DE102019009130A1 - Approximating compression method for ultrasonic sensor data - Google Patents

Abstract

The invention relates to a method for operating an ultrasonic sensor, the associated sensor, the data transmission to a computer system and the associated decompression method in the computer system that controls the ultrasonic sensor. The compression method comprises the steps of acquiring an ultrasonic received signal from an ultrasonic transducer, providing a signal-free ultrasonic echo signal model (610, FIG. 15a), performing the following steps of subtracting a reconstructed ultrasonic echo signal model at least once and optionally repeatedly ( 610) of the ultrasonic received signal (1) and the formation of a residual signal (660), the step of carrying out a method for recognizing signal objects in the residual signal (660), the addition of the parameterized signal profile to the ultrasonic echo signal model (610) a recognized signal object (600 to 605); and the termination of the repetition of these steps if the absolute values of the residual signal are below the absolute values of a predetermined threshold value signal. This is followed by the transmission of at least some of the recognized signal objects in the form of symbols for these recognized signal objects to the computer system, and there the use of at least some of the recognized signal objects. These are preferably reconstructed in the computer system to form a reconstructed ultrasonic echo signal model.

Description

Generic term

The invention is directed to a method for transmitting data from an ultrasonic sensor system for parking aids in vehicles to a computer system and a corresponding device. The basic idea is the detection of potentially relevant structures in the measurement signal and the compression of this measurement signal by transmitting only these identified, potentially relevant structures instead of the measurement signal itself. The actual detection of objects, e.g. Obstacles to the parking process only take place after the measurement signal has been reconstructed as a reconstructed measurement signal in the computer system, where typically several such decompressed measurement signals from several ultrasonic sensor systems come together. Here, the decompression of the data in the control unit is claimed after the compression in the sensor through structure recognition in the measurement signal.

General introduction

Ultrasonic sensor systems for parking aids in vehicles are enjoying increasing popularity. Here, there is a desire to transmit more and more data from the actual measurement signal of the ultrasonic sensor to a central computer system, where these are combined with data from other ultrasonic sensor systems and also from other types of sensor systems, e.g. Radar systems, can be converted into so-called environment or environmental maps by sensor fusion. There is therefore a desire not to carry out object recognition in the ultrasonic sensor system itself, but only through this sensor fusion in the computer system in order to avoid data loss and thus lower the probability of incorrect information and thus incorrect decisions and thus accidents. At the same time, however, the transmission bandwidth of the available sensor data buses is limited. Their exchange should be avoided as they have proven themselves in the field. At the same time, there is therefore a desire not to let the amount of data to be transmitted increase. In short: the information content of the data and its relevance for the obstacle detection (object detection) later running in the computer system must be increased without having to increase the data rate too much - better: without having to increase the data rate at all. Quite the opposite: the data rate requirement should preferably even be reduced in order to allow data rate capacities for the transmission of status data and self-test information from the ultrasonic sensor system to the computer system, which is absolutely necessary in the context of functional safety (FuSi). The disclosure presented here is dedicated to this problem.

Various methods of processing an ultrasonic sensor signal are already known in the prior art.

From the WO 2012/016 834 A1 For example, such a method for evaluating an echo signal for vehicle surroundings detection is known. It is proposed there to transmit a measurement signal with a predeterminable coding and form and to search for the components of the measurement signal in the received signal in the received signal by means of a correlation with the measurement signal and to determine these. The level of the correlation and not the level of the envelope curve of the echo signal is then weighted with a threshold value.

From the DE 4 433 957 A1 It is known to periodically emit ultrasonic pulses for obstacle detection and to deduce the position of obstacles from the transit time, with echoes that remain correlated over several measurement cycles being amplified during the evaluation, while echoes that remain uncorrelated are suppressed.

From the DE 10 2012 015 967 A1 a method for decoding a received signal received by an ultrasonic sensor of a motor vehicle is known, in which a transmitted signal of the ultrasonic sensor is transmitted in coded form and for decoding the received signal is correlated with a reference signal, with a frequency shift of the received signal with respect to the reference signal prior to correlating the received signal with the reference signal Transmission signal is determined and the reception signal is correlated with the transmission signal, which is shifted in frequency by the determined frequency shift, as a reference signal, the same reception signal being subjected to a Fourier transformation to determine the frequency shift of the reception signal and the frequency shift being determined on the basis of a result of the Fourier transformation.

From the DE 10 2011 085 286 A1 is a method for detecting the surroundings of a vehicle by means of ultrasound, whereby ultrasound pulses are emitted and the ultrasound echoes reflected on objects are detected, characterized in that the detection area is divided into at least two distance areas, wherein the ultrasonic pulses used for the detection in the respective distance area are emitted independently of one another and are coded by different frequencies.

From the WO 2014 108 300 A1 A device and a method for environment sensors by means of a signal converter and an evaluation unit are known, signals received from the environment with a first impulse response length at a first point in time during a measurement cycle and with a second, longer impulse response length at a second later point in time within the same measurement cycle, signals being filtered depending on the runtime .

From the DE 10 2015 104 934 A1 a method for providing information is known which depends on an object detected in a surrounding area of a motor vehicle. In the case of the DE 10 2015 104 934 A1 the surrounding area of the motor vehicle is detected with a sensor device of the motor vehicle and information is provided at a communication interface in the motor vehicle. In this case, raw sensor data are stored in a control unit on the sensor device side as information about a free space recognized between the sensor device and an object detected in the surrounding area. These raw sensor data are processed in accordance with the technical teaching of DE 10 2015 104 934 A provided at a communication interface connected to the control unit on the sensor device side for transmission to and for further processing with a further control unit of a processing device comprising the two control units and designed to create a map of the surrounding area.

The technical teachings of the above property rights are all guided by the idea of recognizing an object in front of the vehicle with the aid of the ultrasonic sensor in the ultrasonic sensor and only then transmitting the object data after the objects have been recognized. However, synergy effects are lost when using several ultrasonic transmitters.

From the DE 10 2010 041424 A1 A method for detecting the surroundings of a vehicle with a number of sensors is known, in which at least one echo information about the surroundings is detected by at least one sensor during at least one echo cycle and is compressed with an algorithm, and in which the at least one is compressed Echo information is transmitted to at least one processing unit. Although the focused DE 10 2010 041 424 A1 also to the compressed transmission of recognized object data, but it is already recognized that it makes general sense to transmit data extracted from the received echo signal (echo information) in compressed form to the control unit without specifying a compression method becomes.

From the DE 10 2013 226 376 A1 a method for connecting sensors is known. In section [0006] of the DE 10 2013 226 376 A1 explain the authors of the DE 10 2013 226 376 A1 the state of the art. In claim 1 of the DE 10 2013 226 376 A1 claim the authors of the DE 10 2013 226 376 A1 this self-cited state of the art. From the DE 10 2013 226 376 A1 a method for operating a sensor system, in particular an ultrasonic sensor system, with an ultrasonic sensor and with a control device is known. Data is transmitted from the ultrasonic sensor to the control unit in a current-modulated manner. Data are transmitted from the control unit to the at least one ultrasonic sensor in a voltage-modulated manner.

From the DE 10 024 959 A1 a device for the unidirectional or bidirectional exchange of data between a measuring unit for measuring a physical quantity and a control / evaluation unit is known, which uses the measurement data provided by the measuring unit to determine the physical quantity with a predetermined accuracy. / Evaluation unit connected via a data line. The technical teaching of the DE 10 024 959 A1 The underlying task is to optimize the data transfer between a measuring unit and a remote control / evaluation unit. The object is achieved in that at least one compression unit is provided which is assigned to the measurement unit and which compresses the measurement data supplied by the measurement unit in a predetermined cycle in such a way that, on the one hand, the information content of the measurement data is completely or reduced via the data line to the control / Evaluation unit is transmitted that the physical quantity is determined with the specified accuracy and that, on the other hand, the amount of data transmitted is minimal.

From the DE 10 2013 015 402 A1 a method for operating a sensor device of a motor vehicle is known in which, by means of a sensor, in particular an ultrasonic sensor, the sensor device emits a transmission signal into a surrounding area of the motor vehicle and a signal in the surrounding area ( 7th ) Signal portion of the transmitted signal reflected at a target object ( 11 ) as echo signal ( 12 ) Will be received. Data are transmitted between the sensor and a control unit of the sensor device transmit a data bus. A measured variable for the target object is determined by means of the sensor device. Using a transducer of the sensor, the echo signal is converted into a digital echo signal, and the digital echo signal is transmitted from the sensor via the data bus to the control unit, which determines the measured variable by relating the digital echo signal to a reference signal corresponding to the transmitted signal. The DE 10 2013 015 402 A1 does not provide for any decompression of the data in the control unit.

From the US 2006/0 250 297 A1 a system for sensor fusion is known.

From the WO 2018/210 966 A1 a method for transmitting data via a vehicle bus from an ultrasound system to a data processing device is known, on which the system presented here is essentially based. In this respect, the technical teaching presented here has a large overlap with this document. The claims made on the publication presented here result from the claims. The technical teaching of the WO 2018/210 966 A1 discloses an efficient method of transferring data. However, it does not disclose how object recognition is now carried out.

Also from the DE 10 2018 106 244 B3 a method for the periodically continuous transmission of compressed data of an ultrasound system in a vehicle is known, on which the system presented here is also based in essential parts. In this respect, the technical teaching presented here also has a large overlap with this document. The claims made on the publication presented here result from the claims. The technical teaching of the DE 10 2018 106 244 B3 discloses an efficient method of transferring data. However, it does not disclose how object recognition is now carried out.

The disclosure registered at the same time as the document presented here DE 10 2019 106 190 A1 is essentially the same text with regard to the description of the document presented here. Extensive sections of scripture DE 10 2019 106 190 A1 have an earlier priority date than the document presented here and are therefore expressly not claimed here. The claims of the document presented here are therefore corresponding to the technical teaching of the DE 10 2019 106 190 A1 with the earlier priority date, new technical teaching is now restricted.

task

The invention is therefore based on the object of creating a solution which achieves the above object of reducing the requirement for bus bandwidth for the transmission of measurement data from the ultrasonic sensor system to the computer system and possibly has further advantages.

This object is achieved by means of a method according to claim 1. Further refinements are the subject of the subclaims.

solution

As explained above, the technical teachings from the prior art are all guided by the idea of already recognizing an object in front of the vehicle with the aid of the ultrasonic sensor in the ultrasonic sensor and only then to transmit the object data after the objects have been recognized. However, since synergy effects are lost when using several ultrasonic transmitters, it was recognized in the course of the elaboration of the disclosure presented here that it does not make sense to only transmit the echo data of the ultrasonic sensor itself, but all data and only transfer the data in a central computer system evaluate several sensors. For this, however, the compression of the data for the transmission via a data bus with a smaller bus bandwidth must take place differently than in the prior art. This then enables synergy effects to be developed. For example, it is conceivable that a vehicle has more than one ultrasonic sensor. In order to be able to differentiate between the two sensors, it makes sense if these two sensors transmit with different coding. In contrast to the prior art, however, both sensors should now detect the ultrasonic echoes of both emissions from both ultrasonic sensors and transmit them in a suitably compressed form to the central computer system, where the ultrasonic received signals are reconstructed. The objects are only recognized after the reconstruction (decompression). This also allows the ultrasonic sensor data to be merged with other sensor systems such as radar etc.

A method for transmitting sensor data from a sensor to a computer system is proposed. The method is particularly suitable for use for transmitting data of an ultrasonic reception signal from an ultrasonic sensor to a control device as a computer system in one Vehicle. The procedure is based on the 1 explained. According to the proposed method, an ultrasound burst is first generated and transmitted into a free space, typically in the vicinity of the vehicle (step α of the 1 ). An ultrasonic burst consists of several sound pulses that follow one another at an ultrasonic frequency. This ultrasonic burst is created by the fact that a mechanical oscillator in an ultrasonic transmitter or ultrasonic transducer slowly starts to oscillate and then oscillates again. The ultrasonic burst transmitted in this way by the exemplary ultrasonic transducer is then reflected on objects in the vicinity of the vehicle and received as an ultrasonic signal by an ultrasonic receiver or the ultrasonic transducer itself and converted into an electrical received signal (step β of the 1 ). The ultrasound transmitter is particularly preferably identical to the ultrasound receiver and is then referred to below as the transducer. The principle explained below can also be used for separate receivers and transmitters. In the proposed ultrasonic sensor there is a signal processing unit which now analyzes and compresses the electrical received signal received in this way (step γ of the 1 ), in order to minimize the necessary data transmission and to create space for the said status messages and further control commands from the control computer to the signal processing unit or the ultrasonic sensor system. The compressed electrical reception signal is then transmitted to the computer system (step δ of the 1 ).

The associated method is used to transmit sensor data, in particular an ultrasonic sensor, from a sensor to a computer system, in particular in a vehicle. It begins by sending out an ultrasonic burst (step α of the 1 ). An ultrasonic signal is then received and an electrical reception signal is generated (step β of the 1 ) as well as performing a data compression of the received signal (step γ of the 1 ) to generate compressed data (step γ of the 1 ) and the detection of at least two or three or more predetermined properties. The electrical received signal is preferred by sampling (step γa of the 2 ) is converted into a sampled received signal, which consists of a time-discrete stream of sampled values. A sampling time can typically be assigned to each sampling value as a time stamp of this sampling value. The compression can be done, for example, by a wavelet transformation (step γb of the 2 ) respectively. For this purpose, the received ultrasonic signal in the form of the sampled received signal can be compared with predetermined basic signal forms, which are stored in a library, for example, by forming a correlation integral (see also Wikipedia on this term) between the predetermined basic signal forms and the sampled received signal. The predetermined basic signal forms are also referred to below as signal object classes. By forming the correlation integral, associated spectral values of this prototypical signal object class are determined for each of these prototypical signal object classes. Since this happens continuously, the spectral values themselves represent a stream of time-discrete instantaneous spectral values, with each spectral value being able to be assigned a time stamp. An alternative, but mathematically equivalent method is the use of optimal filters (English matched filter) for each predetermined signal object class (basic signal form). Since usually several prototypical signal object classes are used, which can also be subjected to different temporal spreads (see also "Wavelet Analysis"), this typically results in a time-discrete stream of multidimensional vectors of spectral values of different prototypical signal object classes and their different respective temporal classes Spreads. A time stamp is assigned to each of these multidimensional vectors. Each of these multidimensional vectors is a so-called feature vector. It is therefore a time-discrete stream of feature vectors. A time stamp is preferably assigned to each of these feature vectors. (Step γb of the 2 )

The ongoing time shift therefore also results in a time dimension. As a result, the feature vector of the spectral values can also be supplemented by values from the past or values that depend on them, for example time integrals or derivatives or filter values of one or more of these values, etc. This can further increase the dimension of these feature vectors within the feature vector data stream. In order to keep the effort in the following small, it makes sense to restrict it to a few prototypical signal object classes during the extraction of the feature vectors from the scanned input signal of the ultrasonic sensor. Thus, for example, matched filters can then be used to continuously monitor the occurrence of these prototypical signal object classes in the received signal.

The isosceles triangle and the double point can be named as particularly simple prototypical signal object classes. A prototypical signal object class generally consists of a predetermined spectral coefficient vector, that is to say a predetermined prototypical feature vector value.

To determine the relevance of the spectral coefficients of a feature vector of an ultrasonic echo signal, the amount of a distance between these properties, the elements of the vector of the current spectral coefficients (feature vector), is determined for at least one prototypical combination of these properties (prototype) in the form of a prototypical signal object class, which is symbolized by a predetermined prototypical feature vector (prototype or prototype vector) from a library of predetermined prototypical signal object class vectors (step γd of 2 ). The spectral coefficients of the feature vector are preferably normalized prior to correlation with the prototypes (step γc of the 2 ). The distance determined in this distance determination can, for example, consist of the sum of all differences between a respective spectral coefficient of the given prototypical feature vector (prototype or prototype vector) of the respective prototype and the corresponding standardized spectral coefficient of the current feature vector of the ultrasonic echo signal. A Euclidean distance would be formed by the root of the sum of the squares of all differences between one spectral coefficient of the given prototypical feature vector (prototype or prototype vector) of the prototype and the corresponding normalized spectral coefficient of the current feature vector of the ultrasonic echo signal. However, this distance formation is usually too complex. Other methods of spacing are conceivable. A symbol and possibly also a parameter, for example the distance value and / or the amplitude, can then be assigned to each predefined prototypical feature vector (prototype or prototype vector) before normalization. If the distance determined in this way falls below a first threshold value and if it is the smallest distance between the current feature vector value and one of the specified prototypical feature vector values (prototypes or values of the prototype vectors), its symbol will continue to be used as a recognized prototype. A pairing of the recognized prototype and time stamp of the current feature vector is thus created. The data are then preferably transmitted (step δ of the 2 ), here the determined symbol that best symbolizes the recognized prototype and, for example, the distance and the time of occurrence (time stamp) to the computer system only if the amount of this distance is below the first threshold value and the recognized prototype is one to be transmitted Prototype is. It is possible that prototypes that cannot be recognized, for example for noise, for example the absence of reflections, etc., are stored. These data are irrelevant for obstacle detection and should therefore not be transmitted if necessary. A prototype is recognized when the amount of the determined distance between the current feature vector value and the specified prototypical feature vector value (prototype or value of the prototype vector) is below this first threshold value (step γe of the 2 ). The ultrasonic echo signal itself is no longer transmitted, but only a sequence of symbols for recognized typical time signal curves and time stamps associated with these signal curves in a specific time segment (step δ of the 2 ). It is then preferred for each recognized signal object only one symbol for the recognized signal form prototype, its parameters (e.g. amplitude of the envelope (of the ultrasonic echo signal) and / or temporal stretching) and a time reference point of the occurrence of this signal form prototype (the time stamp) as transmitted signal object detected. The transmission of the individual sampled values or of times at which threshold values are exceeded by the envelope curve of the sampled received signal (the ultrasonic echo signal), etc., is omitted. In this way, this selection of the relevant prototypes leads to massive data compression and a reduction in the required bus bandwidth.

A quantitative recording of the presence of a combination of properties takes place with the formation of an estimated value - here e.g. the inverse distance between the representative of the prototypical signal object class in the form of the specified prototypical feature vector (prototype or prototype vector) - and the subsequent transmission of the compressed data to the computer system if the amount of this estimated value (e.g. inverse distance) is above a second threshold value or the inverse estimate is below a first threshold value. The signal processing unit of the ultrasonic sensor thus performs data compression on the received signal to generate compressed data.

The determination of the distance should be explained again here for better clarity.

This distance determination is also known as a classification from the signal processing of statistical signals. Examples of classifiers include logistic regression, the cuboid classifier, the distance classifier, the closest neighbor classifier, the polynomial classifier, the cluster method, artificial neural networks, and latent class analysis.

An example of a classifier is in 7th shown.

