DE102019128023B4 - Method for classifying the height of an object by a driving assistance system - Google Patents

Abstract

A method for classifying the height of an object (24) using an ultrasonic sensor (16), the classification comprising the height of the object (24) relative to the installation height of the ultrasonic sensor (16), comprising the method steps: a) emitting ultrasonic waves; b ) Detecting a reflection echo of ultrasonic waves reflected on the object (24); c) detecting the time interval between two peaks (26, 28) of a double echo from one another of the reflection echo; d) modifying the double echo by adjusting the time interval between the peaks (26, 28) the double echo to each other to a predetermined value; and e) classifying the height of the object (24) based on the modified double echo.

Description

The present invention relates to a method for classifying the height of a detected object by a driving support system using an ultrasonic sensor. The present invention further relates to a driving assistance system which is designed to carry out such a method at least in part.

Driving support systems, for example for motor vehicles, are widely known. Environment detection sensors are used in driving assistance systems in order to detect the surroundings of a vehicle. This is a prerequisite, for example, in order to identify routes and to recognize obstacles in the area of the routes. For example, applications for driving support systems based on ultrasonic sensors are known.

In today's ultrasonic systems, the target is detected by finding peaks in the response signal of an ultrasonic sensor. Typically, an ultrasonic sensor sends out a specific signal for this purpose. Depending on the nature of the subsurface, a certain proportion of the ultrasonic waves is reflected and can be received by the sensor again. Today's systems use this to determine objects and their distance. However, it is often desirable not only to detect the position of the object but also to estimate whether the object has a height that is greater or less than the installation height of the ultrasonic sensor.

EP 2 073 038 B1 describes a method comprising the transmission of measurement signals and the reception of measurement signals from a sensor, which are reflected on a distant object. The time between sending and receiving the measurement signals is determined and the time is correlated with a distance. The statistical distribution of the distance data is correlated with the height of the reflective object by means of standard deviation and / or variation. Ultrasonic or microwaves are used as measurement signals, and the time between two measurements is varied. A frequency of the transmitted measurement signals is varied.

DE 10 2014 114 999 A1 describes a method for detecting at least one object in a surrounding area of a motor vehicle by means of a driver assistance system, in which a transmission signal is transmitted in successive measurement cycles with a distance sensor and a first and a second echo of the transmission signal reflected by the at least one object is received and a first distance value is determined by means of a control device on the basis of the first echo, a second distance value is determined on the basis of the second echo and a height of the at least one object is determined on the basis of the first and the second distance value.

DE 10 2010 025 552 A1 describes a method for detecting the surroundings of a motor vehicle to determine possible parking spaces by means of an ultrasound-based environment sensor system, in which the detection of the environment of the motor vehicle is based on the detection of first and second echoes of the ultrasonic waves emitted by the environment sensor system and reflected on objects. By means of the simple detection of first and second echoes, a precise analysis of the motor vehicle environment is possible without the need for complicated investigations of, for example, the temporal course of the backscattered amplitude of the sound wave, since it has been found that objects that only generate first echoes with respect to a measurement point , have a different height than objects that generate first and second echoes with respect to a measuring point. This is used in a method for assisted parking and a corresponding parking assistant.

DE 10 2012 200 308 A1 relates to a method for detecting the surroundings, in particular for detecting the surroundings of a motor vehicle, comprising at least one sensor with at least one transmitter and at least one receiver, or a combined transmitter and receiver for detecting the distance to an object using the pulse-echo method, with a signal from the sensor is sent and an echo of the signal reflected by an object is received and the distance to the object is determined from the transit time between sending the signal and receiving the echo. After receiving an echo, the recording of measured values is stopped and a new signal is sent.

EP 1 764 630 B1 describes a parking space regulation method including an evaluation of the height of an object on the basis of the detection of two echo signals as a double echo, the time interval between which is smaller than a predetermined maximum distance.

