method can be used. Particularly preferably, the object feature candidate 41 thus created includes the deviation determined in this way (sum of the squares of the distances 51, 52, 54, 55) as the precision measure of the object feature candidate 43.

[0162] Referring to Fig. 8, in a further preferred embodiment, not only is a best possible intersection point determined approximately for each of the combinations in the manner described above, but a best possible tangent line is also approximately drawn to the loci of the combination, for example, to loci 21, 23, 25. While in Fig. 8 an exact tangent line can be drawn to the three loci 21, 23, 25, this is not possible in the general case; rather, the tangent line whose deviation (sum of the squared distances) from loci 21, 23, 25 of the respective combination is minimal is determined again by means of an analytical or numerical method.

[0163] According to a preferred further development that combines the aforementioned preferred further developments, the deviation of the best possible intersection point can then be compared with the deviation of the best possible tangent line, and depending on which of the deviations is smaller, either a point-like object feature candidate 43 or a line-like object feature candidate 44 is created. The corresponding, i.e., the smaller of the two deviations, is then used as the precision measure of the created point-like or line-like object feature candidates 43, 44.

[0164] The description of a basic embodiment is now continued with reference to Figures 1-3 and Figures 5 to 8. In step S4, the fourth unit 74 of the control device 7 determines, for each of the object feature candidates 41-44 determined in step S3, based on the information in 2023PF00270

[0165] 31 the object feature candidate 41-44 and / or in the echo features used to create it, information included an evaluation.

[0166] Any suitable criterion can be used for evaluation. Preferably, the precision measure mentioned above is used for evaluation; for example, the precision measure or a value derived from it can be adopted as the evaluation criterion. Other criteria for evaluation include, for example, the qualities of the echo features used to create the object feature candidate 41-44, as described above, and / or a match or mismatch of the types of these echo features. However, these are only a few of many possible examples for carrying out the evaluation. Further examples are discussed below with reference to other preferred developments. All these criteria can be combined with one another in any suitable manner.

[0167] Finally, in step S5, the fifth unit 75 of the control device 7 determines whether the evaluation of the highest-rated object feature candidate 41-44 meets a predefined minimum criterion (for example, whether a predefined minimum precision is achieved). If so, the highest-rated object feature candidate 41-44 is adopted as a newly created object feature.

[0168] Accordingly, the exemplary embodiment describes an advantageous method in which, using an exhaustive brute-force approach, all combinations of three or more echo features (loci 21-25) are generated from all determined echo features; an object feature candidate 41-44 is generated for each combination; the object feature candidates 41-44 are evaluated; and only the highest-evaluated object feature candidate 41-44 is adopted as the newly created object feature candidate. This advantageously results in a better adaptation of the generated object feature to physical reality. 2023PF00270

[0169] 32

[0170] With reference to Fig. 1 and Figs. 5-8, the control device 7 can thus reliably determine, based on the generated object features 41-44, whether and to what extent each of the side doors 3, 4 can be opened. The risk of a collision between the side doors 3, 4 and an incorrectly detected obstacle is advantageously reduced. Likewise, the risk of the side doors 3, 4 being blocked (not opened) due to a falsely detected phantom obstacle is advantageously reduced, even though at least one of the side doors 3, 4 could have been opened. Such a situation could conventionally occur, in particular, if, instead of a point-like object feature 41-43, which might only block one of the two side doors 3, 4, an extended linear object feature 44 was incorrectly determined, seemingly blocking both of the side doors 3, 4.

[0171] Further preferred embodiments of the exemplary embodiment will now be described, which can be combined arbitrarily with the basic form of the exemplary embodiment or with the preferred embodiments already described.