An exemplary physical interface ( 101 ) controls an exemplary ultrasonic transducer ( 100 ) and causes it to send out an exemplary ultrasonic transmission pulse or Ultrasonic bursts. The ultrasonic transducer ( 100 ) does not receive the from one in 7th The drawn object reflected exemplary ultrasonic transmission pulse or ultrasonic transmission burst with an amplitude change, a delay and typically with a distortion which is caused by the nature of the reflecting object. In addition, there are typically several objects in the vicinity of the vehicle that contribute to the reflected ultrasonic wave in a modifying manner. The ultrasonic transducer ( 100 ) converts the received reflected ultrasonic wave into an ultrasonic transducer signal ( 102 ) around. The physical interface converts this ultrasonic transducer signal, typically through filtering and / or amplification, into an ultrasonic echo signal ( 1 ) around. The feature vector extractor ( 111 ) extracted from this ultrasonic echo signal ( 1 ). The physical interface preferably transmits ( 101 ) to the feature vector extractor ( 111 ) the ultrasonic echo signal ( 1 ) begins as a time-discrete received signal consisting of a sequence of sample values. A date (time stamp) is preferably assigned to each sample value. The feature vector extractor ( 111 ) shows in the example 7th a plurality of m (with m as an integer positive number) optimal filters (optimal filter 1 to optimal filter m). These serve to determine m intermediate parameter signals ( 123 ) each relating to the presence of a basic signal object preferably assigned to the respective intermediate parameter signal, preferably with the help of a suitable filter (e.g. an optimal filter) from the sequence of sampled values of the ultrasonic echo signal ( 1 ). The resulting intermediate parameter signals ( 123 ) are also designed as a time-discrete sequence of respective intermediate parameter signal values, which are preferably each correlated with a date (time stamp). Thus, exactly one time date (time stamp) is preferably assigned to each intermediate parameter signal value.

A subsequent increase in significance ( 125 ) performs a matrix multiplication of the vector of the intermediate parameter signal values of the intermediate parameter signal ( 123 ) with a so-called LDA matrix ( 126 ) by. This is typically determined at the time of construction using statistical methods of statistical signal processing and pattern recognition. In this way, the significance enhancer generates the feature vector signal ( 138 ). Here it forms the m intermediate parameter signal values on n (with n as an integer) parameter signal values of the feature vector signal ( 138 ) from. Typically the feature vector signal ( 138 ) from n parameter signals. Preference is given to n <m. At this point it should only be noted that the term “feature vector” is often referred to as a feature vector in statistical signal theory and in pattern recognition. The feature vector signal ( 138 ) is thus, inter alia, as a time-discrete sequence of feature vector signal values with n parameter signal values each of the preferably n parameter signals of the feature vector signal ( 138 ) formed, which included the parameter signal values and further parameter signal values each with the same date (time stamp). Here, n is the dimension of the individual feature vector signal values, which are preferably the same from one feature vector value to the next feature vector value. In this sense, a feature vector signal value is a vector with a time stamp which comprises several, preferably n, parameter signal values. Each feature vector signal value formed in this way is assigned this respective date (time stamp).

The evaluation of the time course of the feature vector signal now follows ( 138 ) in the resulting n-dimensional phase space as well as inferring a recognized basic signal object while determining an evaluation value (eg the distance).

To do this, a distance finder (or classifier) compares ( 112 ) the current feature vector signal value of the feature vector signal ( 138 ) with a plurality of prototypes previously stored in a prototype database ( 115 ) stored feature vector signal value prototypes. This is explained in more detail below. The distance finder (or classifier) ( 112 ) determines an evaluation value for each of the examined feature vector signal value prototypes of the prototype database ( 115 ), which indicates the extent to which the respective feature vector signal value prototype of the prototype database ( 115 ) equals the current feature vector signal value. This evaluation value is referred to below as the distance. The feature vector signal value prototypes are preferably each assigned to exactly one basic signal object. The feature vector signal value prototype of the prototype database ( 115 ) with the smallest distance to the current feature vector signal value then best matches it. If the distance is smaller than a specified threshold value, this feature vector signal value prototype represents the prototype database ( 115 ) the signal base object recognized with the highest probability ( 121 ).

This recognition process is carried out several times in succession, so that a determined signal basic object sequence is derived from the chronological sequence of the recognized signal basic objects ( 121 ) results.

This is followed by the determination of the signal object that is probably present ( 122 ) by determining that sequence of specified signal basic object sequences of a signal object database ( 116 ), which is most similar to the determined signal base object sequence. As previously explained, a signal object consists of one temporal sequence of basic signal objects. The signal object is typically predefined with a symbol in the signal object database ( 116 ) assigned.

For example, this estimation of the signal basic object sequence can be done by a Viterbi estimator ( 113 ) respectively. In the simplest case, the number of basic signal objects recognized within a specified period of time which, according to their position in the sequence of recognized basic signal objects, is compared with the position of an expected basic signal object in a signal object database ( 116 ) expected sequence of basic signal objects specified as a given signal object match minus the number of basic signal objects recognized within the specified period of time which, according to their position in the sequence of basic signal objects recognized, correspond to the position of an expected basic signal object in a signal object database ( 116 ) the expected sequence of basic signal objects given as a given signal object do NOT match, taken as the evaluation value for the match. In this way, the Viterbi estimator determines for each of the specified signal objects in the signal object database ( 116 ) an evaluation value, whereby the signal objects of the signal object database ( 116 ) consist of predetermined sequences of expected basic signal objects and are correlated with a respective symbol.

Figuratively speaking, it is checked here whether the point to which the n-dimensional feature vector signal ( 138 ) in the n-dimensional phase space shows, on its way through the n-dimensional phase space in a predetermined time sequence, it approaches predetermined points in this n-dimensional phase space closer than a predetermined maximum distance. The feature vector signal ( 138 ) thus has a course over time. An evaluation value (for example a distance) is then calculated, which can, for example, reflect the probability of the existence of a specific sequence. It is then preferably carried out within the Viterbi estimator ( 113 ) the comparison of this evaluation value, which is again assigned a time date (time stamp), with a threshold value vector with formation of a Boolean result that can have a first and a second value. If this Boolean result has the first value for this time date (time stamp), the symbol of the signal object and the time date (time stamp) assigned to this symbol are transmitted from the sensor to the computer system. This causes the detected signal object ( 122 ) preferably transmitted with its parameters. Possibly. further parameters can be transferred depending on the detected signal object.

At this point we would like to return to the processing of the feature vector signal ( 138 ). The ultrasonic echo signal is preferred ( 1 ) a sequence of quantization vectors - the ultrasonic echo signal values - whose components, the measured parameters, will in general, however, not be completely independent of one another. Each ultrasonic echo signal value of an ultrasonic echo signal ( 1 ) by itself usually has too little selectivity for precise signal object detection in complex contexts. This is precisely the shortcoming in the prior art. By creating one or more such quantization vectors from the continuous stream of analog physical values of the ultrasonic echo signal ( 1 ) at typically regular or regular intervals through the physical interface ( 101 ) (please refer 7th ) the multi-dimensional ultrasonic echo signal value data stream quantized in terms of time and value is created in the form of the ultrasonic echo signal ( 1 ).

This multidimensional ultrasonic echo signal ( 1 ) in the form of one or more streams of quantization vectors is first subdivided into individual frames of a defined length, filtered, normalized, then orthogonalized and, if necessary, suitably distorted by a non-linear mapping - e.g. logarithmization and cepstrum analysis - in a first exemplary processing step. This is due to the block of the matched filter in the feature vector extractor ( 111 ) of the 7th indicated. Instead of the block of the block of the optimal filter, you can also use other signal processing structures for generating the intermediate parameter signals ( 123 ) imagine. For example, derivations of the ultrasonic echo signal values generated in this way can also be formed here. Finally there is an increase in significance of the determined intermediate parameter signal ( 123 ) in a unit of significance increase ( 125 ) to the actual feature vector signal ( 138 ) instead of. As described, this can be done, for example, by multiplying the multi-dimensional quantification sector with a so-called, predetermined LDA matrix ( 126 ) respectively.

1. a) durch ein Neuronales Netz oder
2. b) durch einen HMM-Erkenner
3. c) durch ein Petri-Netz

The next step of recognition in the distance finder (or classifier) ( 112 ), for example, can be done in two different ways:

1. a) through a neural network or
2. b) by an HMM recognizer
3. c) by a Petri net

• Mit Hilfe der besagten vordefinierten LDA-Matrix (126) wird der Zwischenparameterdatenstrom (123) so vom mehrdimensionalen Eingangsparameterraum auf einen neuen Parameterraum durch den Signifikanzsteigerer (125) abgebildet, wodurch seine Selektivität maximal wird. Die Komponenten der dabei gewonnenen neuen transformierten Merkmalsvektoren werden hierbei nicht nach realen physikalischen oder sonstigen Parametern, sondern nach maximaler Signifikanz ausgewählt, was besagte maximale Selektivität zur Folge hat.

Here again the HMM recognizer ( 7th ) described:

• With the help of the predefined LDA matrix ( 126 ) the intermediate parameter data stream ( 123 ) so from the multi-dimensional input parameter space to a new parameter space through the significance enhancer ( 125 ), which maximizes its selectivity. The components of the new transformed feature vectors obtained in this way are not selected according to real physical or other parameters, but rather according to maximum significance, which results in said maximum selectivity.

The LDA matrix ( 126 ) is usually calculated beforehand at the time of construction on the basis of sample data streams with known signal object data records, that is, data records that have been obtained with specified structures of the signal course, beforehand by a training step offline.

If care is taken that all elements of the distance finder (or the classifier) ( 112 ) perform at least locally reversible functions performed method, then deviations in the signal course can be taken into account in the form of an approximately linear transformation function.

The prototypes from sample data streams for the given prototypical signal curves (prototypical basic signal objects) in the coordinates of the new parameter space are calculated in the construction phase and stored in a prototype database ( 115 ) for later recognition. In addition to these statistical data, this can also contain instructions for a computer system as to what should happen in the event of a successful or unsuccessful detection of the respective basic signal object prototype. As a rule, the computer system will be the computer system of the sensor system.

The feature vector signal values of the feature vector signal ( 138 ) that the feature extractor ( 111 ) for these prototypical signal curves of the given basic signal objects of the ultrasonic echo signal ( 1 ) are issued in the laboratory, are therefore in this prototype database ( 115 ) as signal base object prototypes.

1. 1. Entspricht der erkannte Feature-Vektor-Wert des Feature-Vektorsignals (138) einem der vorgespeicherten Signalgrundobjektprototypen der Prototypendatenbank (115) oder nicht und mit welcher Wahrscheinlichkeit und Zuverlässigkeit?
2. 2. Wenn es einer der bereits gespeicherten Signalgrundobjektprototypen der Prototypendatenbank (115) ist, welcher ist es und mit welcher Wahrscheinlichkeit und Zuverlässigkeit?

In later operation, the feature vector signal values of the feature vector signal ( 138 ) now with these previously stored, i.e. learned signal basic object prototypes ( 115 ) For example, by calculating the Euclidean distance between a quantization vector in the coordinates of the new parameter space and all of these previously stored basic signal object prototypes ( 115 ) in the distance finder ( 112 ) compared. At least two detections are made here:

1. 1. Corresponds to the recognized feature vector value of the feature vector signal ( 138 ) one of the pre-stored signal basic object prototypes of the prototype database ( 115 ) or not and with what probability and reliability?
2. 2. If there is one of the already saved basic signal object prototypes in the prototype database ( 115 ) is, which one is it and with what probability and reliability?

In order to make the first recognition, dummy prototypes are usually also stored in the prototype database ( 115 ) the basic signal object prototypes are stored, which should cover all parasitic parameter combinations occurring during operation as far as possible. The above-mentioned basic signal object prototypes are saved in the prototype database ( 115 ).

For the signal base object prototypes of the prototype database ( 115 ) at the time of construction the distance can be determined according to the distance calculator ( 112 ) applied procedure for each pairing from two different signal base object prototypes of the prototype database ( 115 ) be determined. There is then a minimal prototype distance. This is stored in the prototype database ( 115 ) or in the distance finder, preferably also halved as half the minimum prototype distance.

For example, if this minimum half prototype distance is determined by the distance calculation ( 112 ) determined distance between the current feature vector signal value of the feature vector signal ( 138 ) and a signal base object prototype from the prototype database ( 115 ), this signal basic object prototype is assessed as recognized. From this point in time it can be ruled out that further distances calculated in the course of a further search to other signal base object prototypes in the prototype database ( 115 ) can deliver even smaller distances. The search can then be aborted, which halves the time on average and thus saves the sensor's resources.

The minimum Euclidean distance can be calculated using the following formula, for example: $$\mathrm{D.}is{t}_{\mathrm{F.}V\_\mathrm{C.}b\mathrm{E.}}=\underset{\mathrm{C.}b\_cnt=\mathrm{C.}b\_anz}{\overset{1}{{\displaystyle \mathrm{M.}in}}}\left[{{\displaystyle \sum _{\text{dim\_}cnt=\text{dim}}^1\left(\mathrm{F.}{V}_{\text{dim\_}cnt}-\mathrm{C.}{b}_{\mathrm{C.}b\_cnt,din\_cnt}\right)}}^2\right]$$

Dim_cnt stands for the dimension index up to the maximum dimension of the feature vector ( 138 ) dim passes.

FV _{dim_cnt} stands for the parameter value of the feature vector signal value of the feature vector ( 138 ) according to the index dim_cnt.

Cb_cnt stands for the number of the signal basic object prototype in the prototype database ( 115 ).

Cb _{CB_cnt} , _{dim_cnt} accordingly stands for the parameter value corresponding to dim_cnt of the entry of the signal basic object prototype in the prototype database ( 115 ), which is assigned to the signal base object prototype corresponding to the Cb_cnt.

Dist _{FV_CbE} stands for the minimal Euclidean distance obtained here as an example. When searching for the smallest Euclidean distance, the number Cb_cnt that produces the smallest distance is noted.

An exemplary assembler code is given for clarification: Beginning of code Mov Cb_cnt, # Cb_anz initialize prototype database - vector counter Mov C, # 0 initialize register C with 0 Mov dist, maxvalue initialize the distance with maximum value Mov num, not_valid_num initialize nearest neighbor number with invalid value Mov Cb_adr, Cb_badr initialize prototype database address with the base address of the prototype database Label_A: // next vector Mov SP, # 0 initialize cache Mov dim_cnt, #dim initialize dimension counter with feature vector dimension Label B: // next dimension MovA, $ Cb_adr load value absolutely from prototype database address SubA, $ FV_adr, dim_cnt subtract value absolutely relative from feature vector value Mov BA load B register with result MulA B multiply A and B (= A ² ) AddA, SP add the result to the intermediate result Mov SP, A and remember Dec dim_cnt next vector component Inc Cb_adr increment prototype database pointer by one jnz dim_cnt, label B but only if it wasn't the last Cmp SP, dist evaluate prototype database entry (signal base object prototypes) jmpgt label C Mov dist, SP if better entry than previous optimum Label Mov num, Cb_cnt Note the entry number and spacing C: dec_Cb_cnt next prototype database entry jnz Cb_cnt, label A but only if it wasn't the last End of code

The degree of confidence for correct detection is derived from the spread of the underlying basic data streams for a signal basic object prototype and the distance between the current feature vector value of the feature vector signal ( 138 ) from their focus.

8th demonstrates different detection cases. For the sake of simplicity, the representation is selected for a two-dimensional feature vector signal in which each feature vector signal value comprises a first parameter value and a second parameter value. This is only used here to better represent the methodology on a two-dimensional sheet of paper.In reality, the feature vector signal values of the feature vector signal ( 138 ) typically always multidimensional.

The focus is on various prototypes ( 141 , 142 , 143 , 144 ). In the prototype database ( 115 ), as described above, half the minimum distance between these basic signal object prototypes in the prototype database ( 115 ) must be saved. This would then be a global parameter that applies to all signal base object prototypes in the prototype database ( 115 ) would be equally valid. This decision is made with the help of this minimal distance. However, it assumes that the scattering of the signal base object prototypes, which is caused by their center of gravity ( 141 , 142 , 143 , 144 ) are marked, are more or less identical. Is the distance determination (or other evaluation) by the distance finder ( 112 ) (or classifier) is optimal, this is also the case. This would correspond to a circle around the center of gravity ( 141 , 142 , 142 , 144 ) each of the signal base object prototypes of the prototype database ( 115 ).

However, this can rarely be achieved in reality. An improvement in the recognition performance can therefore be achieved if the spread for the signal base object prototype of the prototype database ( 115 ) would be saved in each case. This would correspond to a circle around each of the basic signal object prototypes in the prototype database ( 115 ) with a specific radius for the basic signal object prototype. The disadvantage is an increase in computing power.

A further improvement in the recognition performance can be achieved if the spread for the signal base object prototype of the prototype database ( 115 ) is modeled by an ellipse. For this purpose, instead of the radius, as before, the main axis diameter of the scattering ellipse and its tilting against the coordinate system in the prototype database ( 115 ) for preferably each of the basic signal object prototypes of the prototype database ( 115 ) can be saved. The disadvantage is a further massive increase in computing power and memory requirements.

Of course, the calculation can be made even more complicated, but this usually only increases the effort massively and the recognition performance for the basic signal object prototypes of the prototype database ( 115 ) no longer increases significantly.

It is therefore recommended here to use the simplest of the options described.

The position of the distance calculation ( 112 ) in the exemplary two-dimensional parameter space of 8th determined current feature vector signal values of the feature vector signal ( 138 ) can now be very different. It is conceivable that such a first such exemplary feature vector signal value ( 146 ) too far from the center of gravity coordinates ( 141 , 142 , 143 , 144 ) the center of gravity of any signal basic object prototype in the prototype database ( 115 ) is away. This distance threshold value can, for example, be said minimum half prototype distance. It can also be that the scatter areas of the signal base object prototypes are centered around their respective focal points ( 143 , 142 ) overlap and a second exemplary determined current feature vector signal value ( 145 ) of the feature vector signal ( 138 ) lies in the overlap area. In this case, a hypothesis list can contain both signal basic object prototypes with different probabilities as attached parameters, since they have different distances. So there is then not a signal basic object as the most likely to the Viterbi estimator ( 113 ), but a vector from possibly existing basic signal objects to the Viterbi estimator ( 113 ) to hand over. From the chronological sequence of these hypothesis lists, the Viterbi estimator then searches for the possible sequence that has the greatest probability of one of the predetermined signal basic object sequences in its signal object database compared to all possible paths through which it has recognized as possible Basic signal objects ( 121 ) from the distance detector ( 112 ) by the Viterbi estimator ( 113 ) received hypothesis lists. In this case, exactly one recognized basic signal object prototype of this hypothesis list must be run through this path in such a path for each hypothesis list.