DE 10 2011 088 401 A1 describes a method for operating an assistance system of a vehicle with at least one ultrasonic sensor, the ultrasonic sensor sending and receiving ultrasonic signals as a function of at least one operating condition of the vehicle, wherein a control device calculates distances to objects in the vicinity of the vehicle from the signal transit times of reflected ultrasonic signals carries out a classification of the objects on the basis of the reflected ultrasound signals, the control device performing the classification on the basis of an assessment of a ratio of the detected reflected signals when the vehicle is driving slowly and / or at a standstill Performs ultrasonic signals with multiple echoes to the total number of detected reflected ultrasonic signals of the ultrasonic sensor within an evaluation period.

DE 10 2009 047 012 A1 relates to a method for detecting objects, wherein at least one sensor emits a transmission pulse as a wave, in particular as an acoustic or electromagnetic wave, which is at least partially reflected by objects in the propagation space, the reflected wave being detected as a received signal by at least one receiver. It is provided that the received signal of the reflected wave is divided into segments, information being obtained from the individual segments that is used to determine an object hypothesis.

From the DE 10 2017 128 983 A1 is a method for estimating a height of an object in a surrounding area of a motor vehicle by means of an ultrasonic sensor with statistical evaluation of a received signal.

Such solutions known from the prior art can, however, offer further potential for improvement, in particular with regard to the classification of the height of an object detected with an ultrasonic sensor.

It is the object of the present invention to at least partially overcome the disadvantages known from the prior art. In particular, the object of the present invention is to provide a solution by means of which it is possible to classify the height of an object detected with an ultrasonic sensor in an effective and reliable manner.

The object is achieved according to the invention by a method with the features of claim 1. The object is also achieved according to the invention by a driving support system with the features of claim 10. Preferred embodiments of the invention are described in the dependent claims, in the description or the figures, further features described or shown in the dependent claims or in the description or the figures, individually or in any combination, may constitute an object of the invention if the context does not clearly indicate the contrary.

1. a) Aussenden von Ultraschallwellen;
2. b) Erfassen eines Reflexionsechos von an dem Objekt reflektierten Ultraschallwellen;
3. c) Detektieren des zeitlichen Abstands zweier Peaks eines Doppelechos zueinander des Reflexionsechos;
4. d) Modifizieren des Doppelechos durch Einstellen des zeitlichen Abstands der Peaks des Doppelechos zueinander auf einen vorbestimmten Wert; und
5. e) Klassifizieren der Höhe des Objekts basierend auf dem modifizierten Doppelecho.

A method is proposed for classifying the height of an object using an ultrasonic sensor, the classification comprising the height of the object relative to the installation height of the sensor, having the method steps:

1. a) emitting ultrasonic waves;
2. b) detecting a reflection echo of ultrasonic waves reflected on the object;
3. c) detecting the time interval between two peaks of a double echo from one another of the reflection echo;
4. d) modifying the double echo by setting the time interval between the peaks of the double echo to one another to a predetermined value; and
5. e) Classifying the height of the object based on the modified double echo.

Such a method allows the height of a detected object to be classified in an effective and reliable manner using data from an ultrasonic sensor. With the technology described here, additional added value can be gained from the data from the ultrasonic sensor.

The method described here thus serves to classify the height of a detected object, such a height classification being based in particular on whether the object has a low height or the same height or a greater height compared to the installation height of the ultrasonic sensor. It can be assumed, for example, that the installation angle of the ultrasonic sensor is exactly horizontal, or the installation angle of the ultrasonic sensor can typically be known from the calibration or application of the sensor to the vehicle. Any deviations in the installation angle can, if necessary, be taken into account in the evaluation.

Such a method can be carried out at least in part, for example completely, by a driving support system. The method described here can be used for completely autonomous driving, that is to say the method can be designed for driving the vehicle without a driving intervention by a driver. Alternatively, it is possible that the method is only used to support the driver while driving, that is to say for example to output driving instructions, with the driver having to carry out certain driving interventions himself. The method can thus be suitable for partially autonomous driving or a supported method. In principle, the method can be started automatically or carried out or triggered manually.