[0172] According to a preferred further development process, after the highest-rated object feature candidate 41-44 has been adopted as an object feature in step S5, the echo features used to create it are discarded. Discarding can include deleting the echo features. However, if the echo features are retained for further measurement cycles, as described below, discarding can also include marking the echo features to be discarded as no longer to be used to create object feature candidates 41-44 in the current measurement cycle. Steps S2 to S5 are then repeated. If the highest-rated object feature candidate 41-44 resulting in step S5 again exceeds the specified minimum criterion, this object feature candidate 41-44 is also adopted as a newly created object feature. This loop continues until no further newly created object features are adopted in step S5.Accordingly, all sufficiently well-rated object characteristics can advantageously be determined in an iterative process. 2023PF00270.

[0173] 33

[0174] According to a particularly preferred further development, step S1 of determining echo features is repeated in several successive measurement cycles. The echo features generated in each cycle are retained across multiple measurement cycles. An echo feature generated in the first measurement cycle is updated in a subsequent second measurement cycle if another echo 81' (Fig. 10) corresponding to the echo feature generated in the first measurement cycle is determined.

[0175] For explanation, reference is made to Fig. 10. Fig. 10 shows envelopes of several echo signal waveforms 8', 8', 8" from different successive measurement cycles, which were recorded at the same real or virtual measurement positions 31-34 (Fig. 5).

[0176] The ultrasound signal waveform 8 recorded in the first of the measurement cycles shows a single echo 81 at an echo position (signal transit time or signal path distance) ts. The ultrasound signal waveform 8' received in a subsequent second measurement cycle shows a single echo 81' at a slightly shifted echo position ts'. Due to the similarity of the signal shapes and echo positions ts and ts', the echo 81' can be recognized as corresponding to the echo feature created in the first measurement cycle, and the echo feature (the associated locus 21-25, Fig. 5) is updated according to the updated echo position ts'.

[0177] It should be noted that the successive measurement cycles, such as the first measurement cycle and the second measurement cycle, do not need to follow each other immediately, as the echo characteristics are maintained over several measurement cycles.

[0178] When updating an echo characteristic, various additional pieces of information can be stored within it. For example, the created or updated echo characteristic can contain information about the measurement cycle in which it was created or last updated. This information can include a time or, for example, a sequential number. Furthermore, the created or updated echo characteristic can include a counter indicating how many measurement cycles have already been used for the echo characteristic 2023PF00270.

[0179] 34 echoes 81, 81' corresponding to this feature (associable with it) have been detected. Furthermore, in the second measurement cycle, a speed can be determined at which the echo 81, 81' moved over the two measurement cycles (shown in Fig. 10 as an arrow from 81 to 81') based on the known echo positions ts and ts' and the known time interval between the two measurement cycles. If an echo (not shown) is then detected again at the measurement position in question in a third measurement cycle, this newly detected echo, or rather its echo position, can be compared with a predicted echo 83, or rather its predicted echo position ts". The predicted echo 83 can be predicted based on the echo position 81' and the speed determined in the second measurement cycle.A degree of agreement between the echo position of the echo from the third measurement cycle and the echo position ts" of the predicted echo 83 can be stored as a prediction accuracy value in the echo feature. The determination of velocities and the predictions can be further improved if the velocity determination and the predictions are based on multiple echo positions ts determined in several previous measurement cycles.

[0180] It should be noted that updating, comparing, determining a velocity and predicting echo positions ts, ts', ts" is equivalent to updating, comparing and predicting the locus curves 21-25 defined by the echo features (Fig. 5).

[0181] Reference is made to Figures 5 to 10. Advantageously, any of the information just discussed can be used in the evaluation of the object feature candidates 41-44 created from the corresponding echo features by lateration.

[0182] For example, one criterion for evaluation could be the recency of the object feature candidate 41-44. The recency can be determined based on information about the respective measurement cycles in which the echo features used to create the object feature candidate 41-44 were created or last updated. Object feature candidates 41-44 created using current echo features can be evaluated more favorably than object feature candidates 41-44 created using older echo features. 2023PF00270

[0183] 35 echo features have been created that have not yet been deleted, but are also no longer entirely up-to-date.