In the best case, the current feature vector signal value is ( 148 ) in the scatter range (threshold ellipsoid) ( 147 ) around the center of gravity ( 141 ) of a single signal base object prototype ( 141 ) which is thus safely through the distance finder ( 112 ) recognized and recognized as a recognized basic signal object ( 121 ) to the Viterbi estimator ( 113 ) is passed on.

It is conceivable, in order to improve the modeling of the scatter range of an individual signal base object prototype, to model it using several here circular signal base object prototypes with associated scatter ranges. Several basic signal object prototypes from the prototype database ( 115 ) represent the same basic signal object prototype in the understanding of a basic signal object class. The danger here is that, due to the division of the probability of a basic signal object prototype into several such basic subsignal object prototypes, the probability of the individual basic subsignal object prototype can become smaller than that of another basic signal object prototype, the probability of which was lower than that of the basic signal object prototype. This other signal basic object prototype can thus possibly be falsely established.

• Ein entscheidender Punkt ist, dass der Rechenaufwand mit Cb_anz * dim steigt.
• Bei einem nicht optimierten HMM-Erkenner beträgt die Anzahl der Assemblerbefehle, die ausgeführt werden müssen, um eine Vektorkomponente zu berechnen, ca. 8 Schritte.

Another major problem is the computing power that has to be made available to create the basic signal object prototypes in the prototype database ( 115 ) sure to recognize. This should be discussed a little more:

• A crucial point is that the computational effort increases with Cb_anz * dim.
• With a non-optimized HMM recognizer, the number of assembly instructions that must be executed to compute a vector component is approximately 8 steps.

The number A_Abst of the assembler steps required to calculate the distance between an individual signal base object prototype (CbE) in the prototype database and a single feature vector signal value (FV) is calculated roughly as follows: $$\text{A\_Abst}=\text{FV\_Dimension}*\mathrm{8th}+\mathrm{8th}$$

This leads to the number A_CB of assembler steps for the determination of the basic signal object prototype of the prototype database ( 115 ) with the smallest distance: $$\text{A\_CB}=\text{Cb\_anz}*\left(\text{A\_Abst}\right)+\mathrm{4th}=\text{Cb\_Anz}\left(\text{FV\_Dimension * 8 + 8}\right)+\mathrm{4th}$$

Using the example of a medium-sized HMM recognizer with 50,000 signal base object prototypes (number of signal base object prototype entries in the prototype database = CB_anz) and 24 FV_Dimensions (number of parameter values in a feature vector signal value = feature vector dimension = FV_Dimension), the number of steps is : $$\begin{array}{c}50000*\left(24*\mathrm{8th}+\mathrm{8th}\right){\mathrm{4th}}^{\sim }10\text{Millions of operations per feature vector signal value of the feature}\\ \text{Vector finals}\left(138\right)\end{array}$$

At a relatively low sampling rate of 8kHz = 8000 FV per second (feature vector signal values of the feature vector signal ( 124 ) per second) you need a computing power of 8GlpS (8 billion instructions per second).

In view of the challenges of saving energy in electromobility and / or reducing the CO2 photo print, this is not acceptable.

In the case of an optimized HMM detection method being carried out by the distance finder ( 112 ) or the classifier ( 112 ) therefore, as already mentioned, the smallest distance between two signal base object prototypes in the prototype database ( 115 ) pre-calculated and in the prototype database ( 115 ) or in the distance calculation ( 112 ) filed. This has the advantage that the search is then carried out by the Distance finder ( 112 ) can be canceled if there is a distance between a current feature vector signal value of the feature vector signal ( 138 ) and a signal base object prototype from the prototype database ( 115 ) by the distance finder ( 115 ) that is smaller than half of this smallest distance. This halves the mean search time for the distance finder ( 112 ). Further optimizations can be made if the prototype database ( 115 ) according to the statistical occurrence of the basic signal object prototypes in real ultrasonic echo signals ( 1 ) is sorted. This ensures that the most common signal base object prototypes can be found much faster, which reduces the computing time of the distance finder ( 112 ) or classifier ( 112 ) is further shortened and power consumption is further reduced.

For a distance finder that executes such an optimized HMM recognition process, the computing power requirement is now as follows:

Again there are 8 steps to calculate the distance of a vector component. The steps for calculating the distance A_Abst of a signal base object prototype entry (CbE) in the prototype database ( 115 ) to the current feature vector signal value (FV) of the feature vector signal ( 138 ) are again: $$\text{A\_Abst}=\text{FV\_Dimension}*\mathrm{8th}+\mathrm{8th}$$

The number of steps for determining the signal base object prototype entry in the prototype database (115) with the smallest distance A_CB with optimization is a little higher: $$\text{A\_CB}=\text{Cb\_anz *}\left(\text{A\_Abst}\right)+\mathrm{4th}=\text{Cb\_anz}*\left(\text{FV\_Dimension}*\mathrm{8th}+10\right)+\mathrm{4th}$$

The two additional assembler commands are necessary to check whether the determined distance between the current feature vector signal value and the signal base object prototype of the prototype database that has just been examined ( 115 ) below half the smallest distance between the signal base object prototypes of the prototype database ( 115 ) is located.

Furthermore, the number CB_Anz of the basic signal object prototype entries in the prototype database ( 115 ) for mobile and energy self-sufficient applications on 4000 prototype database entries of signal base object prototypes in the prototype database ( 115 ) or even less restricted.

In addition, the number of feature vector signal values per second within the feature vector signal ( 138 ) by filtering in the feature vector extractor ( 111 ) and lowering the sampling rate in the feature vector extractor ( 111 ) lowered.

• Der besagte Abstandsermittler (112) oder Klassifikator (112), der ein mittleres HMM-Erkennungsverfahren ausführt, wird nun mit einer Prototypendatenbank (115) mit nur noch einem knappen Zehntel der Einträge z.B. mit 4000 Einträgen (CbE) und weiterhin mit 24 Feature-Vektor-Signal-Dimensionen (also 24 Parametersignalen) betrieben.

This is explained using a simple example:

• The said distance finder ( 112 ) or classifier ( 112 ), which carries out a medium HMM recognition process, is now used with a prototype database ( 115 ) operated with just under a tenth of the entries, e.g. with 4000 entries (CbE) and still with 24 feature vector signal dimensions (i.e. 24 parameter signals).

The number of steps is now $$\begin{array}{c}4000*\left(24*\mathrm{8th}+10\right)+{\mathrm{4th}}^{\sim }\underset{\_}{808004}\text{Operations per feature vector signal value of the feature vector signal}\\ \left(138\right)\end{array}$$

When the feature vector signal value rate is reduced to 100 feature vector signal values per second extracted from a 10ms time window in the feature vector extractor ( 111 ) over, for example, 80 samples each and the search is terminated when the distance between the current feature vector signal value and the processed signal base object prototype in the prototype database ( 115 ) is below half the smallest prototype database entry distance, the effort is at least halved if the prototype database is sorted appropriately ( 115 ).

The required computing power then drops to <33DSP-Mips (33 million operations per second). In reality, the prototype database sorting leads to even lower computing power requirements of for example 30Mips. This makes the system real-time capable and can be integrated into a single IC and thus into a sensor in the first place.

The search area can be restricted by preselection. A prerequisite is an even distribution of the data = focal points of the quadrants in the geometric quadrant center.

• Eine Verringerung der Eintragsanzahl der Prototypendatenbank (115) erhöht sowohl die False-Acceptance-Rate (FAR), also die falschen Signalgrundobjektprototypen, die als Signalgrundobjektprototypen akzeptiert werden, als auch die False-Rejection-Rate (FRR), also die tatsächlich vorhandenen Signalgrundobjektprototypen, die nicht erkannt werden.

The necessary reduction in the size of the prototype database ( 115 ) has advantages and disadvantages:

• A reduction in the number of entries in the prototype database ( 115 ) increases both the false acceptance rate (FAR), i.e. the wrong basic signal object prototypes that are accepted as basic signal object prototypes, and the false rejection rate (FRR), i.e. the actually existing basic signal object prototypes that are not recognized.

On the other hand, this reduces the resource requirements (computing power, chip areas, memory, power consumption, etc.).

In addition, the previous history, i.e. the previously recognized basic signal object prototypes, can be used when forming hypotheses by the distance calculator ( 112 ) can be used. A suitable model for this is the so-called Hidden Markov Model (HMM).

For each signal base object prototype, a degree of confidence and a distance to the measured current feature vector signal value of the feature vector signal ( 138 ) derived from the Viterbi estimator ( 113 ) can also be further processed. It is also useful to create a list of hypotheses for each of the recognized basic signal object prototypes ( 121 ), which contains, for example, the ten most likely signal basic object prototypes with the respective probability and reliability of detection.

Since not every temporal and spatial signal basic object sequence can be assigned to a signal object and is therefore meaningful, it is possible to determine the temporal sequence of the hypothesis lists of the successive frames using a Viterbi estimator ( 113 ) to evaluate.

Here, the signal base object prototype sequence path can be found through the successive hypothesis lists that has the highest probability and is in the signal object database ( 116 ) of the Viterbi estimator ( 113 ) is available.

1. 1. Ist die wahrscheinlichste Folge von Signalgrundobjektprototypen eine der bereits eingespeicherten Folge von Signalgrundobjektprototypen oder nicht und mit welcher Wahrscheinlichkeit und Zuverlässigkeit?
2. 2. Wenn sie eine der bereits gespeicherten Folge von Signalgrundobjektprototypen ist, welche ist es und mit welcher Wahrscheinlichkeit und Zuverlässigkeit?

Here, too, at least two detections are made:

1. 1. Is the most likely sequence of signal base object prototypes one of the already stored sequence of signal base object prototypes or not and with what probability and reliability?
2. 2. If it is one of the already stored sequence of signal basic object prototypes, which one is it and with what probability and reliability?

For this purpose, on the one hand, a training program can be used to create such a prototype in the form of an entry consisting of a specified sequence of basic signal object prototypes in a signal object database ( 116 ). On the other hand, this can also be done manually using a type-in tool that enables these sequences of signal base object prototypes to be entered using a keyboard.

Using the Viterbi estimator ( 113 ), can be derived from the sequence of hypothesis lists of the distance finder ( 112 ) or the classifier ( 112 ) the most likely of the predefined sequences of signal base object prototypes for a sequence of recognized signal base object prototypes ( 121 ) be determined. This also applies in particular if individual signal base object prototypes due to measurement errors by the distance finder ( 112) or the classifier ( 112 ) were recognized incorrectly. In this respect, a takeover of sequences of signal base object prototype lists from the distance determiner ( 112 ) or classifier ( 112 ), as described above, by the Viterbi estimator ( 113 ) very useful. The result is the most likely detected signal object ( 122 ) or analogous to the emission calculation of the distance estimator described above ( 112 ) or classifier ( 112 ) a signal object hypothesis list.

Finally, we consider the functional components of the signal object recognition machine. This is in the in 7th as Viterbi estimator ( 113 ) entered. This search of the Viterbi estimator ( 113 ) accesses the signal object database ( 116 ) to. The signal object database ( 116 ) is fed on the one hand by a learning tool and on the other hand by a tool in which these sequences of basic signal object prototypes are carried out textual input can be specified The possibility of a download in production is only mentioned here for the sake of completeness.

Basis of the sequence recognition for the temporal sequence of the signal basic object prototypes in the Viterbi estimator ( 119 ) is an example of a hidden Markov model. The model is built up from different states. In the in 9 given example, these states are symbolized by numbered circles. In the example mentioned in 9 the circles are numbered from Z1 to Z6. There are transitions between the states. These transitions are made in the 9 with the letter a and two indices i, j. The first index i denotes the number of the output node and the second index j the number of the destination node. In addition to the transitions between two different nodes, there are also transitions a _ii or a _jj that lead back to the starting node. In addition, there are transitions that make it possible to skip nodes from the sequence, thus in each case a probability of actually observing a k-th observable b _k . This results in sequences of observables that can be observed with precalculable probabilities b _k .

It is important that every hidden Markov model consists of non-observable states q ⁱ . The transition probability a _ij exists between two states q ⁱ and q ^j .

Thus the probability p for the transition from q ⁱ to q ^{j can be} written as: $$p\left(q_n^i|q_{n-1}^i\right)\equiv a_{ij}$$

Here n stands for a discrete point in time. The transition therefore takes place between step n with state q ⁱ and step n + 1 with state q ^j .

The emission distribution b _i (Ge) depends on the state q ⁱ . As already explained, this is the probability of observing the elementary gesture Ge (the observable) when the system (hidden Markov model) is in state q ': $$p\left(Ge|q^i\right)\equiv b_i\left(Ge\right)$$

In order to be able to start the system, the initial states must be defined. This is done using a probability vector π _i . After that can indicate that a state q ⁱ with probability π _i, an initial state is: $$p\left(q^i{}_1\right)\equiv {\pi }_i$$

bestimmt werdenIt is important that a new model must be created for each sequence of signal base object prototypes. In a model M, the observation probability for a temporal observation sequence of signal base object prototypes is intended $$G\overrightarrow{e}=\left(\mathrm{<figure-callout\; id="1"\; label="G\; e"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">G\; </figure-callout>}{e}_{}{}_1,\mathrm{<figure-callout\; id="2"\; label="G\; e"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0028.png"\; state="\{\{state\}\}">G\; </figure-callout>}{e}_{}{}_2\mathrm{,\; ...}G{e}_N\right)$$

to be determined

This corresponds to a temporal sequence of states that cannot be observed directly, which corresponds to the following sequence: $$\overrightarrow{Q}=\left({\mathrm{<figure-callout\; id="1"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">q</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0028.png"\; state="\{\{state\}\}">q</figure-callout>}}_2\mathrm{,\; .....}{q}_N\right)$$

The probability p, depending on the model M, the temporal sequence of states Q and the temporal observation sequence Ge, of observing this sequence of states Q is: $$\begin{array}{l}p\left(G\overrightarrow{e}|\overrightarrow{Q},\mathrm{M.}\right)=p\left(\mathrm{<figure-callout\; id="1"\; label="G\; e"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">G\; </figure-callout>}{e}_{}{}_1,\mathrm{<figure-callout\; id="2"\; label="G\; e"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0028.png"\; state="\{\{state\}\}">G\; </figure-callout>}{e}_{}{}_2\mathrm{,\; ...,}G{e}_N|{\mathrm{<figure-callout\; id="1"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">q</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0028.png"\; state="\{\{state\}\}">q</figure-callout>}}_2\mathrm{,\; .....}{q}_N\right)\\ =p\left(\mathrm{<figure-callout\; id="1"\; label="G\; e"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">G\; </figure-callout>}{e}_{}{}_1|q_1\right)\cdot p\left(\mathrm{<figure-callout\; id="2"\; label="G\; e"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0028.png"\; state="\{\{state\}\}">G\; </figure-callout>}{e}_{}{}_2|q_2\right)\cdot \mathrm{........}p\left(G{e}_N|q_N\right)\\ ={\displaystyle \prod _{n=1}^{N}p\left(G{e}_n|q_n\right)}\\ ={\displaystyle \prod _{n=1}^N{b}_n\left(G{e}_n\right)}\end{array}$$

im Modell M: $$\begin{array}{l}p\left(\overrightarrow{Q}|M\right)=p\left(q_1,q_2\mathrm{,.....}{q}_N|M\right)\\ =p\left(q_1\right)\cdot p\left(q_2|q_1\right)\cdot p\left(q_3|q_1,q_2\right)\cdot \mathrm{.....}p\left(q_N|q_1,q_2\mathrm{,......}{q}_{N-1}\right)\\ =p\left(q_1\right){\displaystyle \prod _{n=2}^{N}p\left(q_n|q_{n-1}\right)}\\ ={\pi }_1{\displaystyle \prod _{n=2}^N{a}_{\left(n-1\right)n}}\end{array}$$

This gives the probability of a sequence of states $\overrightarrow{Q}=\left({\mathrm{<figure-callout\; id="1"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">q</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0028.png"\; state="\{\{state\}\}">q</figure-callout>}}_2\mathrm{,\; .....}{q}_N\right)$

in model M: $$\begin{array}{l}p\left(\overrightarrow{Q}|\mathrm{M.}\right)=p\left({\mathrm{<figure-callout\; id="1"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">q</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0028.png"\; state="\{\{state\}\}">q</figure-callout>}}_2\mathrm{,\; .....}{q}_N|\mathrm{M.}\right)\\ =p\left(q_1\right)\cdot p\left({\mathrm{<figure-callout\; id="2"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0028.png"\; state="\{\{state\}\}">q</figure-callout>}}_2|q_1\right)\cdot p\left({\mathrm{<figure-callout\; id="3"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0036.png"\; state="\{\{state\}\}">q</figure-callout>}}_3|{\mathrm{<figure-callout\; id="1"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">q</figure-callout>}}_1,q_2\right)\cdot \mathrm{.....}p\left(q_N|{\mathrm{<figure-callout\; id="1"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">q</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="q"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0028.png"\; state="\{\{state\}\}">q</figure-callout>}}_2\mathrm{,\; ......}{q}_{N-1}\right)\\ =p\left(q_1\right){\displaystyle \prod _{n=2}^{N}p\left(q_n|q_{n-1}\right)}\\ ={\mathrm{<figure-callout\; id="1"\; label="\pi "\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">\pi </figure-callout>}}_1{\displaystyle \prod _{n=2}^N{a}_{\left(n-1\right)n}}\end{array}$$

Thus, the probability for the detection of a signal object is the same as a sequence of basic signal object prototypes (see also 10 ): $$\begin{array}{l}p\left(G\overrightarrow{e}|{\mathrm{M.}}_j\right)={\displaystyle \sum _{all{Q}_k}p\left(G\overrightarrow{e}|Q_k,{\mathrm{M.}}_j\right)}p\left({\overrightarrow{Q}}_k|{\mathrm{M.}}_j\right)\\ ={\displaystyle \sum _{all{Q}_k}\left({\displaystyle \prod _{n=1}^N{b}_n\left(G{e}_n\right)}\right)\left({\mathrm{<figure-callout\; id="1"\; label="\pi "\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">\pi </figure-callout>}}_1{\displaystyle \prod _{n=2}^N{a}_{\left(n-1\right)n}}\right)}\end{array}$$

The most likely signal object model (signal object) for the observed emission Ge is determined by this summation of the individual probabilities over all possible paths Q _k , which lead to this observed sequence of signal base object prototypes Ge. $$\begin{array}{l}p\left(G\overrightarrow{e}|{\mathrm{M.}}_j\right)={\displaystyle \sum _{allQ}p\left(G\overrightarrow{e}|Q_{k,{\mathrm{M.}}_j}\right)}p\left({\overrightarrow{Q}}_k|{\mathrm{M.}}_j\right)\\ ={\displaystyle \sum _{all{Q}_k}\left({\displaystyle \prod _{n=1}^N{b}_n\left(G{e}_n\right)}\right)}\left({\mathrm{<figure-callout\; id="1"\; label="\pi "\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">\pi </figure-callout>}}_1{\displaystyle \prod _{n=2}^N{a}_{\left(n-1\right)n}}\right)\end{array}$$

The summation over all possible paths Q is not without problems due to the possible computational effort. As a rule, it is therefore canceled very early. It is therefore proposed to only use the most probable path Q _k . This is discussed below.