In addition, the method can be carried out using only one or a plurality of ultrasonic sensors.

In order to classify the height of a detected object according to the method described here, the method comprises the following method steps.

According to method step a), the method comprises the emission of ultrasonic waves. This method step can take place in a manner known per se using an ultrasonic sensor or a plurality of ultrasonic sensors, in that ultrasonic waves are emitted by a transmitter or a plurality of transmitters of an ultrasonic sensor or a plurality of ultrasonic sensors.

The same applies in principle to method step b), according to which a reflection echo of the emitted ultrasonic waves, that is to say in other words the ultrasonic radiation reflected by an object located in front of the ultrasonic sensor, is detected or recorded.

The above-described steps can thus in particular be carried out by an ultrasonic sensor or a plurality of ultrasonic sensors, which is or is provided in a plurality of vehicles anyway.

It is therefore an advantage of the method described here that such an ultrasonic sensor can be used for the classification of the height of the object, which ultrasonic sensor is already present in the motor vehicle, in particular is provided for another purpose in the motor vehicle. Specifically, the invention preferably makes use of such an ultrasonic sensor which is part of a sensor arrangement which is set up to carry out distance measurements, in particular for maneuvers into and out of parking spaces. The method is therefore particularly advantageous in that it does not require any hardware additions and only precautions have to be taken on the software side in order to carry out the method according to the invention.

The ultrasonic sensor can thus take on two functions, on the one hand, namely a distance measurement, which is preferably carried out at low speeds of the motor vehicle, e.g. when the motor vehicle is parked in or out, and on the other hand, the classification of the height of the detected object.

In order to make this possible based on the recorded ultrasonic signals, the method further comprises, according to method step c), the detection of the time interval between two peaks of a double echo from one another of the reflection echo. This step can be carried out in a simple manner by creating a time-resolved signal of the detected echo and thereby determining the time interval between the peaks of the double echo.

This step is based on the fact that ultrasonic sensors send out a signal and measure the waves reflected by objects as a function of time. The distance between the sensor and the object can be determined via the transit time. In the case of objects that are higher than the sensor, two reflected waves appear, which are represented as a double echo or an arrangement of two peaks. A wave with a shorter transit time that is reflected back directly from the surface normal of the object and occurs as the first peak, resolved in time, and one that is reflected on the ground at the point of contact of the object and occurs as the second peak, resolved in time. Whether the detected peaks are a double echo, i.e. the peaks belong to the same object, can be determined by determining the distance to the object from the position of the first peak and then estimating where and which one is based on this Runtime should be the second peak. A run-time-based verification of a double echo can thus take place.

It can thus be estimated whether an object is higher or lower than the sensor height or the installation height of the ultrasonic sensor. For this it is necessary to detect and classify the two peaks. Detecting peaks in the time-resolved signal is usually carried out as standard.

In the prior art, the altitude is classified on the basis of whether or not two echoes occurred at a defined distance from one another. However, the distance between the two echoes changes depending on the object distance. For this reason, a distance-dependent height estimate results. This distance dependency can lead to significant detection errors for the detection of two echoes, in particular at distances greater than 2 m. For other approaches based on machine learning, for example, differently trained classifiers or networks are currently used for different distance ranges. Although this measure can prevent the loss of accuracy in certain distance ranges, it requires a significantly higher training effort and the storage of many different classifiers.

In order to circumvent this disadvantage, according to method step d), the modification of the double echo is provided by setting the time interval between the peaks of the double echo to one another to a predetermined value. According to this step, the time interval between the peaks of the double echo is not in accordance with the retained the detected measurement signal, but rather the spacing of the peaks from one another is adjusted to a value that would result from a predetermined spacing of the detected object. In other words, the peaks, in particular the second, that is to say later occurring, peak of the double echo are shifted along the time axis. The predetermined distance can preferably be in a range of less than or equal to 2 m, for example 1 m, since, as indicated above, susceptibility to errors occurs in particular when the distance between the detected object and the ultrasonic sensor is more than 2 m.