[0184] For example, a criterion for evaluating a candidate object feature 41-44 can be a degree of reliability. The degree of reliability can be determined based on the number of measurement cycles in which an echo 81, 81' corresponding to the echo feature was determined. An echo feature for which an echo 81 was determined only once can be considered unreliable, while an echo feature for which a corresponding echo 81, 81' has been determined 5 times or more can be considered absolutely reliable and achieve the best possible rating.

[0185] Another criterion could be, for example, the degree of agreement between the last determined velocities of the echo features used to create the object feature candidate 41-44. For instance, if a locus 21-25 resulting from an echo feature whose corresponding echoes 81, 81' (Fig. 10) have moved away from the associated measurement position 31-35 is combined with a locus 21-25 resulting from an echo feature whose corresponding echoes (not shown) have moved towards the associated measurement position 31-35, the associated object feature candidate 41-44 receives a lower rating than an object feature candidate 41-44 whose corresponding echoes 81, 81' have moved in the same direction. The evaluation can also be better the less the velocities stored in the respective echo features used to determine the object feature candidate 41-44 differ from each other.

[0186] Finally, the predictive accuracy, which is stored in the echo feature when associating a newly determined echo with an existing echo feature and indicates a deviation of the newly determined echo from the echo 83 predicted based on the stored velocity, can also be a criterion in the evaluation. 2023PF00270

[0187] 36

[0188] According to a particularly preferred further development, the evaluation of the object characteristic candidates 41-44 (Figs. 6-8) in step S4 (Fig. 2) is calculated as an overall evaluation 1 10 (Fig. 1 1 ) from a plurality of individual evaluations 1 11 , 1 12 (Fig. 11 ) of each of several criteria. Most preferably, this determination is optionally carried out by normalizing as well as by weighting and summing the several individual evaluations 11 1 , 1 12 (Fig. 11 ).

[0189] Reference is made to the table in Fig. 11. The table shows in the column "ID" a consecutive number of a combination of echo features or the associated object feature candidate 41-44 (Fig. 6-8) generated and evaluated within the framework of the exhaustive brute-force search. In the column "EB1", the table shows a first individual evaluation 11 1 of the respective object feature candidate 41-44 (Fig. 6-8) according to a first criterion, and in the column "EB2", the table shows a second individual evaluation 112 of the respective object feature candidate 41-44 (Fig. 6-8) according to a second criterion.

[0190] The individual ratings 11 1 , 112, ... (more than two individual ratings per combination are possible) are numerical ratings 11 1 , 112 on a predefined scale (minimum and maximum values) and according to a predefined convention that indicates whether increasing numerical values ​​represent better or worse ratings. In this example, the scale is a floating-point scale from 0 to 1, and higher values ​​signify better ratings. If different scales or conventions are used for different individual ratings, normalization is first performed so that all normalized individual ratings 11 1 , 112 are on the same scale and according to the same convention. The normalized individual ratings thus obtained are then summed to obtain a total rating 110 for each of the combinations.

[0191] The values ​​from Fig. 1 1 were obtained during experiments by the inventors with echo characteristics that correspond to those in Fig.

[0192] The 5 shown loci 21-25 define the result. According to these values, the highest overall score of 1.10 out of 0.985 is obtained for the combination with ID 6. This combination- 2023PF00270

[0193] 37 tion corresponded to the lateration of a point-like object feature candidate 42 (Fig. 6) based on the loci 23, 24 and 25 (Fig. 6). Accordingly, the object feature candidate 42 (Fig. 6) was adopted as the object feature and not, for example, the point-like object feature candidate 43 (Fig. 7) or the line-like object feature candidate 44 (Fig. 8).

[0194] This corresponded to the actual measurement situation, in which two rods were arranged at the positions labeled 41 and 42 in Fig. 6. Thus, thanks to the proposed exhaustive brute-force approach, the incorrect determination of a linear object feature 44 (Fig. 8) could be prevented, and the incorrect determination of a point-like object feature 43, which is located closer to the motor vehicle 1 and could have prevented the opening of the side doors 3, 4 (Fig. 1) of the motor vehicle 1, could be avoided.