The calculation is carried out by recursive calculation. The probability a _n (i) at time n to observe the system in state q ⁱ can be calculated as follows: $$\begin{array}{c}{\alpha }_n\left(i\right)=p\left(\mathrm{<figure-callout\; id="1"\; label="G\; e"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">G\; </figure-callout>}{e}_{}{}_1,\mathrm{<figure-callout\; id="2"\; label="G\; e"\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0028.png"\; state="\{\{state\}\}">G\; </figure-callout>}{e}_{}{}_2\mathrm{,\; ......}G{e}_n;q_n=q^i\right)\equiv p\left(G{e}_i^n,q_n^i\right)\\ {\alpha }_{n+1}\left(j\right)=\left[{\displaystyle \sum _{i=1}^{\mathrm{S.}}{\alpha }_n\left(i\right)\cdot a_{ij}}\right]b_j\left(G{e}_{n+1}\right)\end{array}$$

This adds up over all S possible paths that lead into the state q ^{i + 1}

It is assumed here that the overall probability of reaching the state q ⁱ _{n + 1 is} dominated by the best path. Then the sum can be simplified with little error. $${\alpha }_{n+1}*\left(j\right)=\left[\underset{i}{{\displaystyle \text{Max}}}\left({\alpha }_n*\left(i\right)\cdot a_{ij}\right)\right]b_j\left(c_{n+1}\right)$$

By tracing back from the last state one now gets the best path.

The probability of this path is a product. Therefore a logarithmic calculation reduces the problem to a pure summation problem. The probability for the detection of a signal object, which corresponds to the detection of a model M _j, corresponds to the determination of the most likely signal object model for the observed emission X. This is now carried out exclusively via the best possible path Q _best $$p\left(G\overrightarrow{e}|{\mathrm{M.}}_j\right)={\displaystyle \sum _{all{Q}_k}\left({\displaystyle \prod _{n=2}^N{b}_n\left(G{e}_n\right)}\right)\left({\mathrm{<figure-callout\; id="1"\; label="\pi "\; filenames="DE102019009130A1\_0024.png,DE102019009130A1\_0025.png"\; state="\{\{state\}\}">\pi </figure-callout>}}_1{\displaystyle \prod _{n=2}^N{a}_{\left(n-1\right)n}}\right)}$$

This becomes with it $$\begin{array}{l}p\left(G\overrightarrow{e}|{\mathrm{M.}}_j\right)=p\left(G\overrightarrow{e}|Q_{best},{\mathrm{M.}}_j\right)p\left({\overrightarrow{Q}}_{best}|{\mathrm{M.}}_j\right)\\ =\text{exp}\left(\text{ln}\left({\pi }_1\right)+\text{ln}\left(b_1\left(G{e}_1\right)\right)+{\displaystyle \sum _{n=2}^N\text{ln}\left(b_n\left(G{e}_n\right)\right)+\text{ln}\left(a_{\left(n-1\right)n}\right)}\right)\end{array}$$

It is now of particular importance that the prototype database ( 115 ) contains only signal base object prototypes.

The recognized signal objects are transmitted with the recognized parameters instead of the sampled values. This leads to a compression of the data without leaving the signal character.

IT IS PARTICULARLY IMPORTANT THAT THERE IS NO DETECTION OF OBJECTS; THAT ARE IN THE AREA OF THE VEHICLE. MORE THAN THE STRUCTURES WITHIN THE ULTRASOUND ECHO SIGNAL ( 1 ) DETECTED AND USED FOR COMPRESSION.

Only this enables evaluation-free reconstruction of the signal in the control unit after the data has been received.

In contrast to the state of the art, the aim of the present proposal is not to detect and classify objects that are in the vicinity of the vehicle and thereby bring about data compression, but rather the ultrasonic echo signal ( 1 ) to compress and transmit as loss-free as possible by restricting to application-relevant signal shape components.

The ultrasonic sensor then transmits the data compressed in this way, preferably only the codes (symbols) of the prototypes recognized in this way, their amplitude and / or temporal expansion and the time of occurrence (time stamp), to the computer system. This minimizes the EMC load caused by the data transmission between the ultrasonic sensor and the computer system via the data bus and status data from the ultrasonic sensor for system error detection can be transmitted to the computer system via the data bus between the ultrasonic sensor and the computer system, which improves the latency period. When working out the proposal, it was recognized that the transmission of data via the data bus must be prioritized. However, as is known from the prior art, this prioritization does not affect the prioritization with respect to other bus users. Rather, the data connection between the sensor and the control computer of the vehicle is usually a point-to-point connection. Rather, the prioritization here is to be understood to mean which date of the data determined by the sensor system must be transmitted first to the control device. Messages of safety-critical errors of the sensor, that is to say here, for example, of the ultrasonic sensor, to the computer system have the highest priority, since it is very likely that they impair the validity of the measurement data of the ultrasonic sensor. This data is sent from the sensor to the computer system. Inquiries from the computer system to carry out safety-relevant self-tests have the second highest priority. Such commands are sent from the computer system to the sensor. The third highest priority is given to the data of the Ultrasonic sensor itself, since the latency time must not be increased. All other data have a lower priority for transmission over the data bus.

It is particularly advantageous if the method for transmitting sensor data, in particular an ultrasonic sensor, from a sensor to a computer system, in particular in a vehicle, includes the transmission of an ultrasonic burst with a beginning ( 57 ) and end ( 56 ) of sending the ultrasonic burst and comprising receiving an ultrasonic signal and forming a received signal for a reception time (T _E ) at least from the end ( 56 ) the transmission of the ultrasound burst and the transmission of the compressed data via a data bus, in particular a single-wire data bus, to the computer system is designed so that the transmission ( 54 ) the data from the sensor to the computer system with a start command ( 53 ) from the computer system to the ultrasonic sensor via the data bus and before the end ( 56 ) the transmission of the ultrasonic burst begins or after a start command ( 53 ) from the computer system to the sensor via the data bus and before the start ( 57 ) of the transmission of the ultrasonic burst begins. The transfer ( 54 ) then takes place after the start command ( 53 ) periodically and continuously until the end of the data transmission ( 58 ). This end of data transmission ( 58 ) is then after the end of the reception time (T _E ).

A further variant of the proposed method thus provides, as the first step of data compression, the formation of a feature vector signal (stream of feature vectors with n feature vector values and n as the dimension of the feature vector) from the received signal. Such a feature vector signal can comprise several analog and digital data signals. It therefore represents a time sequence of more or less complex data / signal structures. In the simplest case, it can be understood as a vector signal consisting of several partial signals.

For example, it can be useful to form a first and / or higher time derivative of the received signal or the single or multiple integral of the received signal, which are then partial signals within the feature vector signal.

An envelope curve signal of the received signal (ultrasonic echo signal) can also be formed, which is then a partial signal within the feature vector signal ( 138 ) can be.

Furthermore, it can be useful to fold the received signal with the transmitted ultrasonic signal and thus to form a correlation signal, which can then be a partial signal within the feature vector signal. On the one hand, the signal that was used to control the driver for the transmitter can be used as the emitted ultrasonic signal, or on the other hand, for example, a signal that was measured at the transmitter and thus better corresponds to the actually emitted sound wave.

Finally, it can be useful to use an optimal filter (English matched filter) to detect the occurrence of predetermined signal objects and to form an optimal filter signal for the respective signal object of some of the predetermined signal objects. A matched filter is a filter that optimizes the signal-to-noise ratio (SNR). The predefined signal objects should be recognized in the disturbed ultrasonic reception signal. The terms correlation filter, signal matched filter (SAF) or just matched filter are also often found in the literature. The optimal filter is used to optimally determine the presence (detection) of the amplitude and / or the position of a known signal shape, the predetermined signal object, in the presence of interference (parameter estimation). These disturbances can be signals from other ultrasonic transmitters and / or ground echoes, for example.

The optimal filter output signals are then preferably partial signals within the feature vector signal.

Certain events can be signaled in separate partial signals of the feature vector signal. These events are basic signal objects in the sense of this disclosure. Basic signal objects therefore do not include signal shapes, such as square-wave pulses or wavelets or wave trains, but rather distinctive points in the course of the received signal and / or in the course of signals derived therefrom, such as an envelope signal (ultrasonic echo signal), which is generated, for example, by filtering the received signal can be obtained.

Another signal, which can be a partial signal of the feature vector signal, can detect, for example, whether the envelope curve of the received signal, the ultrasonic echo signal, is a predetermined third Threshold crosses. It is therefore a signal that signals the presence of a basic signal object within the received signal and thus the feature vector signal.

Another signal, which can be a partial signal of the feature vector signal, can, for example, detect whether the envelope curve of the received signal, the ultrasonic echo signal, crosses a predetermined fourth threshold value, which can be identical to the third threshold value, in ascending order. It is therefore a signal that signals the presence of a basic signal object within the received signal and thus the feature vector signal.

Another signal, which can be a partial signal of the feature vector signal, can, for example, detect whether the envelope curve of the received signal, the ultrasonic echo signal, descends a predetermined fifth threshold value, which can be identical to the third or fourth threshold value . It is therefore a signal that signals the presence of a basic signal object within the received signal and thus the feature vector signal.

Another signal, which can be a partial signal of the feature vector signal, can, for example, detect whether the envelope curve of the received signal, the ultrasonic echo signal, is a maximum above a sixth threshold value, which is identical to the aforementioned third to fifth threshold values can, has. It is therefore a signal that signals the presence of a basic signal object within the received signal and thus the feature vector signal.

Another signal, which can be a partial signal of the feature vector signal, can for example detect whether the envelope curve of the received signal, the ultrasonic echo signal, is a minimum above a seventh threshold value, which is identical to the aforementioned third to sixth threshold values can, has. It is therefore a signal that signals the presence of a basic signal object within the received signal and thus the feature vector signal.

It is preferably evaluated whether the at least one preceding maximum of the ultrasonic echo signal is at a minimum distance from the minimum in order to avoid the detection of noise. Other filters are conceivable at this point. It can also be checked whether the time interval between this minimum and a preceding maximum is greater than a first time minimum interval. The fulfillment of these conditions sets a flag or signal, which in turn is preferably a partial signal of the feature vector signal.

Likewise, it should be checked in an analogous manner whether the time and amplitude spacings between the other signal objects meet certain plausibility requirements, such as minimum time intervals and / or minimum spacing in amplitude. Further, also analog, binary or digital partial signals can also be derived from these tests, which thus further increase the dimensionality of the feature vector signal.

Possibly. the feature vector signal can be transformed into a significant feature vector signal e.g. can still be transformed by a linear mapping or a matrix polynomial of a higher order. In practice, however, it has been shown that this is not yet necessary, at least for today's requirements.

According to the proposed method, signal objects are recognized and classified into recognized signal object classes within the received signal on the basis of the feature vector signal or the significant feature vector signal.

If, for example, the amplitude of the output signal of an optimal filter, and thus a partial signal of the feature vector signal, is above a possibly optimal filter-specific eighth threshold value, the signal object, for whose detection the optimal filter is designed, can be considered recognized. Other parameters are preferably also taken into account here. If, for example, an ultrasound burst with an increasing frequency was sent during the burst (called chirp-up), an echo is also expected that has this modulation property. If the signal shape of the ultrasonic echo signal, for example a triangular signal shape of the ultrasonic echo signal, coincides locally with an expected signal shape in terms of time, but not the modulation property, then it is not an echo from the transmitter, but a Interference signal that can come from other ultrasonic transmitters or from overreaches. In this respect, the system can then differentiate between self-echoes and external echoes, whereby one and the same signal form is assigned to two different signal objects, namely internal echoes and external echoes. The Transmission of the intrinsic echoes via the data bus from the sensor to the computer system is preferably prioritized over the transmission of external echoes, since the former are generally safety-relevant and the second are generally not safety-relevant.

Typically, during the detection, at least one assigned signal object parameter is assigned to each recognized signal object or is determined for this signal object. This is preferably a time stamp that indicates when the object was received. In this case, the time stamp can relate, for example, to the temporal beginning of the signal object in the received signal or the temporal end or the temporal position of the temporal focus of the signal object etc. Other signal object parameters, such as amplitude, stretching, etc. are also conceivable. In a variant of the proposed method, at least one of the assigned signal object parameters is thus transmitted with a symbol for the at least one recognized signal object class. The signal object parameter is preferably a time value as a time stamp and indicates a time position that is suitable for inferring the time since the transmission of a previous ultrasonic burst. A determined distance of an object is preferably determined therefrom later as a function of a time value determined and transmitted in this way.

Finally, the prioritized transmission of the recognized signal object classes in the form of assigned symbols with time stamps, preferably in each case together with the assigned signal object parameters, follows. The transfer can also take place in more complex data structures (English: records). For example, it is conceivable to first transmit the times of the recognized safety-relevant signal objects (e.g. identified obstacles) and then to transmit the recognized signal object classes of the safety-relevant signal objects. This further reduces the latency.

The proposed method comprises, at least in one variant, the determination of a chirp value as an assigned signal object parameter, which indicates whether the detected signal object is an echo of an ultrasonic transmission burst with chirp-up or a chirp-down or a No- Chirp properties. Chirp-Up means that the frequency within the received signal object increases in the received signal. Chirp-Down means that the frequency falls within the received signal object in the received signal. No-chirp means that the frequency within the received signal object remains essentially the same in the received signal. At this point, reference is made to the as yet unpublished German patent applications DE 10 2017 100 837.3 and DE 10 2017 100 835.7 that are fully part of this disclosure.

In a variant of the method, a confidence signal is thus also generated by forming the correlation, e.g. by forming a time-continuous or time-discrete correlation integral between the received signal or instead of the received signal with a signal derived from the received signal on the one hand and a reference signal, for example the ultrasonic transmission signal or another expected wavelet, on the other. The confidence signal is then typically a partial signal of the feature vector signal, that is to say a component of the feature vector, which consists of a sequence of vectorial sampled values (feature vector values).

In a variant of the method, a phase signal is also formed on this basis, which indicates the phase shift, for example of the received signal or a signal formed therefrom (e.g. the confidence signal) in relation to a reference signal, for example the ultrasonic transmission signal and / or another reference signal. The phase signal is then typically also a partial signal of the feature vector signal, that is to say a component of the feature vector, which consists of a sequence of vectorial sampled values.

In a similar way, in a further variant of the proposed method, a phase confidence signal can be formed by forming the correlation between the phase signal or a signal derived therefrom and a reference signal and used as a partial signal of the feature vector signal. The phase confidence signal is then typically also a partial signal of the feature vector signal, that is to say a component of the feature vector, which consists of a sequence of vectorial sampled values.

When evaluating the feature vector signal, it then makes sense to compare the phase confidence signal with one or more threshold values in order to generate a discretized phase confidence signal, which in turn becomes a partial signal of the feature vector signal can.

The evaluation of the feature vector signal and / or the significant feature vector signal can take place in a variant of the proposed method in such a way that one or more distance values between the feature vector signal and one or more signal object prototype values are formed for recognizable signal object classes . Such a distance value can be Boolean, binary, discrete, digital or analog. All distance values are preferably linked to one another in a non-linear function. In the case of an expected chirp-up echo in triangular form, a received chirp-down echo in triangular form can be discarded. For the purposes of this disclosure, this discarding is a non-linear process.

Conversely, the triangle in the received signal can be different. First and foremost, this affects the amplitude of the triangle in the received signal. If the amplitude in the received signal is sufficient, then, for example, the optimal filter assigned to this triangular signal delivers a signal above a predetermined ninth threshold value. In this case, for example, this signal object class (triangular signal) can be assigned a signal object recognized at this point in time when it is exceeded. In this case, the distance value between the feature vector signal and the prototype (here the ninth threshold value) falls below one or more predetermined, binary, digital or analog distance values (here 0 = intersection).

In a further variant of the method, at least one signal object class is wavelets, which are estimated and thus detected by estimation devices (e.g. optimal filter) and / or estimation methods (e.g. estimation programs that run in a digital signal processor). The term wavelet refers to functions that can be used as the basis for a continuous or a discrete wavelet transformation. The word “wavelet” is a new creation from the French “ondelette”, which means “small wave” and is partly translated literally (“onde” → “wave”), partly phonetically (“-lette” → “-let”) into English has been. The term "wavelet" was coined in the 1980s in geophysics (Jean Morlet, Alex Grossmann) for functions that generalize the short-term Fourier transform, but has only been used since the late 1980s in the meanings customary today. In the 1990s, a real wavelet boom arose, triggered by the discovery of compact, continuous (up to any order of differentiability) and orthogonal wavelets by Ingrid Daubechies (1988) and the development of the fast wavelet transformation (FWT) algorithm. with the help of MultiResolution Analysis (MRA) by Stephane Mallat and Yves Meyer (1989).

In contrast to the sine and cosine functions of the Fourier transformation, the most commonly used wavelets not only have locality in the frequency spectrum, but also in the time domain. "Locality" is to be understood in the sense of small spread. The probability density is the normalized square of the absolute value of the function under consideration or of its Fourier transform. The product of the two variances is always greater than a constant, analogous to Heisenberg's uncertainty principle. Out of this restriction, the Paley-Wiener theory (Raymond Paley, Norbert Wiener), a forerunner of the discrete wavelet transformation, and the Calderón-Zygmund theory (Alberto Calderón, Antoni Zygmund), which of the continuous wavelet Transformation corresponds.

The integral of a wavelet function is always 0 in technical usage, so the wavelet function usually takes the form of outwardly tapering (decreasing) waves (ie "waves" = ondelettes = wavelets). For the purposes of this disclosure, however, wavelets should also be permissible which have an integral other than zero. The rectangle and triangle wavelets described below are mentioned here as examples. This further interpretation of the term "wavelet" is widespread in the American-speaking area and is therefore known. This further interpretation should also apply here.

Important examples of wavelets with a 0 integral are the Haar wavelet (Alfred Haar 1909), the Daubechies wavelets named after Ingrid Daubechies (around 1990), the Coiflet wavelets that she also constructed, and the theoretically important Meyer wavelet (Yves Meyer, around 1988).

There are wavelets for spaces of any dimension, mostly a tensor product of a one-dimensional wavelet base is used. Due to the fractal nature of the two-scale equation in the MRA, most wavelets are complex in shape, most are not closed in shape. This is of particular importance because the previously mentioned feature vector signal is multi-dimensional and therefore allows the use of multi-dimensional wavelets for signal object detection.

A special variant of the proposed method is therefore the use of multidimensional wavelets with more than two dimensions for signal object detection. In particular, the use of appropriate optimal filters for recognizing such wavelets with more than two dimensions is proposed in order to supplement the feature vector signal with further partial signals suitable for recognition if necessary.