Then, according to step e), the height of the object is classified based on the modified double echo. In this step, the second peak of the double echo can thus be compared with the first peak or normalized to the first peak. If defined features of the peaks, for example the areas under the peaks, are then compared with one another with reference to the now defined distance, it can be reliably and securely classified whether the detected object has a height that is lower than the installation height of the ultrasonic sensor or whether the height of the detected object is equal to or greater than the installation height of the ultrasonic sensor.

In principle, this evaluation can be carried out with a suitable classifier.

The proposed solution therefore consists of a type of mapping to a fixed, predetermined distance. For this purpose, the area in which the second peak should be located based on geometric calculations is shifted for each detected peak, so that the distance between them corresponds to the theoretically expected distance at the fixed distance. The empty area between the peaks can be filled with gamma-distributed noise, as one would expect from the ground at the predefined distance. This can further improve the classification with regard to computing power. Alternatively, the classifier or neural network can also only be given the two areas without the noise inserted in between. Therefore, you can only use one network or one classifier, regardless of the distance, saving you time and money.

In other words, in the method described here, as a departure from the solutions from the prior art, provision is made for the measurement signal or the envelope curve to be modified before it is fed to the classifier. The transit time distance between the two peaks of the double echo is adjusted in such a way that the transit time distance between the peaks corresponds to a transit time distance at a predetermined object distance.

A major advantage of the method described here can be seen in the fact that the classifier only needs to be trained on this predetermined object distance. Extensive training attempts with different distances from detected objects are not necessary. This can significantly reduce the amount of data and, at the same time, the computing power.

In addition, this enables a reliable and effective classification of the height of the detected objects despite the reduced amount of data, so that in other words, despite the reduced amount of data, there is no loss of reliability of the classification and thus the safety of subsequent driving sequences based on the classification is to be expected.

Following the above, it is therefore expedient for the method to be carried out using previously generated training data. In other words, it can be provided that the classification of the height of the detected object based on the determined modified peaks takes place by means of a comparison with previously generated training data by means of the classifier.

• a_T) Aussenden von Ultraschallwellen;
• b_T) Erfassen eines Reflexionsechos von an dem Objekt reflektierten Ultraschallwellen bei vorgegebener Distanz des Ultraschallsensors zu dem Objekt; und
• f_T) Speichern von Daten hinsichtlich vordefinierter Merkmale der Peaks des Doppelechos mit der vorgegebenen Distanz des Ultraschallsensors zu dem Objekt, beispielsweise in einem Speicher des Fahrunterstützungssystems.

In other words, it can be provided that the training has the following procedural steps, the designations with index T denoting training steps:

• a _T ) emitting ultrasonic waves;
• b _T ) detecting a reflection echo of ultrasonic waves reflected on the object at a predetermined distance of the ultrasonic sensor from the object; and
• f _T ) storage of data with regard to predefined features of the peaks of the double echo with the predetermined distance of the ultrasonic sensor to the object, for example in a memory of the driving assistance system.

It is preferably provided during training that the training data are generated only at the predetermined value of the time interval between the peaks of the double echo, which corresponds to a predetermined distance between the ultrasonic sensor and the object, in order to reduce the computational effort during training. This is possible, as described above, since the evaluation of the measurement is based on a predetermined distance between the ultrasonic sensor and the detected object. Further training data is therefore not necessary.

It can furthermore be preferred that the accuracy of the method and thus in particular the training data is checked during training. This can be implemented, for example, in that, during a training process, individual measurements are not carried out as training using the steps, linking the results with a predetermined distance, but rather by classifying the height of the object during the training according to steps a) to e) . On the other hand, since the distance between the ultrasonic sensor and the object is known and thus the result of the classification of the height of the object can be checked for correctness, the reliability of the measurement can be determined. In the case of a plurality of checks, a probability of the correctness of the measurement can then be determined, which can also be referred to as a determination of the accuracy.

Another advantage of the method described here can be seen in the fact that normalizing the distances to a predetermined value removes the distance dependency of the echoes and therefore simplifies the height classification.