[0195] It should be noted that in Table 1, the individual ratings 11, 112 are simply summed to obtain the overall rating 110. In other words, the weighting factor for each of the individual ratings 11, 112 is exactly 0.5, or put another way, both individual ratings 11, 112 of both criteria have been weighted equally. However, particularly when more than two criteria are used, it is advisable to deviate from this equal weighting. The weighting factors can be determined by a person skilled in the art according to their usual knowledge and the resulting expectations, and / or those weighting factors that yield the best results in relevant scenarios can be determined through trials.

[0196] The intention is also to determine the weighting factors of the individual ratings 1 11 , 1 12 in an automated process based on numerical optimization, machine learning, or a neural network. The determination of the weights takes place under controlled conditions in which several typical obstacles are successively positioned next to the vehicle 1 in typical positions. For each obstacle position, weights are determined that result in object characteristics best adapted to reality. A deviation of 2023PF00270 is possible.

[0197] 38

[0198] Object characteristics, obtained with respective weights chosen within the iterative process, are derived from known real-world objects in the vicinity of vehicle 1 and used as feedback for the automated method employed to determine the weights. This is a multidimensional optimization problem, as the weights must be optimized not just for one, but for a multitude of different obstacle arrangements, such that they yield the best possible results for all of them.

[0199] Although the present invention has been described using exemplary embodiments, it can be modified in many ways.

[0200] The invention has been described by way of example in the context of a door opening assistance function using ultrasonic sensors 1 1-14 installed on the front and rear side doors 3, 4. However, the invention can also be implemented with ultrasonic sensors installed on other body components, such as the front fender 2, the rear fender 5, the side sill 6, a front bumper or a rear bumper, and / or in connection with other assistance functionalities, such as a parking assistance or a driving assistance system.

[0201] For the basic embodiment, it has been described that the evaluation is an evaluation of a precision measure that is included in the information from which the respective object feature candidate 41-44 is formed. While the use of the precision measure for the evaluation is preferred, it is not a necessary feature. In other variants, only a minimum value can be specified for the precision measure, above which an object feature candidate 41-44 is created in step S3, and the evaluation can be based on other criteria described above as preferred embodiments.

[0202] As part of the exhaustive combination according to the brute-force approach of step

[0203] For certain echo characteristics, it is not necessary to consider all possible combinations of all 2023PF00270.

[0204] 39 specific echo features can be combined, provided that exhaustive combinations are performed using a brute-force approach. For example, only combinations of three or more echo features can be combined, and the corresponding loci can be laterated. The formation of pairs can be omitted, as these generally exhibit low reliability. It is also conceivable to set a maximum number of echo features to be combined, e.g., a maximum of 10, preferably a maximum of 8, and particularly preferably a maximum of 5, to avoid the fruitless trial and error of over-determined combinations. Furthermore, additional criteria can be applied to reduce the subset of echo features tried and thus save computation time.In particular, it is planned to form different subsets of the total set of determined echo features, whereby each subset must include at least one echo feature that was created or updated in the current measurement cycle, i.e., for which a corresponding echo 81, 81' (Fig. 10) was determined in the current measurement cycle. This prevents combinations from being tried that do not include a current echo feature and therefore would inevitably receive a low score.

[0205] Figure 4 shows a raw ultrasound signal waveform 8. This is for illustrative purposes only. The described signal processing can be performed on the raw ultrasound signal waveform 8, or on an envelope of the raw ultrasound signal waveform 8, as shown in Figure 10. It is also conceivable to use a downsampled or IQ-converted ultrasound signal waveform. Advantageously, the generation of the envelope, the downsampling, and / or IQ conversion can be performed directly by the respective ultrasound sensor 11-14 (Figure 1), thus reducing the amount of data that needs to be communicated to the control device 7 (Figure 1).