A particularly suitable wavelet is, for example, a triangle wavelet. This is characterized by a start time of the triangle wavelet, a temporally essentially linear increase in the wavelet amplitude that follows the start time of the triangle wavelet up to a maximum of the amplitude of the triangle wavelet and one the maximum of the triangle wavelet and one temporally subsequent, temporally essentially linear decrease in the wavelet amplitude up to one end of the triangle wavelet.

Another particularly suitable wavelet is a rectangular wavelet, which in the context of this disclosure also includes trapezoidal wavelets. A rectangular wavelet is characterized by a starting time of the rectangular wavelet, which is followed by an increase in the wavelet amplitude of the rectangular wavelet with a first temporal steepness of the rectangular wavelet up to a first plateau time of the rectangular wavelet. The first plateau point in time of the rectangular wavelet is followed by a persistence of the wavelet amplitude with a second steepness of the wavelet amplitude up to a second plateau point in time of the rectangular wavelet. The second plateau point in time of the rectangular wavelet is followed by a drop with a third temporal steepness up to the temporal end of the rectangular wavelet. The amount of the second temporal steepness is less than 10% of the amount of the first temporal steepness and less than 10% of the amount of the third temporal steepness.

Instead of the wavelets described above, it is also possible to use other two-dimensional wavelets, such as a half-sine wave wavelet, which also has an integral other than 0.

It is proposed that, when using wavelets, the time shift of the relevant wavelet of the recognized signal object is used as a signal object parameter. For example, this shift can be determined by correlation. It is further proposed that when using wavelets, the time at which the level of the output of an optimal filter suitable for detecting the relevant wavelet is exceeded is preferably used via a predefined tenth threshold value for this signal object or this wavelet becomes. The ultrasonic echo signal (the envelope curve) of the received signal and / or a phase signal and / or a confidence signal etc. are preferably evaluated.

Another possible signal object parameter that can be determined is a temporal compression or expansion of the relevant wavelet of the recognized signal object. An amplitude of the wavelet of the recognized signal object can also be determined.

When working out the proposal for the method disclosed here, it was recognized that it is advantageous to first transmit the data of the detected signal objects of the very quickly arriving echoes from the sensor to the computer system and only then to transfer the subsequent data of the later detected signal objects. Preferably, at least the recognized signal object class and a time stamp are always transmitted, which should preferably indicate when the signal object has arrived at the sensor again. Within the scope of the recognition process, scores can be assigned to the various signal objects that come into consideration for a section of the received signal, which indicate which probability is assigned to the presence of this signal object in accordance with the estimation algorithm used. Such a score is binary in the simplest case. However, it is preferably a complex, real or integer number. It can be the determined distance, for example. If several signal objects have a high score value, it makes sense in some cases to also transmit the data of recognized signal objects with lower score values. In order to enable the computer system to handle it correctly, not only the date (symbol) of the detected signal object and the time stamp for the respective signal object should be transmitted in this case, but also the determined score value. Instead of only transmitting the date (symbol) of the detected signal object and the time stamp for the signal object corresponding to this symbol, the date (symbol) of the signal object with the second smallest distance and its time stamp for the signal object corresponding to this second most likely symbol can also be transmitted. In this case, a list of hypotheses consisting of two recognized signal objects and their positions over time as well as additionally assigned score values are transmitted to the computer system. Of course, a list of hypotheses consisting of more than two symbols for more than two recognized signal objects and their temporal positions as well as additionally assigned score values can also be transmitted to the computer system.

The data of the recognized signal object class and the assigned data, such as time stamps and score values of the respective recognized signal object classes, that is to say the assigned signal object parameters, are preferably transmitted according to the FIFO principle. This ensures that the data of the reflections of the closest objects are always transmitted first, so that the safety-critical case of the vehicle colliding with an obstacle is prioritized according to probability.

In addition to the transmission of measurement data, error states of the sensor can also be transmitted. This can also take place during a reception time (T _E ) when the sensor uses self-test devices to determine that there is a defect and the previously transmitted data could potentially be incorrect. This ensures that the computer system can gain knowledge of a change in the evaluation of the measurement data at the earliest possible point in time and can discard it or treat it differently. This is of particular importance for emergency braking systems, since emergency braking is a safety-critical intervention that may only be initiated if the underlying data have a corresponding trustworthiness. On the other hand, the transmission of the measurement data, that is to say, for example, the date of the recognized signal object class and / or the transmission of the one assigned signal object parameter, is therefore postponed and thus given a lower priority. It is of course conceivable that the transmission could be aborted if an error occurs in the sensor. In some cases, however, it can happen that an error appears possible but is not certain. In such cases, it may be advisable to continue a transmission. The transmission of safety-critical errors of the sensor is therefore preferably carried out with a higher priority.

In addition to the already described wavelets with an integration value of 0 and the signal sections also referred to here as wavelet with an integration value other than 0, certain times in the course of the received signal can also be understood as a basic signal object in the sense of this disclosure, which can be used for data compression and instead of samples of the received signal can be transmitted. This subset of the set of possible basic signal objects is referred to below as signal times. In the context of this disclosure, the signal times are therefore a special form of the basic signal objects.

A first possible signal point in time and thus a basic signal object is a crossing of the amplitude of the ultrasonic echo signal ( 1 ) with the amplitude of an eleventh threshold value signal (SW) in an ascending direction.

A second possible signal point in time and thus a basic signal object is a crossing of the amplitude of the ultrasonic echo signal ( 1 ) with the amplitude of a twelfth threshold value signal (SW) in descending direction.

A third possible signal point in time and thus a basic signal object is a maximum of the amplitude of the ultrasonic echo signal ( 1 ) above the amplitude of a thirteenth threshold value signal (SW).

A fourth possible signal point in time and thus a basic signal object is a minimum of the amplitude of the ultrasonic echo signal ( 1 ) above the amplitude of a fourteenth threshold value signal (SW).

Possibly. it can be useful to use threshold value signals (SW) specific to the signal time point for these four exemplary types of signal times and other types of signal times.

1. 1. das Auftreten eines ersten möglichen Signalzeitpunkts mit einem Kreuzen der Amplitude des Ultraschall-Echo-Signals (1) mit der Amplitude eines Schwellwertsignals (SW) in aufsteigender Richtung und daran zeitlich anschließend
2. 2. das Auftreten eines dritten möglichen Signalzeitpunkts mit einem Maximum der Amplitude des Ultraschall-Echo-Signals (1) oberhalb der Amplitude eines Schwellwertsignals (SW) und daran zeitlich anschließend
3. 3. das Auftreten eines zweiten möglichen Signalzeitpunkts mit einem Kreuzen der Amplitude des Ultraschall-Echo-Signals (1) mit der Amplitude eines Schwellwertsignals (SW) in absteigender Richtung

The time sequence of basic signal objects is typically not arbitrary. This is exploited in this disclosure since it is preferred not to transmit the basic signal objects, which are of a simpler nature, but rather recognized patterns of sequences of these basic signal objects, which then represent the actual signal objects. For example, if a triangular wavelet is included in the ultrasonic echo signal ( 1 ) sufficient amplitude is expected, in addition to a corresponding minimum level at the output of an optimal filter suitable for the detection of such a triangular wavelet

1. 1. the occurrence of a first possible signal instant with a crossing of the amplitude of the ultrasonic echo signal ( 1 ) with the amplitude of a threshold value signal (SW) in an ascending direction and subsequently in time
2. 2. the occurrence of a third possible signal instant with a maximum of the amplitude of the ultrasonic echo signal ( 1 ) above the amplitude of a threshold value signal (SW) and subsequent to it in time
3. 3. the occurrence of a second possible signal instant with a crossing of the amplitude of the ultrasonic echo signal ( 1 ) with the amplitude of a threshold value signal (SW) in descending direction

can be expected in temporal correlation with the exceeding of said minimum level at the output of said optimal filter. In this example, the exemplary signal object of a triangular wavelet consists of the predefined sequence of three basic signal objects, which are recognized, replaced by a symbol and preferably transmitted as this symbol together with its time of occurrence, the time stamp. This exceeding of said minimum level at the output of said optimal filter is Incidentally, another example of a fifth possible signal time and thus another possible basic signal object.

The resulting grouping and temporal sequence of recognized basic signal objects can itself be recognized, for example by a Viterbi decoder, as a predefined, expected grouping or temporal sequence of basic signal objects and can thus itself represent a basic signal object. Thus, a sixth possible signal instant and thus a basic signal object is the occurrence of such a predefined grouping and / or temporal sequence of other basic signal objects.

If such a grouping of basic signal objects or a temporal sequence of such signal object classes is recognized in the form of a signal object, the symbol of this recognized, summarizing signal object class and at least the one assigned signal object parameter are transmitted, preferably instead of transmitting the individual basic signal objects, as this saves considerable data bus capacity. However, there may be cases in which both are transmitted. In this case, the date (symbol) of the signal object class of a signal object is transmitted, which is a predefined time sequence and / or grouping of other basic signal objects. In order to achieve compression, it is advantageous if at least one signal object class (symbol) at least one of these other basic signal objects is not transmitted.

A temporal grouping of basic signal objects is present in particular when the time interval between these basic signal objects does not exceed a predefined interval. In the example mentioned above, the runtime of the signal in the optimal filter should be considered. Typically, the matched filter should be slower than the comparators. Therefore, the change in the output signal of the optimal filter should have a fixed temporal connection with the temporal occurrence of the relevant signal times.

A method for transmitting sensor data, in particular an ultrasonic sensor, from a sensor to a computer system, in particular in a vehicle, is proposed here, which consists of transmitting an ultrasonic burst and receiving an ultrasonic signal and forming a discrete-time received signal a sequence of samples begins. A date (time stamp) is assigned to each sample value. At least two parameter signals are then determined, each relating to the presence of a basic signal object assigned to the respective parameter signal with the aid of at least one suitable filter (e.g. an optimal filter) from the sequence of sampled values of the received signal. The resulting parameter signals (feature vector signals) are also designed as a time-discrete sequence of respective parameter signal values (feature vector values), each of which is correlated with a date (time stamp). Thus, each parameter signal value (feature vector value) is preferably assigned exactly one time date (time stamp). These parameter signals together are referred to below as the feature vector signal. The feature vector signal is thus designed as a time-discrete sequence of feature vector signal values, each with n parameter signal values, which consist of the parameter signal values and further parameter signal values each with the same date (time stamp). Here, n is the dimensionality of the individual feature vector signal values, which are preferably the same from one feature vector value to the next feature vector value. Each feature vector signal value formed in this way is assigned this respective date (time stamp). This is followed by the evaluation of the time course of the feature vector signal in the resulting n-dimensional phase space as well as the conclusion that a signal object has been recognized by determining an evaluation value (e.g. the distance). As explained above, a signal object consists of a chronological sequence of basic signal objects. A symbol is typically assigned to the signal object in a predefined manner. Figuratively speaking, it is checked here whether the point to which the n-dimensional feature vector signal points in the n-dimensional phase space, on its way through the n-dimensional phase space in a predetermined time sequence, is at predetermined points in this n-dimensional phase space closer than a predetermined maximum distance. The feature vector signal therefore has a time profile. An evaluation value (e.g. a distance) is then calculated which can, for example, reflect the probability of the existence of a specific sequence. This evaluation value, which is again assigned a time date (time stamp), is then compared with a threshold value vector with formation of a Boolean result that can have a first and a second value. If this Boolean result has the first value for this time date (time stamp), the symbol of the signal object and the time date (time stamp) assigned to this symbol are transmitted from the sensor to the computer system. Possibly. further parameters can be transferred depending on the detected signal object.

The reconstruction of a reconstructed ultrasonic echo signal model is particularly advantageous ( 610 ) from the detected signal objects ( 122 ). This ultrasonic echo signal model ( 610 ) becomes from the Adding linearly superimposed signal curve models of the individual detected signal objects are generated. It is then followed by the ultrasonic echo signal ( 1 ) deducted. Thereby a residual signal ( 660 ) educated. As a result, signal objects similar to the selected signal object recognized with a higher probability are better suppressed. The weaker signal objects appear in the residual signal ( 660 ) stand out better and can be better recognized. (See also 18th ). A subtraction of the ultrasonic echo signal model reconstructed from the detected signal objects is therefore also preferred in a proposed method ( 610 ) from the ultrasonic echo signal ( 1 ) to form a residual signal ( 660 ) intended. The residual signal formed in this way ( 660 ) is then used again for the formation of the feature vector signal ( 138 ) is used and the next signal object is recognized with the next higher probability. Since the signal object recognized first was removed from the input signal according to its weighting, it can no longer influence this recognition. This form of recognition therefore provides a better result.

However, this detection method is usually slower. Therefore, it makes sense to first carry out a direct first object recognition without subtraction while the measurement is still in progress and then, after recording all the sampled values of an ultrasonic echo response from the surroundings of the vehicle, after sending an ultrasonic pulse or an ultrasonic burst, repeat pattern recognition with subtraction of the Ultrasonic echo signal model ( 610 ), which takes longer but is more precise.

This reducing classification of the ultrasonic echo signal is preferred ( 1 ) in signal objects with the help of the ultrasonic echo signal model ( 610 ) ends when the amounts of the samples of the residual signal ( 660 ) are below the amounts of a specified threshold value curve.

The data transmission in the vehicle is particularly preferably carried out via a serial bidirectional single-wire data bus. The electrical return line is preferably ensured by the body of the vehicle. The sensor data are preferably sent to the computer system in a current-modulated manner. The data for controlling the sensor are sent to the sensor by the computer system, preferably voltage-modulated. It was recognized that the use of a PSI5 data bus and / or a DSI3 data bus is particularly suitable for data transmission. It was also recognized that it is particularly advantageous to transfer the data to the computer system at a transfer rate of> 200 kBit / s and from the computer system to the at least one sensor with a transfer rate> 10 kBit / s, preferably 20 kBit / s . Furthermore, it was recognized that the transmission of data from the sensor to the computer system should be modulated with a transmission current on the data bus whose current intensity should be less than 50 mA, preferably less than 5 mA, preferably less than 2.5 mA. These buses must be adapted accordingly for these operating values. The basic principle remains, however.

• • Optimalfilter,
• • Komparatoren,
• • Schwellwertsignalerzeugungsvorrichtungen zur Erzeugung eines oder mehrerer Schwellwertsignale (SW),
• • Differenzierer zur Bildung von Ableitungen,
• • Integrierer zur Bildung integrierter Signale,
• • sonstige Filter,
• • Hüllkurvenformer zur Erzeugung eines Hüllkurvensignals (Ultraschall-Echo-Signal) aus dem Empfangssignal,
• • Korrelationsfilter zum Vergleich des Empfangssignals oder von daraus abgeleiteten Signalen mit Referenzsignalen.

A computer system with a data interface to said data bus, preferably said single-wire data bus, which supports the decompression of the data compressed in this way is required for carrying out the methods described above. As a rule, however, the computer system will not undertake a complete decompression, but only evaluate the time stamp and the recognized signal object type, for example. The sensor, which is required to carry out one of the methods described above, has at least one transmitter and at least one receiver for generating a received signal, which can also be present in combination as one or more transducers. Furthermore, it has at least devices for processing and compressing the received signal and a data interface for transmitting the data via the data bus, preferably the said single-wire data bus, to the computer system. For the compression, the device for compression preferably has at least one of the following sub-devices:

• • optimal filter,
• • comparators,
• • Threshold signal generating devices for generating one or more threshold signals (SW),
• • Differentiators for the formation of derivations,
• • integrator for the formation of integrated signals,
• • other filters,
• • Envelope shaper for generating an envelope signal (ultrasonic echo signal) from the received signal,
• • Correlation filter for comparing the received signal or signals derived from it with reference signals.

• Es beginnt mit dem Aussenden eines Ultraschall-Bursts. Es folgt das Empfangen eines Ultraschallsignals, also typischerweise eine Reflektion und das Bilden eines zeitdiskreten Empfangssignals bestehend aus einer zeitlichen Folge von Abtastwerten. Dabei ist jedem Abtastwert ein zeitliches Datum (Zeitstemple) zugeordnet. Dieses gibt den Zeitpunkt der Abtastung typischerweise wieder. Auf Basis dieses Datenstroms erfolgt die Bestimmung eines ersten Parametersignals einer ersten Eigenschaft mit Hilfe eines ersten Filters aus der Folge von Abtastwerten des Empfangssignals. Bevorzugt wird dabei das Parametersignal wieder als zeitdiskrete Folge von Parametersignalwerten ausgebildet. Jedem Parametersignalwert wird wieder genau ein zeitliches Datum (Zeitstempel) zugeordnet. Bevorzugt entspricht dieses Datum dem jüngsten zeitlichen Datum eines Abtastwertes, der zur Bildung dieses jeweiligen Parametersignalwertes benutzt wurde. Zeitlich parallel dazu erfolgt bevorzugt die Bestimmung mindestens eines weiteren Parametersignals einer diesem weiteren Parametersignal zugeordneten Eigenschaft mit Hilfe eines diesem weiteren Parametersignal zugeordneten weiteren Filters aus der Folge von Abtastwerten des Empfangssignals, wobei die weiteren Parametersignale jeweils wieder als zeitdiskrete Folgen von weiteren Parametersignalwerten ausgebildet sind. Auch hier wird jedem weiteren Parametersignalwert jeweils das gleiche zeitliche Datum (Zeitstempel) wie dem entsprechenden Parametersignalwert zugeordnet.

In a particularly simple form, the proposed method for transmitting sensor data, in particular an ultrasonic sensor, from a sensor to a computer system, in particular in a vehicle, is carried out as follows:

• It starts by sending out an ultrasonic burst. This is followed by the reception of an ultrasonic signal, that is to say typically a reflection and the formation of a time-discrete received signal consisting of a temporal sequence of sample values. A date (time stamp) is assigned to each sample value. This typically reflects the time of the scan. On the basis of this data stream, a first parameter signal of a first property is determined with the aid of a first filter from the sequence of sampled values of the received signal. The parameter signal is preferably again designed as a time-discrete sequence of parameter signal values. Exactly one date (time stamp) is assigned to each parameter signal value. This datum preferably corresponds to the most recent temporal datum of a sample that was used to form this respective parameter signal value. At least one further parameter signal of a property assigned to this further parameter signal is preferably determined from the sequence of sampled values of the received signal with the aid of a further filter assigned to this further parameter signal, the further parameter signals each again being in the form of discrete-time sequences of further parameter signal values. Here, too, the same date (time stamp) is assigned to each further parameter signal value as the corresponding parameter signal value.

The first parameter signal and the further parameter signals are referred to together below as the feature vector signal. This feature vector signal (or also parameter vector signal) thus represents a time-discrete sequence of feature vector signal values, which consist of the parameter signal values and further parameter signal values each with the same date (time stamp). This respective time date (time stamp) can thus be assigned to each feature vector signal value (= parameter signal value) formed in this way.