Based on an ascertained accuracy, the accuracy can also be taken into account when the method is used to classify the height of the object, for example in the assessment or in further steps that are dependent on the classification. This can improve a further improvement of driving assistance sequences.

It can furthermore be preferred that method step e) takes place using a support vector machine (SVM classifier) or a neural network. It is thus provided that the method is implemented using machine learning tools and in particular using a neural network and / or an SVM classifier. This can in particular make it possible to obtain the result based on a large number of measurements by means of a classifier and also to carry out a corresponding training beforehand, so that the height of the detected object can be estimated at any time.

It can furthermore be preferred that method step d) is implemented with the geometry of the double echo peaks unchanged, that is, the envelope curve is modified in such a way that in particular the area that includes the second peak is shifted unchanged along the time axis. In this embodiment, the exact peak shape can thus be retained, which allows further information about the peak and thus the classification possibly has a more exact result, since the classifier has further data available with regard to the peak.

Alternatively, it can be provided that the first peak and the second peak of the double echo are modeled as a Gaussian curve, i.e. a Gaussian curve is adapted to the first peak and the second peak and the further evaluation is carried out based on the Gaussian curve . The adjustment of the transit time can thus be done by first adapting the Gaussian curves to the peaks and then shifting the Gaussian curves. The actual object distance and the predetermined object distance to which the classifier is trained are taken into account. The transit time distance of the measured peaks to one another is shifted so that it corresponds to a transit time distance that would be expected with the predetermined object distance. The advantage of using a Gaussian curve can be seen, for example, in a further reduced computing power when modifying the measurement signal.

It can furthermore be preferred that the predetermined value used in method step d) corresponds to a spacing of the peaks from one another which occurs in the case of a tall object which is at a distance from the ultrasonic sensor that is in a range of less than or equal to 2 m. In this embodiment, it is thus taken into account that the distance between the two peaks of the double echo from one another changes depending on the object distance, in particular at distances of greater than 2 m. Therefore, the advantages of the method described here can be particularly effective in this embodiment in particular.

With regard to further technical features and advantages of the method, reference is made to the description of the driving assistance system, the figures and the description of the figures, and vice versa.

A driving support system for a vehicle for supporting a driver of the vehicle is also described, the driving support system having at least one ultrasonic sensor and also a control unit which can be fed with data from the at least one ultrasonic sensor, the driving support system being configured to carry out a method such as this is described in detail above.

The driving assistance system can be part of a vehicle, such as a car or a truck. The driving assistance system has at least one ultrasonic sensor for monitoring the surroundings or for detecting the surroundings. The ultrasonic sensor can preferably be part of the driving support system that also executes the method and can optionally be used together with further sensors for detecting the surroundings, such as radar sensors optical sensors, capacitive sensors or LED sensors can be provided as non-limiting examples.

A control unit such as a processor is also provided. The control unit is suitable for evaluating the sensor data supplied by the sensor or sensors and for acquiring, storing and processing the sensor data. In particular, the control unit is designed to enable the height of a detected object to be classified based on sensor data acquired by the ultrasonic sensor and thus to carry out the method, for example, as described in detail elsewhere.

This can be possible, for example, in that a computer program product for outputting control commands for the driving assistance system of the vehicle is set up to at least partially, for example completely, execute a method described above. The computer program product can have programs or program parts that are loaded into the processor, e.g. of the driving assistance system, and are set up to carry out the above-mentioned method. On the basis of these programs or program parts, it can be made possible that the method described above can be carried out, for example, to output control commands for a vehicle or to carry out further driving sequences.

The driving assistance system described here can thus make it possible in an effective and reliable and therefore particularly safe manner that the height of a detected object, in particular relative to the installation height of the ultrasonic sensor, is determined. As a result, further improvements to the functionalities of the vehicle or the driving support system are possible.

With regard to further technical features and advantages of the driving assistance system, reference is made to the description of the method, the figures and the description of the figures, and vice versa.