[0206] It was discussed that echo features are maintained over several measurement cycles and that new measurements of associated echoes are used to update the echo features. This provided the possibility of determining the speed at which the echo position ts moves over several measurement cycles, based on the 2023PF00270

[0207] 40

[0208] Speeds are used to predict future echo positions ts. Based on the speeds and / or predictions, criteria are developed for evaluating the object feature candidates 41-44 determined based on such updated echo features. In the same way, certain object features can also be maintained over several measurement cycles. If a new object feature candidate 41-44 is to be adopted that can be associated with an object feature from a past measurement cycle, a new object feature cannot be created; instead, the existing object feature can be updated. In this case as well, a speed can be determined at which the object has moved relative to the vehicle 1 in its vicinity over past measurement cycles. Based on the speed, a prediction of future positions of the object feature can be made.In this context, a deviation of a last determined position of an object feature candidate 41-44 from a predicted position, which was predicted based on the velocity of the object feature with which it is to be associated, can also represent a criterion for evaluating the object feature candidate 41-44.

[0209] 2023PF00270

[0210] 41

[0211] REFERENCE MARK LIST

[0212] 1 motor vehicle

[0213] 2 front fenders

[0214] 3 front vehicle door

[0215] 4 rear vehicle door

[0216] 5 Rear fenders

[0217] 6 side skirts

[0218] 7 Control device

[0219] 8, 8', 8" Ultrasound signal waveform

[0220] 11-14 ultrasonic sensors

[0221] 21-25 loci (circles, ellipses)

[0222] 31-35 measuring positions

[0223] 41-43 point-like object feature candidates

[0224] 44 linear object feature candidate

[0225] 51, 52, 54, 55 Distances (deviation)

[0226] 71-75 first to fifth unit

[0227] 81, 81' Echo

[0228] 82 Follow-up Echo

[0229] 83 predicted echo

[0230] 111, 112 individual ratings

[0231] 110 ratings, overall rating

[0232] A Amplitude

[0233] Ath amplitude threshold t signal propagation time ts echo position

[0234] S1-S5 process steps

Claims

2023PF00270 42 PATENTANSPRÜCHE 1. Method for measuring the environment of a motor vehicle (1) equipped with a number of ultrasonic sensors (1 1 - 14), comprising: a) performing (S1 ) several measurements with the number of ultrasonic sensors (1 1 - 14) at different measurement positions (31 - 35), each of the measurements comprising: emitting an ultrasonic pulse with one of the number of ultrasonic sensors (1 1 - 14); determining an echo (81 ) in an ultrasonic signal waveform (8) received by one of the ultrasonic sensors (1 1 - 14) in response to the emission of the ultrasonic pulse; and creating an echo feature formed from information defining at least one locus (21 - 25) of possible reflection points in the environment of the motor vehicle (1 ), based on the measurement position (31 - 35) and an echo position (ts) of the echo (81 ) in the ultrasonic signal waveform (8);b) exhaustively forming (S2) combinations of several echo features from at least a subset of the echo features created in step a); c) creating (S3) an object feature candidate (41-44) for the respective combination by latering the locus curves (21-25) defined by the echo features, wherein the respective object feature candidate (41-44) is formed from information that includes at least one presumptive position of an object in the vicinity of the motor vehicle (1) obtained as a result of the latering; d) determining (S4) an evaluation (110) for each of the object feature candidates (41-44) determined in step c) based on the information contained in the object feature candidate (41-44) and / or in the echo features used to create the object feature candidate (41-44);and e) Adopt (S5) the object characteristic candidate (41-44) determined in step d) that has the best rating (1-10), provided that the best rating (1-10) meets a predefined minimum criterion, as a newly created object characteristic.; 2. Method according to claim 1, characterized by: 2023PF00270 43 f) Discard the echo features used in step e) to form the object feature candidate (41 -44) adopted as the newly created object feature and repeat steps b) to f) until no further object feature candidate (41 - 44) is adopted as the newly created object feature in step e).