The comparison of the feature vector signal values of a date (time stamp) with a threshold value vector, which is preferably a prototype vector, is then preferably carried out quasi-continuously, with the formation of a Boolean result that can have a first and a second value. For example, it is conceivable to compare the amount of the current feature vector signal value, which for example represents a first component of a feature vector signal value, with a fifteenth threshold value, which represents a first component of the threshold value vector, and to set the Boolean result to a first value if the The amount of the feature vector signal value is smaller than this fifteenth threshold value, and to a second value if this is not the case. If the Boolean result has a first value, it is then also conceivable to combine the amount of the additional feature vector signal value, which represents, for example, a further component of this feature vector signal, with a further threshold value, which represents a further component of the threshold value vector to compare and to leave the Boolean result at the first value if the amount of the further feature vector signal value is smaller than this further threshold value, and to set the Boolean result to the second value if this is not the case. In this way all further feature vector signal values can be checked. Of course, other classifiers are also conceivable. A comparison with several different threshold value vectors is also possible. These threshold value vectors thus represent the prototypes of given signal forms. They come from the library mentioned. A symbol is again preferably assigned to each threshold value vector.

The last step in this case is the transfer of the symbol and possibly also the feature vector signal values and the date (time stamp) assigned to this symbol or feature vector signal value from the sensor to the computer system if the Boolean result has the first value for this time date (time stamp).

This means that all other data is no longer transmitted. In addition, the multi-dimensional evaluation avoids disturbances.

On this basis, a sensor system is therefore proposed with at least one computer system that is capable of performing one of the previously presented methods and with at least two Sensors that are also able to carry out one of the previously presented methods, so that these at least two sensors can communicate with the computer system through signal object recognition and are also able to transmit extraneous echoes in compact form and to make this information available to the computer system. The sensor system is accordingly typically provided so that the data transmission between the at least two sensors and the computer system takes place or can take place in accordance with the method described above. Within the at least two sensors of the sensor system, one ultrasonic reception signal, that is to say at least two ultrasonic reception signals, is typically compressed by means of one of the previously proposed methods and transmitted to the computer system. In this case, the at least two ultrasonic reception signals are reconstructed into reconstructed ultrasonic reception signals within the computer system. The computer system then uses reconstructed ultrasonic reception signals to recognize objects in the vicinity of the sensors. In contrast to the prior art, the sensors do not carry out this object recognition.

The computer system preferably also uses the reconstructed ultrasonic reception signals and additional signals from further sensors, in particular the signals from radar sensors, to recognize objects in the vicinity of the sensors.

As a last step, the computer system preferably creates an environment map for the sensors or a device of which the sensors are a part based on the detected objects.

The proposed ultrasonic sensor system, as exemplified in 7th and 12 is shown, thus preferably comprises an ultrasonic transducer ( 100 ), a feature extractor ( 111 ) and with an estimator ( 150 , 151 ) or classifier and a physical interface ( 101 ). The ultrasonic transducer ( 100 ) is preferably provided and / or set up to receive an acoustic ultrasonic wave signal and, depending on this, an ultrasonic transducer signal ( 102 ) to build. The feature vector extractor ( 111 ) is provided and / or set up to use the ultrasonic transducer signal ( 102 ) a feature vector signal ( 138 ) to build. The ultrasonic sensor system is preferably set up and provided for using the estimator ( 151 , 150 ) Signal objects in the ultrasonic echo signal ( 1 ) to recognize and to classify in signal object classes, wherein a signal object class can also include only one signal object. Preference is given to each signal object recognized and classified in this way ( 122 ) at least one assigned signal object parameter and a symbol corresponding to the signal object class assigned to this signal object or for each signal object recognized and classified in this way ( 122 ) at least one assigned signal object parameter and a symbol for this signal object are determined. At least the symbol of a recognized signal object class ( 122 ) and at least the one assigned signal object parameter of this recognized signal object class ( 122 ) are transmitted to a higher-level computer system via a data bus.

The estimator preferably has ( 150 ) a distance finder ( 112 ) and a prototype database ( 115 ) on. Likewise, the estimator ( 150 ) a Viterbi estimator and a signal object database. The estimator ( 151 ) use a neural network model.

• • Warten auf ein Ultraschallempfangssignal, das vom Grundrauschen verschieden ist,
• • Wechsel in einen Zustand „kein Signalobjektgrundprototyp“ falls das Ultraschallempfangssignal von Grundrauschen verschieden ist, und Durchführung eines Verfahrens zum Erkennen und zum Klassifizieren von Signalgrundprototypen;
• • Wechsel in eine Folge von Zuständen, die den Signalgrundprototypen einer vorgegebenen Folge von Signalgrundprototypen zugeordnet sind, wenn der erste Signalgrundprototyp dieser Folge erkannt wird;
• • Folgen der Folge der Signalgrundprototypen bis das Ende der Folge von Signalgrundprototypen erreicht ist;
• • Schließen auf das Vorliegen eines dieser Folge von Signalgrundprototypen und auf ein dieser Folge zugeordnetes Signalobjekt, wenn das Ende dieser Folge von Signalgrundprototypen erreicht ist und Signalisierung dieses Signalobjekts;
• • Abbruch (nicht in 11 eingezeichnet) dieser Folge bei Zeitüberschreitung und/oder bei einfache oder mehrfacher Detektion eines Signalgrundprototypen an einer Position dieser Folge, an der dieser Signalgrundprototyp nicht erwartet wird oder an der in der folgenden Position dieser Signalgrundprototyp nicht erwartet wird.
• • Rücksprung in den Zustand „kein Signalobjektgrundprototyp“.

A proposed method for operating an ultrasonic sensor therefore includes the steps accordingly 11 :

• • Waiting for an ultrasound reception signal that is different from the background noise,
• • Change to a state “no signal object basic prototype” if the ultrasonic received signal differs from the noise floor, and implementation of a method for recognizing and classifying signal basic prototypes;
• Change to a sequence of states which are assigned to the basic signal prototypes of a predetermined sequence of basic signal prototypes when the first basic signal prototype of this sequence is recognized;
• • Follow the sequence of basic signal prototypes until the end of the sequence of basic signal prototypes is reached;
• • Inferring the presence of one of this sequence of basic signal prototypes and of a signal object assigned to this sequence when the end of this sequence of basic signal prototypes is reached and signaling this signal object;
• • Abort (not in 11 drawn) of this sequence in the event of a timeout and / or in the case of single or multiple detection of a basic signal prototype at a position of this sequence at which this basic signal prototype is not expected or at which this basic signal prototype is not expected in the following position.
• • Return to the status "no basic signal object prototype".

The compression method described above corresponds to an associated decompression method, which is preferably used to decompress ultrasound received data compressed and transmitted in this signal object-oriented manner in the control computer after it has been received by the sensor via the said data interface. After the new data to be decompressed has been received in the control computer from the sensor, an ultrasonic echo signal model is provided that initially has no signal ( 15a) .

This ultrasonic echo signal model is now converted to signal object by successive addition of the respective prototypical, parameterized signal course of the signal objects ( 160 ) included in the ultrasonic reception data, the ultrasonic echo signal model ( 610 ) slowly filled up and approximated to the measured signal curve. Preferably, only the signal curves of signal objects are added that meet specified conditions, such as a specified chirp direction ( 13b) . It goes without saying that totaling without a selection condition is also possible. ( 15th and 16 ) Based on the ultrasonic echo signal model in the sensor's control computer, the reconstructed ultrasonic received signal is preferably formed from the summed up signal curves of the signal curves of the prototypical signal objects recognized by the sensor, which in the form of the formation of samples of a reconstructed envelope curve of the received signal (ultrasound Echo signal) can happen. This reconstructed ultrasonic reception signal (reconstructed ultrasonic echo signal) can then be used in the control computer for more complicated processes such as sensor fusion than the supposed object data that are transmitted by the sensor to the control unit.

advantage

Such a compressed transmission of data via the data bus between the sensor and the computer system reduces the data bus load and thus the criticality of EMC requirements and creates free data bus capacities during the reception time for the transmission of control commands from the computer system to the sensor and for the transmission of status information and other data from the sensor to the computer system. The proposed prioritization ensures that safety-relevant data are transmitted first and that there are no unnecessary dead times for the sensor.

Figure list

• 1 shows the basic sequence of signal compression and transmission (see description above);
• 2 shows in more detail the basic sequence of signal compression and transmission (see description above);
• 3 a shows a conventional ultrasonic echo signal ( 1 ) and its conventional evaluation.
• 3 b shows a conventional ultrasonic echo signal ( 1 ) and its evaluation, whereby the amplitude is also transmitted.
• 3c shows an ultrasonic echo signal with the chirp direction included.
• 3d shows detected signal objects (triangular signals) in the signal of the 1c discarding undetected signal components.
• 4e shows the unclaimed conventional transmission
• 4f shows the unclaimed transmission of analyzed data after complete reception of the ultrasonic echo
• 4g shows the claimed transmission of compressed data, whereby in this example symbols for basic signal objects largely without compression according to the prior art.
• 5h shows the claimed transmission of compressed data, with symbols for basic signal objects being compressed to symbols for signal objects in this example.
• 6th shows the claimed transmission of compressed data, with symbols for basic signal objects being compressed to symbols for signal objects in this example. Not only the envelope signal (ultrasonic echo signal) but also the confidence signal is evaluated.
• 7th shows an exemplary device in the form of an ultrasonic sensor with detection of signal objects, the data interface, the data bus and the control device are not entered for the sake of simplicity.
• 8th serves to explain the selection of the basic signal object prototypes of the prototype database ( 115 ) by the distance finder ( 112 ). 9 serves to explain the HMM method, which is determined by the Viterbi estimator ( 113 ) is used to identify the signal object as the most likely sequence of signal base object prototypes based on a sequence of recognized signal base object prototypes ( 121 ) as a recognized signal object ( 122 ) to identify.
• 10 shows an exemplary sequence of states in the Viterbi estimator ( 113 ) for the detection of a single signal object.
• 11 shows an exemplary, preferred sequence of states in the Viterbi estimator ( 113 ) for the continuous detection of one of signal objects, as it is typically necessary for the detection of tasks of autonomous driving.
• 12 shows an alternative design with an estimator ( 151 ) that executes a neural network model.
• 13 shows the decompression of the transmitted envelope curve of the ultrasonic signal (ultrasonic echo signal) limited to the chirp-down signal components;
• 14th shows the decompression of the transmitted envelope curve of the ultrasonic signal (ultrasonic echo signal) limited to the chirp-up signal component;
• 15th and 16 show the decompression of the transmitted envelope curve of the ultrasonic signal (ultrasonic echo signal) with chirp-up signal components and chirp-down signal components;
• 17th shows the original signal (a), the reconstructed signal (b) and their superimposition (c).
• 18th shows an improved apparatus for improved compression.

Description of the other figures

The 1 and 2 have already been described above.

Figure 3 a

3a shows the unclaimed time course of a conventional ultrasonic echo signal ( 1 ) and its unclaimed conventional evaluation in freely chosen units. Beginning with the transmission of the transmit burst (SB), a threshold value signal (SW) is carried along. Whenever the envelope signal (ultrasonic echo signal ( 1 )) of the ultrasonic reception signal exceeds the threshold value signal (SW), the output ( 2 ) set to logical 1. It is a temporally analog interface with a digital output level. The further evaluation is then taken over in the control unit of the sensor. A signaling of errors or a control of the sensor is not possible via this state-of-the-art analog interface.

Figure 3 b

3b shows the unclaimed time course of a conventional ultrasonic echo signal ( 1 ) and its unclaimed conventional evaluation in freely chosen units. Beginning with the transmission of the transmit burst (SB), a threshold value signal (SW) is carried along. Whenever the envelope signal (ultrasonic echo signal ( 1 )) of the ultrasonic reception signal exceeds the threshold value signal (SW), the output ( 2 ), however, is now set to a level corresponding to the amplitude of the detected reflection. It is a temporally analog interface with an analog output level. The further evaluation is then taken over in the control unit of the sensor. A signaling of errors or a control of the sensor is not possible via this state-of-the-art analog interface.

Figure 3c

shows the ultrasound echo signal for explanation, with the chirp direction (e.g. A = chirp up; B = chirp down) being marked.

Figure 3d

In 3d the principle of symbolic signal transmission is explained. Instead of the signal off 3c For example, only two types of triangular objects are transferred here. Specifically, these are a first triangular object (A) for the chirp-up case and a second triangular object (B) for the chirp-down case. At the same time, the point in time and the amplitude of the triangular object are transmitted. If the signal is then reconstructed on the basis of this data, a corresponding signal is obtained 3d . The signal components that did not correspond to the triangular signals were removed from this signal. There is thus a discard of undetected signal components and massive data compression.

Figure 4e

shows the conventional analog transmission of the intersection points of the ultrasonic echo signal ( 1 ) of the ultrasonic reception signal with the threshold value signal (SW).

Figure 4f

shows the unclaimed transmission of analyzed data after complete reception of the ultrasonic echo.

Figure 4g

shows the claimed transmission of compressed data, whereby in this example symbols for basic signal objects are transmitted largely without compression.

Figure 5h

shows the claimed transmission of compressed data, with symbols for basic signal objects being compressed to symbols for signal objects in this example. First a first triangle object ( 59 ) characterized by the characteristic time sequence of a threshold value being exceeded, followed by a maximum and a threshold value being exceeded, detected and transmitted. Then a double point with a saddle point ( 60 ) recognized above the threshold value signal. A characteristic here is the sequence in which the threshold value signal (SW) is exceeded by the ultrasonic echo signal ( 1 ) followed by a maximum of the ultrasonic echo signal ( 1 ) followed by a minimum above the threshold value signal (SW) followed by a maximum above the threshold value signal (SW) followed by an undershooting of the threshold value signal (SW). After recognition, the symbol for this double point with saddle point is transmitted. A time stamp is also transmitted. Other parameters of the double peak with saddle point are preferably also transmitted, such as the positions of the maxima and the minimum or a scaling factor. A triangular signal is then recognized again as the basic signal object when the threshold value signal (SW) is exceeded by the ultrasonic echo signal ( 1 ) followed by a maximum of the ultrasonic echo signal ( 1 ) followed by the ultrasonic echo signal falling below the threshold value signal (SW) ( 1 ). A double peak is then recognized again, with the minimum of the ultrasonic echo signal ( 1 ) is below the threshold value signal (SW). This double peak can therefore be treated as a separate signal object, for example. As can easily be seen, this treatment of the signal leads to a massive data reduction.

Figure 6

shows the claimed transmission of compressed data accordingly 3 , whereby in this example not only the envelope signal (ultrasonic echo signal ( 1 )), but also the confidence signal is evaluated.

Figure 7

• • Integratoren und/oder
• • Differenzierer und/oder
• • Filter und/oder
• • Logarithmierer und/oder
• • FF- und DFFT-Vorrichtungen und/oder
• • Korrelatoren und/oder
• • Demodulatoren, die ihr Eingangssignal mit einem vorgegebenen Signal multiplizieren und dann filtern, und/oder
• • andere signalverarbeitende Teilvorrichtungen und/oder
• • deren Kombinationen

je nach Anwendungsfall eingesetzt werden, die dann das n-dimensionale Zwischenparametersignal (123) erzeugen. Die mit „Optimalfilter“ bezeichneten Blöcke in den Zeichnungen sind insofern nur als Platzhalter für solche Signalverarbeitungsblöcke zu verstehen. Auch kann ein solcher mit „Optimalfilter“ bezeichneter Signalverarbeitungsblock auch mehr als einen Ausgang haben der zu dem n-dimensionalen Zwischenparametersignal (123) also mehrfach mit mehr als einem Signal beiträgt. Ein nachfolgender, beispielhafter Signifikanzsteigerer dient dazu, den n-dimensionalen Raum des Zwischenparametersignals (123) auf einen m-dimensionalen Raum des Feature-Vektor-Signals (138) abzubilden. Dabei ist m eine ganze positive Zahl. In der Regel ist m kleiner als n. Dies dient zum einen dazu, dass die Selektivität der Parameterwerte, aus denen sich jeder Feature-Vektor-Signalwert des Feature-Vektor-Signals (138) ergibt, maximiert wird. Dies geschieht bevorzugt durch eine lineare Abbildung mit Hilfe eines im Labor mit statistischen Verfahren bestimmten Offset-Werts, der zu den Werten der Zwischenparametersignal hinzuaddiert wird, und einer so genannten LDA Matrix, mit der der jeweilige Vektor der jeweiligen Abtastwerte der Zwischenparametersignale (123) jeweils zum Feature-Vektor-Signal (138) multipliziert wird. Der Abstandsermittler (112) vergleicht bevorzugt jeden sich so ergebenden Feature-Vektor-Signalwert des Feature-Vektor-Signals (138) mit jedem Signalgrundobjektprototypen der Prototypendatenbank (115). Die Prototypendatenbank (115) enthält hierfür für jeden der Signalgrundobjektprototypen der Prototypendatenbank (115) einen Eintrag mit einem Schwerpunktvektor (z.B. 141, 142, 143, 144) des jeweiligen Signalgrundobjektprototypen der Prototypendatenbank (115). Bevorzugt wird ein Abstand des aktuell untersuchten Feature-Vektor-Signalwerts des Feature-Vektor-Signals (138) zu dem geradeuntersuchten Schwerpunktvektor der Prototypendatenbank (115) durch den Abstandsermittler (112) berechnet. Der Abstandsermittler (112) kann aber auch auf andere Weise eine Bewertung der Ähnlichkeit zwischen dem Schwerpunktvektor (z.B. 141, 142, 143, 144) des jeweiligen Signalgrundobjektprototypen der Prototypendatenbank (115) und dem aktuell untersuchten Feature-Vektor-Signalwert des Feature-Vektor-Signals (138) in Form eines Bewertungswerts, den wir hier der Einfachheit halber stets Abstand nennen, berechnen. Der Abstandsermittler (112) ermittelt auf diese Weise ob ein Schwerpunktvektor (z.B. 141, 142, 143, 144) der Signalgrundobjektprototypen der Prototypendatenbank (115) dem aktuellen Feature-Vektor-Signalwert des Feature-Vektor-Signals (138) ausreichend ähnlich ist, also einen ausreichend geringen Abstand hat und, falls dies der Fall ist, welcher Schwerpunktvektor (z.B. 141, 142, 143, 144) der Signalgrundobjektprototypen der Prototypendatenbank (115) dem aktuellen Feature-Vektor-Signalwert des Feature-Vektor-Signals (138) am ähnlichsten ist, also den geringsten Abstand hat. Ggf. ermittelt der Abstandsermittler (112) eine Liste von Schwerpunktvektoren (z.B. 141, 142, d143, 144) der Signalgrundobjektprototypen der Prototypendatenbank (115), die dem aktuellen Feature-Vektor-Signalwert des Feature-Vektor-Signals (138) ausreichend ähnlich sind, also einen ausreichend geringen Abstand haben. Bevorzugt werden diese nach Abstand geordnet und mit ihrem zugehörigen Abstand als Hypothesenliste an den Viterbi-Schätzer weitergegeben. Die Weitergabe des Erkennungsergebnisses des Abstandsermittlers (der hier auch als Klassifikator bezeichnet wird) erfolgt im Falle eines einzelnen erkannten Signalgrundobjekts (121) als Symbol (Z.B. als Prototypendatenbankadresse, die auf den erkannten Signalgrundobjektprototypen der Prototypendatenbank (115) zeigt.) und im Falle einer Hypothesenliste bevorzugt als Liste von Paaren aus einem Symbol des erkannten Signalgrundobjekts (Z.B. der Prototypendatenbankadresse, die auf den erkannten Signalgrundobjektprototypen der Prototypendatenbank (115) zeigt.) und des Abstands zum Schwerpunkt dieses erkannten Signalgrundobjekts. Der Viterbi-Schätzer sucht dann, die Folge aus Signalgrundobjektprototypen, die einer vordefinierten Abfolge von Signalgrundobjektprototypen in seiner Signalobjektdatenbank (116) am besten entspricht. Hierbei werden Übereinstimmungen beispielsweise positiv gezählt und Nichtübereinstimmungen negativ gezählt, sodass sich für eine zeitliche Folge von erkannten Signalgrundobjektprototypen ein Bewertungswert für jeden Eintrag der Signalobjektdatenbank (116) ergibt. Im Falle von Hypothesenlisten überprüft der Viterbi-Schätzer bevorzugt alle möglichen Pfade durch die zeitliche Sequenz von Hypothesenlisten. Bevorzugt ermittelt er sein Bewertungsergebnis unter Berücksichtigung der zuvor ermittelten Abstände. Dies kann beispielsweise durch Division der hinzuaddierten Werte durch den Abstand vor der Addition bei der Bildung des Bewertungswerts geschehen. Auf diese Weise ermittelt der Viterbi-Schätzer die Erkannten Signalobjekte (122), die dann ggf. mit geeigneten Parametern versehen über den Datenbus übertragen werden. Wichtig ist: Es geht bei dieser Vorgehensweise nicht um das Erkennen von physikalisch im Messraum des Ultraschallwandlers (100) vorhandene Objekte und das Übertragen dieser Information, sondern um das Erkennen von Strukturen innerhalb des Ultraschall-Echo-Signals (1) und deren Übertragung.shows an exemplary device in the form of an ultrasonic sensor with detection of signal objects, the data interface, the data bus and the control device are not entered for the sake of simplicity. The ultrasonic transducer ( 100 ) is transmitted by means of an ultrasonic transducer signal ( 102 ) through a physical interface ( 101 ) controlled and measured. The physical interface ( 101 ) serves to drive the ultrasonic transducer ( 100 ) and for processing the ultrasonic transducer ( 100 ) received ultrasonic transducer signals ( 102 ) to the ultrasonic echo signal ( 1 ) for the subsequent signal object classification. The ultrasonic echo signal is preferred ( 1 ) a digitized signal with temporally spaced samples. The feature vector extractor ( 111 ) has various devices, here as an example n optimal filter (optimal filter 1 to optimal filter n) with n as a whole positive number. The outputs of the optimal filter shown here form the intermediate parameter signal ( 123 ). Instead of or in addition to the optimal filter you can also use