Further advantages and advantageous configurations of the subject matter according to the invention are illustrated by the drawings and explained in the following description. It should be noted that the drawings are only of a descriptive character and are not intended to restrict the invention in any way.

• 1 eine schematische Darstellung zeigend eine Ansicht eines Fahrzeugs in einer Fahrzeugumgebung;
• 2 ein Diagramm darstellend die Laufzeiten von Ultraschallsignalen;
• 3 ein detektiertes Messsignal eines Ultraschallsensors;
• 4 das Messsignal aus 3 in modifizierter Form;
• 5 ein weiteres detektiertes Messsignal eines Ultraschallsensors;
• 6 das Messsignal aus 5 in modifizierter Form;

Show it:

• 1 a schematic illustration showing a view of a vehicle in a vehicle environment;
• 2 a diagram showing the transit times of ultrasonic signals;
• 3 a detected measurement signal of an ultrasonic sensor;
• 4th the measurement signal off 3 in modified form;
• 5 another detected measurement signal from an ultrasonic sensor;
• 6th the measurement signal off 5 in modified form;

1 shows a vehicle 10 according to an embodiment of the present invention in a plan view. The vehicle 10 is designed in the present case as a passenger car. The vehicle 10 includes a driving assistance system 12th . The driving support system 12th again comprises a control unit 14th for example, by an electronic control unit (ECU) of the vehicle 10 can be formed. It also includes the driving assistance system 12th at least one ultrasonic sensor 16 .

In the present case, the driving support system comprises 12th eight ultrasonic sensors 16 that distributed to the vehicle 10 are arranged. There are four ultrasonic sensors 16 at a front area 18th of the vehicle 10 and four ultrasonic sensors 16 at a stern area 20th of the vehicle 10 arranged. The ultrasonic sensors 16 are designed to emit ultrasonic waves into the environment 22nd of the vehicle 10 to send and to detect a reflection echo. In particular, there can be a reflection echo from an object 24 in the neighborhood 22nd of the vehicle 10 originate.

The vehicle 10 or its driving support system 12th can at least partially carry out a method that allows the height of the object 24 , in particular relative to the installation height of the ultrasonic sensor 16 , which the ultrasonic radiation on the object 24 sends and accordingly detected the reflection echo to classify.

For this purpose it is used that ultrasonic sensors 16 Send out an ultrasonic signal, such as an ultrasonic pulse, and, depending on the time, that of objects 24 measure reflected waves. The distance between the ultrasonic sensor can be determined via the transit time of the waves 16 and object 24 to be determined. With objects 24 that are the same height as or higher than the ultrasonic sensor 16 , two reflected waves appear, which appear as a double echo or an arrangement of two peaks 26th , 28 , so a first peak 26th and a second peak 28 represent. The waves include a wave with a shorter transit time that is directly from the surface normal of the object 24 is thrown back and resolved in time as the first peak 26th occurs, and another wave that hits the ground at the point of support of the object 24 is reflected and resolved in time as a second peak 28 occurs.

This is in the 2 clarifies which shows a diagram in which on the x-axis the distance of the object 24 from the ultrasonic sensor 16 in meters and the height in meters on the y-axis. The ultrasonic sensor 16 is at a height corresponding to the line 30th of sh = 0.5 m at a horizontal distance of x = 1 m according to the line 32 to the object 24 that is represented by the line 34 with a height h = 0.6 m, which is higher than the installation height sh of the ultrasonic sensor 16 . The distance d of the ultrasonic sensor 16 to the bottom of the object 22nd according to line 36 can be determined with the Pythagorean theorem. This results in a theoretical path difference ds between the peaks 26th , 28 of the double echo of ds = 2 (dx). This is the theoretically determined distance between the two peaks 26th , 28 . If you now change the distance x according to the line 32 , the path difference ds changes and thus also the distance between the two peaks 26th , 28 . For large x are the peaks 26th , 28 closer together and for small x are the peaks 26th , 28 further apart.

According to the invention, a classification of the height of the detected object can be based on this 22nd take place. The procedure for doing this involves the following steps.