3. Method according to one of the preceding claims, wherein the evaluation is carried out on the basis of at least one precision measure of the object feature candidate (41-44) as a criterion.

4. Method according to claim 3, characterized in that in step c) for the respective combination: depending on at least one deviation of the loci (21-25) defined by the echo features from a best possible intersection point of the loci (21-25) and depending on one deviation of the loci (21-25) from a tangent line best adapted to the loci (21-25), either a point-like object feature candidate (41-43) at the position of the best possible intersection point or a line-like object feature candidate (44) at the position of the best possible tangent line is determined, and the created object feature candidate (41-44) further comprises as information a specification of the type of object feature candidate as point-like or line-like and the precision measure, wherein the precision measure specifies the deviation of the loci (21-25) from the best possible intersection point or from the best possible tangent line.

5. Method according to one of the preceding claims, characterized in that the echo feature created in step a) further comprises as information: a quality of the determination of the echo (81); and / or a type of the determined echo (81).

6. Method according to one of the preceding claims, characterized in that 2023PF00270 44 Step a) is repeated in several successive measurement cycles, the echo features created in each step a) are maintained over several measurement cycles, and one of the echo features created in a first of the measurement cycles is updated in a subsequent second measurement cycle when an echo (81 ') corresponding to the one created in the first measurement cycle is determined again.

7. Method according to claim 6, characterized in that in step b) the respective subset comprises at least one echo feature for which a corresponding echo has been determined in a current measurement cycle.

8. Method according to claim 6 or 7, characterized in that the created or updated echo feature further comprises as information: an indication of a measurement cycle in which the echo feature was created or last updated; and / or an indication of a number of measurement cycles in which an echo (81) corresponding to the echo feature was determined; and / or a speed at which the respective echo (81, 81') corresponding to the echo feature has moved over several measurement cycles; and / or a prediction accuracy that indicates a degree of agreement between a locus (21-25) defined by the updated echo feature and a predicted locus predicted based on the speed at the last determination of the echo (81, 81') corresponding to the echo feature.

9. The method according to claim 8, characterized in that in step d) the evaluation is further carried out on the basis of one or more of the following criteria: the degree of currency of the object feature candidate (41-44) based on the information of the measurement cycles in which the respective echo feature used to create the object feature candidate (41-44) was last updated; and / or 2023PF00270 45 of a reliability level of the object feature candidate (41-44) based on the Information on the number of measurement cycles in which an echo feature corresponding to the respective object feature candidate (41-44) used to create the object feature candidate (41-44) was obtained. Echo (81 , 81 ') was determined; and / or a degree of agreement of the velocities of the echo features used to create the object feature candidate (41-44); and / or the predictive accuracy of the echo features used to create the object feature candidate (41-44).

10. Method according to one of the preceding claims, characterized in that in step d) the evaluation (1 10) is calculated from a plurality of individual evaluations (1 11 , 1 12) of a respective criterion.

11. Method according to claim 10, characterized in that the overall rating (110) is formed by weighting and summing the number of individual ratings (111, 112).

12. Method according to claim 1 1 , further characterized by determining weights of the individual ratings (11 1 , 112) using a method based on numerical optimization or machine learning or a neural network, wherein the determination of the weights is carried out under controlled conditions, wherein a deviation of object features obtained with respective weights chosen within the iterative method from known actual objects in the environment of the motor vehicle (1 ) is used as feedback for the method used to determine the weights.

13. Computer program product comprising instructions which, when the program is executed by a control device (7) of a motor vehicle (1 ) with a number of ultrasonic sensors (1 1 -14), cause the control device (7) to execute the method according to one of claims 1 to 12. 2023PF00270 46 14. Control device (7) for a motor vehicle (1) equipped with a number of ultrasonic sensors (11-14) which is configured to perform the method according to one of claims 1 to 13.

15. Motor vehicle (1) comprising a number of ultrasonic sensors (11-14) and the control device (7) according to claim 14.

Images