• • Integrators and / or
• • Differentiators and / or
• • Filters and / or
• • Logarithmizer and / or
• • FF and DFFT devices and / or
• • Correlators and / or
• • Demodulators, which multiply their input signal with a given signal and then filter, and / or
• • other signal processing sub-devices and / or
• • their combinations

depending on the application, which then use the n-dimensional intermediate parameter signal ( 123 ) produce. The blocks marked with “optimal filter” in the drawings are only to be understood as placeholders for such signal processing blocks. Such a signal processing block, referred to as an "optimal filter", can also have more than one output that corresponds to the n-dimensional intermediate parameter signal ( 123 ) thus contributes several times with more than one signal. The following, exemplary significance enhancer serves to increase the n-dimensional space of the intermediate parameter signal ( 123 ) to an m-dimensional space of the feature vector signal ( 138 ). Where m is a positive whole number. As a rule, m is smaller than n. On the one hand, this serves to ensure that the selectivity of the parameter values from which each feature vector signal value of the feature vector signal ( 138 ) is maximized. This is preferably done by a linear mapping with the help of an offset value determined in the laboratory using statistical methods, which is added to the values of the intermediate parameter signals, and a so-called LDA matrix, with which the respective vector of the respective sample values of the intermediate parameter signals ( 123 ) each to the feature vector signal ( 138 ) is multiplied. The distance finder ( 112 ) preferably compares each resulting feature vector signal value of the feature vector signal ( 138 ) with each signal base object prototype in the prototype database ( 115 ). The prototype database ( 115 ) contains for each of the basic signal object prototypes of the prototype database ( 115 ) an entry with a centroid vector (e.g. 141 , 142 , 143 , 144 ) of the respective basic signal object prototype of the prototype database ( 115 ). A distance between the currently examined feature vector signal value of the feature vector signal is preferred ( 138 ) to the just examined center of gravity vector of the prototype database ( 115 ) by the distance finder ( 112 ) calculated. The distance finder ( 112 ) but an evaluation of the similarity between the center of gravity vector (e.g. 141 , 142 , 143 , 144 ) of the respective basic signal object prototype of the prototype database ( 115 ) and the currently examined feature vector signal value of the feature vector signal ( 138 ) in the form of an evaluation value, which we always call distance here for the sake of simplicity. The distance finder ( 112 ) determines in this way whether a center of gravity vector (e.g. 141 , 142 , 143 , 144 ) the signal base object prototypes of the prototype database ( 115 ) the current feature vector signal value of the feature vector signal ( 138 ) is sufficiently similar, i.e. has a sufficiently small distance and, if this is the case, which center of gravity vector (e.g. 141 , 142 , 143 , 144 ) the signal base object prototypes of the prototype database ( 115 ) the current feature vector signal value of the feature vector signal ( 138 ) is most similar, i.e. has the smallest distance. Possibly. the distance finder determines ( 112 ) a list of centroid vectors (e.g. 141 , 142 , d143, 144 ) the signal base object prototypes of the prototype database ( 115 ), which corresponds to the current feature vector signal value of the feature vector signal ( 138 ) are sufficiently similar, i.e. have a sufficiently small distance. These are preferably sorted according to distance and passed on to the Viterbi estimator with their associated distance as a list of hypotheses. Passing on the The detection result of the distance finder (which is also referred to here as a classifier) occurs in the case of a single detected basic signal object ( 121 ) as a symbol (e.g. as a prototype database address based on the recognized basic signal object prototypes of the prototype database ( 115 ) and in the case of a hypothesis list, preferably as a list of pairs from a symbol of the recognized basic signal object (e.g. the prototype database address, which is based on the recognized basic signal object prototypes of the prototype database ( 115 ) shows.) and the distance to the center of gravity of this recognized basic signal object. The Viterbi estimator then searches for the sequence of signal basic object prototypes that correspond to a predefined sequence of signal basic object prototypes in its signal object database ( 116 ) corresponds best. In this case, matches are counted positively and non-matches counted negatively, so that an evaluation value for each entry in the signal object database ( 116 ) results. In the case of lists of hypotheses, the Viterbi estimator preferably checks all possible paths through the temporal sequence of lists of hypotheses. He preferably determines his evaluation result taking into account the previously determined distances. This can be done, for example, by dividing the added values by the distance before the addition when forming the evaluation value. In this way, the Viterbi estimator determines the recognized signal objects ( 122 ), which are then transferred via the data bus, if necessary with suitable parameters. The important thing is: This procedure is not about the detection of physically in the measurement room of the ultrasonic transducer ( 100 ) existing objects and the transmission of this information, but rather about the recognition of structures within the ultrasonic echo signal ( 1 ) and their transmission.

Figure 8

serves to explain the selection of the basic signal object prototypes of the prototype database ( 115 ) by the distance finder ( 112 ).

The position of the distance calculation ( 112 ) in the exemplary two-dimensional parameter space of 8th determined current feature vector signal values of the feature vector signal ( 138 ) can be very different. It is conceivable that such a first exemplary feature vector signal value ( 146 ) too far from the center of gravity coordinates ( 141 , 142 , 143 , 144 ) the center of gravity of any signal basic object prototype in the prototype database ( 115 ) is away. This distance threshold value can, for example, be the said minimum half the prototype distance among the signal base object prototypes in the prototype database ( 115 ) be. It can also be the case that the scatter areas of signal base object prototypes in the prototype database ( 115 ) by their respective priorities ( 143 , 142 ) overlap and a second exemplary determined current feature vector signal value ( 145 ) of the feature vector signal ( 138 ) lies in such an overlap area. In this case, a hypothesis list can contain both signal basic object prototypes with different probabilities as attached parameters, since they have different distances. These probabilities are preferably represented by the distances. So there is then not a signal basic object as the most likely to the Viterbi estimator ( 113 ), but a vector from possibly existing basic signal objects to the Viterbi estimator ( 113 ) to hand over. From the chronological sequence of these lists of hypotheses, the Viterbi estimator then searches for the possible sequence that has the greatest probability of one of the specified signal basic object sequences in its signal object database compared to all possible paths through the signal basic objects recognized as possible ( 121 ) from the distance finder ( 112 ) by the Viterbi estimator ( 113 ) received hypothesis lists. In this case, exactly one recognized basic signal object prototype of this hypothesis list must be run through this path in such a path for each hypothesis list.

In the best case, the current feature vector signal value is ( 148 ) in the scatter range (threshold ellipsoid) ( 147 ) around the center of gravity ( 141 ) of a single signal base object prototype ( 141 ) which is thus safely through the distance finder ( 112 ) recognized and recognized as a recognized basic signal object ( 121 ) to the Viterbi estimator ( 113 ) is passed on.

It is conceivable, in order to improve the modeling of the scatter range of an individual signal base object prototype, to model it using several here circular signal base object prototypes with associated scatter ranges. Several basic signal object prototypes from the prototype database ( 115 ) represent the same basic signal object prototype in the understanding of a basic signal object class.

The distance finder ( 112 ) preferably compares each resulting feature vector signal value of the feature vector signal ( 138 ) with everybody Signal base object prototypes of the prototype database ( 115 ). The prototype database ( 115 ) contains for each of the basic signal object prototypes of the prototype database ( 115 ) an entry with a centroid vector (e.g. 141 , 142 , 143 , 144 ) of the respective basic signal object prototype of the prototype database ( 115 ). A distance between the currently examined feature vector signal value of the feature vector signal is preferred ( 138 ) to the just examined center of gravity vector of the prototype database ( 115 ) by the distance finder ( 112 ) calculated. The distance finder ( 112 ) but an evaluation of the similarity between the center of gravity vector (e.g. 141 , 142 , 143 , 144 ) of the respective basic signal object prototype of the prototype database ( 115 ) and the currently examined feature vector signal value of the feature vector signal ( 138 ) in the form of an evaluation value, which we always call distance here for the sake of simplicity. The distance finder ( 112 ) determines in this way whether a center of gravity vector (e.g. 141 , 142 , 143 , 144 ) the signal base object prototypes of the prototype database ( 115 ) the current feature vector signal value of the feature vector signal ( 138 ) is sufficiently similar, i.e. has a sufficiently small distance and, if this is the case, which center of gravity vector (e.g. 141 , 142 , 143 , 144 ) the signal base object prototypes of the prototype database ( 115 ) the current feature vector signal value of the feature vector signal ( 138 ) is most similar, i.e. has the smallest distance. Possibly. the distance finder determines ( 112 ) a list of centroid vectors (e.g. 141 , 142 , 143 , 144 ) the signal base object prototypes of the prototype database ( 115 ), which corresponds to the current feature vector signal value of the feature vector signal ( 138 ) are sufficiently similar, i.e. have a sufficiently small distance.

Figure 9

serves to explain the HMM method, which is determined by the Viterbi estimator ( 113 ) is used to identify the signal object as the most likely sequence of signal base object prototypes based on a sequence of recognized signal base object prototypes ( 121 ) as a recognized signal object ( 122 ) to identify. The 9 is explained above in the text.

Figure 10

shows an exemplary sequence of states in the Viterbi estimator ( 113 ) for the detection of a single signal object. The 10 is explained above in the text.

Figure 11

shows an exemplary, preferred sequence of states in the Viterbi estimator ( 113 ) for the continuous detection of one of signal objects, as it is typically necessary for the detection of tasks of autonomous driving. The 11 is explained above in the text.

Figure 12

shows an alternative design with an estimator ( 151 ) that executes a neural network model. The data interface, the data bus and the control unit are not entered for the sake of simplicity. The ultrasonic transducer ( 100 ) is transmitted by means of an ultrasonic transducer signal ( 102 ) through a physical interface ( 101 ) controlled and measured. The physical interface ( 101 ) serves to drive the ultrasonic transducer ( 100 ) and for processing the ultrasonic transducer ( 100 ) received ultrasonic transducer signal ( 102 ) to the ultrasonic echo signal ( 1 ) for the subsequent signal object classification. The ultrasonic echo signal is preferred ( 1 ) a digitized signal with temporally spaced samples. The feature vector extractor ( 111 ) again has various devices, here as an example n optimal filter (optimal filter 1 to optimal filter n) with n as a whole positive number. The outputs of the optimal filter shown here form the intermediate parameter signal ( 123 ). Instead of or in addition to the optimal filters, integrators, filters, differentiators, logarithmizers and / or other signal-processing sub-devices and their combinations can be used depending on the application, which then use the n-dimensional intermediate parameter signal ( 123 ) produce. The following, exemplary significance enhancer serves to increase the n-dimensional space of the intermediate parameter signal ( 123 ) to an m-dimensional space of the feature vector signal ( 138 ). Where m is a positive whole number. As a rule, m is smaller than n. On the one hand, this serves to ensure that the selectivity of the parameter values from which each feature vector signal value of the feature vector signal ( 138 ) is maximized. This is preferably done by a linear mapping with the help of an offset value determined in the laboratory using statistical methods, which is added to the values of the intermediate parameter signals, and a so-called LDA matrix, with which the respective vector of the respective sample values of the intermediate parameter signals ( 123 ) each to the feature vector signal ( 138 ) is multiplied. The estimator ( 151 ) then tries with the help of a neural network model ( 151 ) the signal objects in the data stream of the feature vector signal values of the feature vector signal ( 138 ) to recognize. The each signal base object prototype and the signal objects are connected to the nodes within the encoded neural network model and the parameterization of the neural network model within the neural network model. The estimator ( 151 ) then gives the recognized signal objects ( 122 ) out.

Figure 13

shows the exemplary decompression of the transmitted exemplary envelope signal (ultrasonic echo signal ( 1 )) of the ultrasonic received signal is limited to the chirp-down signals.

First, an ultrasonic echo signal model is generated that has no signal. ( 13a) . The ultrasonic echo signal model is parameterized with a model parameter (SA) which is preferably correlated with the time (t) since the transmission of the ultrasonic pulse or burst.

Then this ultrasonic echo signal model is preferred by adding it to that with data transmission priority 1 first exemplary signal object transmitted from the sensor to the control unit ( 160 ), which describes a triangular shape with Chirp Down (A), is suitably parameterized. ( 13b)

That with data priority 2 transmitted second exemplary signal object ( 161 ) in triangular shape with a chirp up characteristic (B) is not taken into account here for the reconstruction of the chirp down signal.

That with data priority 3 transmitted third exemplary signal object ( 162 ) in double-peak form with a chirp-up characteristic (B) is not taken into account here for the reconstruction of the chirp-down signal.

Then the already supplemented ultrasonic echo signal model is preferred by adding the one with data transmission priority 4th Fourth exemplary signal object transmitted from the sensor to the control unit ( 163 ), which describes a triangular shape with Chirp Down (A), is suitably parameterized. ( 13c )

Then the already supplemented ultrasonic echo signal model is preferred by adding the one with data transmission priority 5 Fifth exemplary signal object transmitted from the sensor to the control unit ( 164 ), which describes a triangular shape with Chirp Down (A), is suitably parameterized. ( 13d )

That with data priority 6th transmitted sixth exemplary signal object ( 165 ) in double-peak form with a chirp-up characteristic (B) is not taken into account here for the reconstruction of the chirp-down signal.

The resulting reconstructed envelope signal is the reconstructed chirp-down ultrasonic echo signal.

The thus decompressed and reconstructed ultrasonic echo signal is then typically used for object recognition in the control unit or at another point in the vehicle.

Figure 14

shows the exemplary decompression of the transmitted exemplary envelope signal (ultrasonic echo signal ( 1 )) of the ultrasonic received signal is limited to the chirp-up signals.

First, an ultrasonic echo signal model is generated that has no signal. ( 14a) . The ultrasonic echo signal model is parameterized with a model parameter (SA) which is preferably correlated with the time (t) since the transmission of the ultrasonic pulse or burst.

That with data priority 1 transmitted first exemplary signal object ( 160 ) in triangular shape with a chirp down characteristic (A) is not taken into account here for the reconstruction of the chirp up signal.

Then this ultrasonic echo signal model is preferred by adding it to that with data transmission priority 2 Second exemplary signal object transmitted from the sensor to the control unit ( 161 ), which describes a triangular shape with Chirp Up (B), is supplemented with suitable parameters. ( 14b)

Then this ultrasonic echo signal model is preferred by adding it to that with data transmission priority 3 Third exemplary signal object transmitted from the sensor to the control unit ( 161 ), which describes a double point shape with chirp up (B), is supplemented with suitable parameters. ( 14c )

That with data priority 4th transmitted fourth exemplary signal object ( 162 ) in triangular shape with a chirp down characteristic (A) is not taken into account here for the reconstruction of the chirp up signal.

That with data priority 5 transmitted fifth exemplary signal object ( 163 ) in triangular shape with a chirp down characteristic (A) is not taken into account here for the reconstruction of the chirp up signal.

Then this ultrasonic echo signal model is preferred by adding it to that with data transmission priority 6th Sixth exemplary signal object transmitted from the sensor to the control unit ( 165 ), which describes a triangular shape with Chirp Up (B), is supplemented with suitable parameters. ( 14d )

The resulting reconstructed envelope signal is the reconstructed chirp-up ultrasonic echo signal.

The thus decompressed and reconstructed ultrasonic echo signal is then typically used for object recognition in the control unit or at another point in the vehicle.

Figures 15 and 16

show the decompression of the transmitted envelope signal (ultrasonic echo signal ( 1 )) the ultrasonic received signal with chirp-up signals and chirp-down signals;
First, an ultrasonic echo signal model is generated that has no signal. ( 15a) . The ultrasonic echo signal model is parameterized with a model parameter (SA) which is preferably correlated with the time (t) since the transmission of the ultrasonic pulse or burst.