First of all, one or more ultrasonic sensors are used 16 Ultrasonic waves or ultrasonic radiation are emitted and these are applied to the object 24 reflected and thrown back. Correspondingly, the on the object 24 reflected ultrasonic waves are detected and recorded as a time-resolved signal.

Such a signal is in the 3 shown, in which a diagram is shown, which a signal of an ultrasonic sensor 16 shows, the X-axis showing the half-path length of the ultrasonic radiation corresponding to the transit time of the signal, i.e. corresponding to the distance of the reflecting point of the object 24 to the ultrasonic sensor 16 and wherein the Y-axis shows the amplitude of the signal in volts, the values being standardized to 1 by way of example. A reflection echo from at the object is thus detected 24 reflected ultrasonic waves and a detection of the time interval between two peaks of a double echo of the reflection echo to one another.

The time-oscillating signal is processed using standard techniques in such a way that we obtain the amplitude as a function of the distance of the reflected wave. The peaks 26th , 28 are also detected using known techniques. The first peak 26th gives the direct distance of the ultrasonic sensor 16 to the object 22nd according to line 32 the end 2 on and the second peak 28 gives the distance of the ultrasonic sensor 16 to the bottom of the object 24 that rests on the floor, according to the line 36 the end 2 . It could in principle be possible that the second peak 28 is reflected directly from another object. Such peak pairs should therefore be identified, which is possible, for example, by means of a transit time-based verification.

The signal in 3 is the same as you would with a tall object 24 received than from an object 24 which is greater than the installation height of the ultrasonic sensor 16 . The two peaks 26th , 28 are each marked with circles. The vertical line 38 marks the position of the first peak detected 26th , i.e. the immediate distance of the ultrasonic sensor 16 to the object 24 according to line 32 the end 2 . The vertical line 40 shows the theoretically expected position of the second peak 28 at which how with reference to 2 discussed is geometrically determinable. Accordingly, it is to be expected when the first peak 26th to the line 38 located in a peak area 42 lies that the second peak 28 in a second peak area 44 is located. The peak area 44 can be defined around the line based on training or given values 40 or the corresponding distance can be adjusted.

In this case the second is peak 28 in the peak area 44 clearly visible. However, it can still be seen that the actual position of the second peak 28 off the line 40 , i.e. the position to be expected, deviates. In order to surround this uncertainty, provision is now made for the distance, in this case for 1.5 m, to be mapped or set to a fixed or predetermined distance of, for example, 1 m. The first peak area is used for this 42 and the second peak region 44 used. Both peak areas 42 , 44 can certainly overlap. For example the second peak area 44 is shifted so far that the distance between the first peak 26th and second peak 28 the horizontal distance between the ultrasonic sensor 16 and object 24 of 1 m. Gamma-distributed noise is inserted in between. The double echo is thus modified by adjusting the time interval between the peaks 26th , 28 of the double echo to a predetermined value.

In 4th is a too 3 corresponding diagram is shown, so that the reference numerals also for 4th apply accordingly. In the diagram of the 4th however, the signal obtained is no longer shown immediately, but rather the modified signal with the shifted second peak area 44 setting a predetermined distance, a normalized second peak and a range 46 which represents inserted noise. The height of the object can thus be classified 24 based on the modified double echo.

This signal is passed into a classifier, such as a neural network or an SVM classifier, for classification. This means that there are no differences depending on the distance between the object and the detector. This has the advantage that it is not necessary to train several classifiers for different distances. This saves a significant amount of time and computation.

In detail can be a classification of the height of the object 24 be done by determining whether at a certain horizontal distance between the ultrasonic sensor 16 and object 24 and thus at a corresponding theoretical distance from the first peak 26th and second peak 28 a defined ratio of the amplitudes of these peaks, in particular determined by training 26th , 28 is present. By having the amplitudes of the peaks 26th , 28 can be determined correctly and only the distance between the peaks 26th , 28 is subject to inaccuracies due to distance, a classification can be carried out in the case of a modified signal. This is done by comparing the amplitudes of the first peak 26th and the second peak 28 or by normalizing the second peak 28 on the first peak 26th .

the 5 again shows a to 3 comparable signal, so basically based on the description 3 is referred. In 5 is the line 38 the position of the first peak 26th and the line 40 again the expected position of the second peak 28 .