Then this ultrasonic echo signal model is preferred by adding it to that with data transmission priority 1 first exemplary signal object transmitted from the sensor to the control unit ( 160 ), which describes a triangular shape with Chirp Down (A), is suitably parameterized. ( 15b)

Then this ultrasonic echo signal model is preferred by adding it to that with data transmission priority 2 Second exemplary signal object transmitted from the sensor to the control unit ( 161 ), which describes a triangular shape with Chirp Up (B), is supplemented with suitable parameters. ( 15c )

Then this ultrasonic echo signal model is preferred by adding it to that with data transmission priority 3 Third exemplary signal object transmitted from the sensor to the control unit ( 162 ), which describes a double point shape with chirp up (B), is supplemented with suitable parameters. ( 15d )

Then this ultrasonic echo signal model is preferred by adding it to that with data transmission priority 4th Fourth exemplary signal object transmitted from the sensor to the control unit ( 163 ), which describes a triangular shape with Chirp Down (A), is suitably parameterized. ( 16d )

Then this ultrasonic echo signal model is preferred by adding it to that with data transmission priority 5 Fifth exemplary signal object transmitted from the sensor to the control unit ( 164 ), which describes a triangular shape with Chirp Down (A), is suitably parameterized. ( 16f)

Then this ultrasonic echo signal model is preferred by adding it to that with data transmission priority 6th Sixth exemplary signal object transmitted from the sensor to the control unit ( 165 ), which describes a triangular shape with Chirp Up (B), is supplemented with suitable parameters. ( 16g)

For clarification, in 16h the reconstructed envelope signal is shown in bold. Otherwise true 16h with the 16g match.

The resulting reconstructed envelope signal is the reconstructed chirp-up / chirp-down ultrasonic echo signal.

The thus decompressed and reconstructed ultrasonic echo signal is then typically used for object recognition in the control unit or at another point in the vehicle.

Figure 17

shows the original signal (a), the reconstructed signal (b) and their superimposition (c). The deviations between the reconstructed ultrasonic echo signal ( 166 ) and the ultrasonic echo signal are very low if the procedure is correct and the basic signal objects and objects are skilfully selected.

Figure 18

18th shows an improved apparatus for improved compression. The one in the 13 to 17th Reconstructor explained in his function ( 600 ) can be used not only in the control unit, but also in the sensor itself before the data transfer. To do this, the reconstructor generates ( 600 ) a reconstructed ultrasonic echo signal model ( 610 ) as in the 13 to 17th shown depending on the purpose. This is derived from the previously received ultrasonic echo signal ( 1 ) in a subtracter ( 602 ) deducted. For this purpose, the ultrasonic echo signal ( 1 ) preferably in the form of digital samples in a memory ( 601 ) preferably filed in chronological order. A stored sample value of the ultra sound echo signal corresponds to this ( 1 ) preferably with exactly one value of the reconstructed ultrasonic echo signal model ( 610 ), which is preferably also stored in a reconstruction memory ( 603 ) is filed. The reconstruction memory ( 603 ) can be part of the reconstructor ( 600 ) be. This additional feedback branch ( 600 , 610 , 603 , 602 , 601 ) can also be used for other classifiers such as the device of 12 be used.

List of reference symbols

α

Sending out the ultrasonic burst;

β

Receiving the ultrasonic burst reflected on an object and converting it into an electrical received signal;

γ

Compression of the electrical received signal;

γa

Sampling of the electrical input signal and formation of a sampled electrical input signal, wherein a time stamp can preferably be assigned to each sampling value of the electrical input signal.

γb

Determination of several spectral values e.g. through matched filters for prototypical signal object classes. These multiple spectral values together form a feature vector. This formation preferably takes place continuously, so that a stream of feature vector values results. A time stamp value can preferably be assigned to each feature vector value;

γc

optional, but preferably carried out normalization of the feature vector spectral coefficients of the respective feature vector of a time stamp value before the correlation with the prototypical signal object classes in the form of predetermined prototypical feature vector values of a prototype library;

γd

Determination of the distance between the current feature vector value and the values of the prototypical signal object classes in the form of predetermined prototypical feature vector values of a prototype library;

γe

Selection of the most similar prototypical signal object class in the form of a given prototypical feature vector value of a prototype library with preferably a minimum distance to the current feature vector and adoption of the symbol of this signal object class as a recognized signal object together with the time stamp value as compressed data. Possibly. further data, in particular signal object parameters, such as e.g. whose amplitude is also taken over as compressed data. These compressed data then form the compressed received signal;

δ

Transmission of the compressed electrical reception signal to the computer system:

1

Envelope curve of the received ultrasonic signal, also referred to here as ultrasonic echo signal;

2

Output signal (transmitted information) of an IO interface according to the prior art;

3

transmitted information of a LIN interface according to the prior art;

4th

exemplary, first intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in downward direction .;

5

exemplary, first intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction;

6th

exemplary first maximum of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW);

7th

exemplary, second intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction;

8th

exemplary, second intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction;

9

exemplary second maximum of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW);

10

exemplary first minimum of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW);

11

exemplary third maximum of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW);

12

exemplary, third intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction;

13

exemplary, third intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction;

14th

exemplary fourth maximum of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW);

15th

exemplary fourth point of intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction;

16

exemplary fourth point of intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction;

17th

exemplary fifth maximum of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW);

18th

exemplary, fifth intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction;

19th

exemplary, fifth intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction;

20th

exemplary sixth maximum of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW);

21st

exemplary, sixth intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction;

22nd

exemplary, sixth intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction;

23

exemplary seventh maximum of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW);

24

exemplary, seventh intersection of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction;

25th

Envelope curve during the ultrasonic burst

26th

Transmission of the data of the exemplary, first intersection point ( 4th ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction via the preferably bidirectional data bus;

27

Transmission of the data of the exemplary, first intersection point ( 5 ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction via the preferably bidirectional data bus;

28

Transmission of the data of the exemplary first maximum ( 6th ) of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW) via the preferably bidirectional data bus;

29

Transmission of the data of the exemplary, second intersection point ( 7th ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction via the preferably bidirectional data bus;

30th

Transmission of the data of the exemplary, second intersection point ( 8th ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction via the preferably bidirectional data bus;

31

Transmission of the data of the exemplary second maximum ( 9 ) of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW) and the data of the exemplary first minimum ( 10 ) of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW) via the preferably bidirectional data bus;

32

Transmission of the data of the exemplary third maximum ( 11 ) of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW) via the preferably bidirectional data bus;

34

Transmission of the data of the exemplary third intersection point ( 12 ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction via the preferably bidirectional data bus;

35

Transmission of the data of the exemplary third intersection point ( 13 ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction via the preferably bidirectional data bus;

36

Transmission of the data of the exemplary fourth maximum ( 14th ) of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW) via the preferably bidirectional data bus;

37

Transmission of the data of the exemplary fourth intersection ( 15th ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction via the preferably bidirectional data bus;

38

Transmission of the data of the exemplary fourth intersection ( 16 ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction via the preferably bidirectional data bus;

39

Transmission of the data of the exemplary fifth maximum ( 17th ) of the ultrasonic echo signal ( 1 ) above the threshold value signal (SW) via the preferably bidirectional data bus;

40

Transmission of the data of the exemplary fifth intersection point ( 18th ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction via the preferably bidirectional data bus;

41

Transmission of the data of the exemplary fifth intersection point ( 19th ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the upward direction via the preferably bidirectional data bus;

42

Transmission of the data of the exemplary sixth intersection ( 21st ) of the ultrasonic echo signal ( 1 ) with the threshold value signal (SW) in the downward direction via the preferably bidirectional data bus;

43

Transmission of the data of the received echoes on the LIN bus according to the prior art after the end of the reception;

44

Transmission of data on the LIN bus according to the state of the art before the ultrasonic burst is transmitted;

45

Transmission of data via the IO interface according to the state of the art before the ultrasonic burst is transmitted;

46

Effect of the ultrasonic transmission burst on the output signal of the IO interface according to the prior art;

47

First echo signal ( 5 , 6th , 7th ) on the IO interface acc. the state of the art;

48

Second echo signal ( 8th , 9 , 10 , 11 , 12 ) on the IO interface acc. the state of the art;

49

Third and fourth echo signal ( 13 , 14th , 15th ) on the IO interface acc. the state of the art;

50

Fifth echo signal ( 16 , 17th , 18th ) on the IO interface acc. the state of the art;

51

Sixth echo signal ( 19th , 20th , 21st ) on the IO interface acc. the state of the art;

52

Seventh echo signal ( 22nd , 23 , 24 ) on the IO interface acc. the state of the art;

53

Start command from the computer system to the sensor via the data bus;

54

periodic automatic data transmission between sensor and computer system, preferably according to the DSI3 standard;

55

Diagnostic bits after the measuring cycle;

56

End of the transmission of the ultrasound burst (end of the transmission burst). The end of the ultrasonic burst preferably coincides with point 4.

57

Beginning of the transmission of the ultrasonic burst (beginning of the transmission burst).

58

End of data transfer

100

Ultrasonic transducer which, for example, can also comprise a separate ultrasonic transmitter and a separate ultrasonic receiver;

101

physical interface for driving the ultrasonic transducer ( 100 ) and for processing the ultrasonic transducer ( 100 ) received ultrasonic transducer signal ( 102 ) to the ultrasonic echo signal ( 1 ) for the subsequent signal object classification;

102

Ultrasonic transducer signal;

111

Feature vector extractor ( 111 );

112

Distance finder (or classifier);

113

Viterbi estimator.

115

Prototype database;

116

Signal object database;

122

recognized signal objects with signal object parameters;

123

Intermediate parameter signals;

125

Significance increase;

126

LDA matrix;

138

Feature vector signal or feature vector signal;

141

Center of gravity coordinate of a first exemplary prototypical basic signal object in the prototype database ( 115 );

142

Center of gravity coordinate of a second exemplary prototypical basic signal object in the prototype database ( 115 );

143

Center of gravity coordinate of a third exemplary prototypical basic signal object of the prototype database (115) ;

144

Center of gravity coordinate of an exemplary prototypical basic signal object of the prototype database ( 115 );

145

Current feature vector signal value of the feature vector signal ( 138 ) in the overlap area of the scatter areas of the two exemplary basic signal objects of the prototype database ( 115 ) with the center of gravity coordinates 142 and 143 ;

146

exemplary feature vector signal value which is too far from the exemplary center of gravity coordinates ( 141 , 142 , 143 , 144 ) the center of gravity of any exemplary signal basic object prototype in the prototype database ( 115 ) is distant;

147

Scatter range (threshold ellipsoid) ( 147 ) around the center of gravity ( 141 ) of a single exemplary signal basic object prototype ( 141 );

148

exemplary current feature vector signal value which is in the scatter range (threshold ellipsoid) ( 147 ) around the center of gravity ( 141 ) of a single exemplary signal basic object prototype ( 141 ) and which is thus safely determined by the distance finder ( 112 ) recognized and recognized as a recognized basic signal object ( 121 ) to the Viterbi estimator ( 113 ) can be passed on.

150

Estimator using an HMM model;

151

Estimator with a neural network model;

160

first exemplary signal object in triangular shape with chirp down (A) and data transmission priority 1 ;

161

second exemplary signal object in triangular shape with chirp up (B) and data transmission priority 2 ;

162

third exemplary signal object in double-peak form with chirp up (B) and data transmission priority 3 (Curbstone profile);

163

fourth exemplary signal object in triangular shape with chirp down (A) and data transmission priority 4th ;

164

fifth exemplary signal object in triangular shape with chirp down (A) and data transmission priority 5 ;

165

sixth exemplary signal object in triangular shape with chirp UP (B) and data transmission priority 6th ;

166

Echo signal transmitted and decompressed with the aid of signal objects;

600

Reconstructor;

601

Memory for the ultra sound echo signal ( 1 ). The memory preferably comprises the data of the echo of an ultrasonic pulse or burst;

602

Subtractor for the subtraction of the reconstructor ( 600 ) calculated reconstructed sampled values of the ultrasonic echo signal that form the reconstructed ultrasonic echo signal model ( 610 ), of those in memory ( 601 ) stored samples of the ultrasonic echo signal ( 1 ) to read the residual signal ( 660 ) as an alternative input signal for the feature vector extractor (111) instead of the ultrasonic echo signal ( 1 ) serves;

603

Reconstruction memory for the reconstructed ultrasonic echo signal model ( 610 ). The reconstruction memory is usually used as part of the reconstructor ( 600 ) executed.

610

reconstructed ultrasonic echo signal model;

660

Residual signal. The residual signal represents the compression error. The more objects are detected, the smaller the compression error. The compression is usually canceled when all samples of the residual signal ( 660 ) lie below a threshold value curve for the compression. The corresponding termination signaling is not shown in the figures.

a

Transmitted information for the transmission of the received ultrasonic echoes by means of an IO interface from the prior art;

A.

Exemplary recognized object with chirp down;

a.u. "Arbritrary units" =

freely chosen units

B.

Exemplary recognized object with chip-up;

b

Transmitted information for the transmission of the received ultrasonic echoes by means of a LIN interface from the prior art;

c

Transferred information for the transfer of the received ultrasonic echoes by means of the proposed method and the proposed device with envelope curve (ultrasonic echo signal ( 1 )) for comparison;

d

Transmitted information for the transmission of the received ultrasonic echoes by means of the proposed method and the proposed device without an envelope curve;

e

schematic signal forms during the transmission of the received echo information by means of an IO interface from the prior art;

En

Amplitude of the envelope curve (ultrasonic echo signal ( 1 )) of the received ultrasonic signal;

f

schematic signal forms during the transmission of the received echo information by means of a LIN interface from the prior art;

G

schematic signal forms during the transmission of the received echo information by means of a bidirectional data interface;

SA

Memory address in memory ( 601 ) or reconstruction memory ( 603 ). This is typically correlated with a time since the emission of an ultrasonic pulse and / or ultrasonic burst.

SB

Send burst

SW

threshold

t

Time;

T _E

Reception time. The reception time typically begins with the end ( 56 ) of sending out the ultrasonic burst. It is possible to start receiving beforehand. However, this can lead to problems that may require additional measures.

QUOTES INCLUDED IN THE DESCRIPTION

This list of the documents listed by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent literature cited

• WO 2012/016834 A1 [0004]
• DE 4433957 A1 [0005]
• DE 102012015967 A1 [0006]
• DE 102011085286 A1 [0007]
• WO 2014108300 A1 [0008]
• DE 102015104934 A1 [0009]
• DE 102015104934 A [0009]
• DE 102010041424 A1 [0011]
• DE 102013226376 A1 [0012]
• DE 10024959 A1 [0013]
• DE 102013015402 A1 [0014]
• US 2006/0250297 A1 [0015]
• WO 2018/210966 A1 [0016]
• DE 102018106244 B3 [0017]
• DE 102019106190 A1 [0018]
• DE 102017100837 [0141]
• DE 102017100835 [0141]

Claims (12)

Method for the transmission of sensor data, in particular an ultrasonic sensor, from a sensor to a computer system, in particular in a vehicle, having or comprising the steps - sending out an ultrasonic burst; Receiving an ultrasonic signal and forming an ultrasonic receiving signal; - Forming a feature vector signal (138) from the ultrasound received signal (1) or a modified residual signal (660) derived therefrom; - Evaluation of at least time segments (Hamming Windows) of the feature vector signal (138) • by forming at least one binary, digital or analog distance value between the feature vector signal and one or more signal object prototype values for recognizable signal object classes and assigning a recognizable signal object class as a recognized signal object if the amount of the distance value is one or more predetermined, binary, digital or analog The amount falls below the distance values, and / or • by assigning a recognizable signal object class as a recognized signal object, in particular by means of an estimator (151), with a neural network model and / or with a Petri network; - Recognition of signal objects and classification of these signal objects in assigned signal object classes within the ultrasonic received signal (1), • where a signal object class can also include only one signal object and • with each signal object (122) thus recognized and classified at least one assigned signal object parameter and a symbol corresponding to the signal object class assigned to this signal object, or • with each signal object (122) thus recognized and classified at least one assigned signal object parameter and a symbol for this signal object being determined; - Transmission of at least the symbol of a recognized signal object class (122) and at least the one assigned signal object parameter of this recognized signal object class (122) - Reconstruction of a reconstructed ultrasound echo signal model (610) from recognized signal objects (122); - Subtraction of the reconstructed ultrasonic echo signal model (610) from the ultrasonic received signal (1) to form a residual signal (660); - Use of the residual signal (660) for the formation of the feature vector signal (138). Procedure according to Claim 1 with the additional step of ending the classification into signal objects if the amounts of the samples of the residual signal (660) are below the amounts of a predetermined threshold value curve; Decompression method, for the decompression of ultrasound received data, which by means of a method according to one or more of Claims 1 to 2 have been compressed, with the steps - receiving the data to be decompressed; - Providing an ultrasonic echo signal model that has no signal ( 13a) ; - Supplementation of the ultrasonic echo signal model by adding a signal object (160) which is contained in the ultrasonic reception data; - Formation of the reconstructed ultrasonic received signal; - Using the reconstructed ultrasound received signal. Decompression procedure, after Claim 3 with the step of using the reconstructed ultrasonic reception signal in a vehicle for the purposes of autonomous driving and / or for the purposes of generating maps of the surroundings. Sensor, in particular ultrasonic sensor, which is used to carry out a method according to one or more of the Claims 1 to 4th is provided. Computer system - that is used to carry out a method according to one or more of the Claims 3 to 4th is provided or - to be used together with a with a sensor Claim 5 is provided and is suitable for receiving data from this sensor, - the sensor data being ultrasound reception data, which by means of a method according to one or more of the Claims 1 to 2 have been compressed. Sensor system - with at least one computer system according to Claim 6 and - with at least one sensor Claim 5 - The computer system for performing a method according to one or more of Claims 3 to 4th is provided. Sensor system - with at least one computer system according to Claim 6 and - with at least two sensors Claim 5 - The data transmission between the sensors and the computer system at least partially according to a method according to one or more of the Claims 1 to 4th expires. Sensor system according to Claim 8 - With one ultrasonic received signal, ie at least two ultrasonic received signals, within the sensors, by means of a method corresponding to a method according to one or more of Claims 1 to 2 is compressed and transmitted to the computer system and - the at least two ultrasonic reception signals being reconstructed within the computer system by decompressing data received from the sensor into reconstructed ultrasonic reception signals. Sensor system according to Claim 8 - the computer system using reconstructed ultrasonic reception signals to recognize objects in the vicinity of the sensors. Sensor system according to Claim 10 - the computer system using reconstructed ultrasound reception signals and additional signals from further sensors, in particular signals from radar sensors, to recognize objects in the vicinity of the sensors. Sensor system according to one or more of the Claims 10 or 11 - The computer system creating an environment map for the sensors or a device of which the sensors are part based on the detected objects.

Images