In the design according to 5 it is now discussed that often the peaks 26th , 28 are somewhat misshapen or have double points. Therefore a clear determination of the amplitude and the position of the peak is possible 26th , 28 only possible to a limited extent. Now the peaks are provided 26th , 28 by a Gaussian bell curve 48 , 50 to represent or to model as a Gaussian curve. The parameters of this curve are determined by considering the energy of the signal. All values that are greater than half of the largest maximum contribute in the following. The position of the maximum of the Gaussian curve 48 , 50 is calculated by the center of gravity of the energy density. The width of the Gaussian curve is determined by the standard deviation of the signal. The amplitude finally results from the condition that the energy of the measured peak 26th , 28 should be equal to the energy of the Gaussian peak. When all the peaks are determined, the peaks will be 26th , 28 shifted so far that the maximum of the first peak 26th is at one meter. In addition, the second peak 28 shifted by the difference between ds at 1 m and ds at the actual distance. Furthermore, the amplitudes of the peaks 26th , 28 normalized. The area between the peaks is filled with gamma distributed noise.

In 6th are the shifted and normalized peaks 26th , 28 shown. This signal is passed to a classifier for analysis. It might also be possible to use just the parameters of the bell curve as features for a support vector machine (SVM) or a neural network. Since the distance has been corrected to a fixed distance, again only one classifier is required, which saves time and computational effort.

List of reference symbols

10

vehicle

12th

Driving support system

14th

Control unit

16

Ultrasonic sensor

18th

Front area

20th

Stern area

22nd

Surroundings

24

object

26th

first peak

28

second peak

30th

line

32

line

34

line

36

line

38

line

40

line

42

Peak area

44

Peak area

46

area

48

Gaussian curve

50

Gaussian curve

Claims (10)

A method for classifying the height of an object (24) using an ultrasonic sensor (16), the classification comprising the height of the object (24) relative to the installation height of the ultrasonic sensor (16), comprising the method steps: a) emitting ultrasonic waves; b) detecting a reflection echo of ultrasonic waves reflected on the object (24); c) detecting the time interval between two peaks (26, 28) of a double echo from one another of the reflection echo; d) modifying the double echo by setting the time interval between the peaks (26, 28) of the double echo to one another to a predetermined value; and e) classifying the height of the object (24) based on the modified double echo. procedure Claim 1 , characterized in that the method is carried out using previously generated training data. Procedure according to Claim 2 , characterized in that the training data are generated only at the predetermined value of the time interval between the peaks (26, 28) of the double echo. Method according to one of the Claims 2 or 3 , characterized in that the accuracy of the method is checked during training. Method according to one of the Claims 1 until 4th , characterized in that method step e) takes place using a support vector machine (SVM classifier) or a neural network. Method according to one of the Claims 1 until 5 , characterized in that method step d) is implemented with the geometry of the peaks (26, 28) of the double echo unchanged. Method according to one of the Claims 1 until 5 , characterized in that the peaks (26, 28) of the double echo are modeled as a Gaussian curve (48, 50). Method according to one of the Claims 1 until 7th , characterized in that the predetermined value used in method step d) corresponds to a spacing of the peaks from one another which occurs in the case of a high object which is at a distance from the ultrasonic sensor (16) which is in a range of less than or equal to 2m. Method according to one of the Claims 1 until 8th , characterized in that method step e) takes place with normalization of the second peak (28) to the first peak (26) of the double echo. Driving support system for a vehicle (10) for supporting a driver of the vehicle (10), the driving support system (12) having at least one ultrasonic sensor (16) and also a control unit (14) which can be fed with data from the at least one ultrasonic sensor (16) , characterized in that the driving assistance system is designed to provide a method according to one of the Claims 1 until 9 to execute.

Images