GB2523904A

GB2523904A - Device for echo-based environment sensor systems

Info

Publication number: GB2523904A
Application number: GB1502620.6A
Authority: GB
Inventors: Matthias Karl
Original assignee: Robert Bosch GmbH
Current assignee: Robert Bosch GmbH
Priority date: 2014-02-17
Filing date: 2015-02-17
Publication date: 2015-09-09
Anticipated expiration: 2035-02-17
Also published as: FR3017718A1; FR3017718B1; DE102014202864A1; GB201502620D0; GB2523904B

Abstract

A means of locomotion (car 12, fig. 4) has an echo-based sensor for the surrounding environment in its front and rear bumpers (13, fig. 4). The device comprises a transducer 2 and an evaluating unit 3. The evaluating unit determines a similarity (X(T)) between a signal (e(I)) coming from the signal transducer (2) and an expected signal (s(T)), and also determines a relative measure (K(T)) of the similarity (X(T)) by the similarity (X(T)) being referred to a predefined reference quantity (X1i.) for the similarity (X(T)) such as their ratio, thereby normalizing the similarity. The result suppresses noise and interference, and takes into account the energy of the signals.

Description

Description Title

Device for echo-based environment sensor systems

Prior art

The present invention relates to a device for echo-based environment sensor systems and to a means of locomotion equipped with such a device. In particular, the present invention relates to a device which is configured to extract environment detection-related signals from transducer signals in an improved manner.

Receiving channels are frequently described as an additive superposition of an expected signal S(i), of a stochastic internal state variable N(T), which is commonly called "noise", and sometimes of a more or less determined quantity 1(t), the interfezer, also referred to as jamming, which is not expected, but not unlike the expected signal.

The instantaneous energy/power of a received signal E(t) can therefore be described as Formula 1.: ELt) = S(t) + 1(T) + N(t) The quantity t in this context denotes the time difference between the end of the emitting of a measuring signal into the environment or surroundings and the receiving instant of the emitted signal at the receiver. Thus, t denotes a time interval, in which the transducer is substantially free from internal or external excitations. The transition between N(i) and 1(i) cannot always be sharply delineated and is frequently also determined by the expected signal S(t). In the context of the present description, however, it is not a matter of distinguishing between the two unexpected quantities N(T) and 1(T). Frequently, the noise is a stationary quantity, so that Formula 2.: N(-r) = N = const applies.

The simplified channel model is thus given as: Formula 3.: E(i) = S(t) + N A fundamental principle of communication and measurement technology is to obtain the expected signal S(-r) by means of suitable filters from the received signal e(i) Customarily, suitable filters, in particular so-called matched filters, are used for this. The output signal of such a filter X(i) is an optimal estimate of the expected signal S(T) for transmission channels disturbed by white noise. In acoustic environment detection, the strength of the received signal S(t) is a measure of the reflexivity and is to be taken into account, according to the invention, in the signal evaluation. In this context, the proven subtractive assessment by means of a threshold or a limit value Xjj(T) is considered as known, where Formula 4.: XK(T) = X(T) -Xnm(T) applies.

The threshold Xijm(T) here is comparable with the threshold used hitherto or the characteristic curve used hitherto. If the filtered signal X(T) exceeds the threshold value X11(T), an object is recognised in the echo signal. The degree to which the threshold value X(t)is exceeded is a measure of the strength of the object reflexivity and thus a measure which is suitable for object detection. The assessment method used here has an associated unit, i.e. XK(T) has the unit of X(i) . An advantage of the absolute, i.e. subtractive assessment of the receiving signal is its simple realisability. A simply realisable threshold value switch is already sufficient. A disadvantage of the subtractive assessment, however, is that relatively high signal dynamics must be provided within an echo cycle owing to the large dynamic range of the filtered received signals X(T) . A correspondingly high resolution is required here for small signals.

A further approach in the environment detection pursues a relative measure of the similarity of the received signal in that, by rearranging Schwarz's inequality, the measure of the similarity of the expected signal X(T) is referred to the total energy E(i) of the signal; Formula 5.: R(r) 4sN Owing to the normalisation, the assessment method here is a unitless assessment method in which the relationship between the two quantities X(t) and E(t) is obtained. Owing to Schwarz's inequality, the value R(T) always lies between o and 1. Only in the presence of a dominant reflective object is R(T) markedly closer to 1, otherwise 0. This interrelationship is substantially independent of the reflexivity of the object.

An advantage of the relative assessment is that the dynamic range can be compressed and the output signal manages with a relatively low resolution/discretisation, regardless of the instantaneous signal strength. The disadvantage, however, is the high computational complexity.

Depending on the technical realisation, the quotient X(i) by E(i) can assume a value close to 1 also at those operating points in which no object generating echoes is present in the environment. One cause lies, on obtaining the quotient, in the fact that the denominator is almost 0, based on the expected meaningful values.

It is therefore an object of the present invention to perform a relative assessment of a received signal of an acoustic environment detection with regard to its similarity with an expected signal profile, in which the aforementioned disadvantages are reduced or eliminated.

Disclosure of the invention

The aforementioned object is achieved according to the invention by a device for echo-based environment sensor systems and a means of locomotion having such a device. The device comprises a signal transducer, which can be formed, for example, as a sound transducer, preferably as an ultrasonic transducer. Additionally, the device comprises an evaluating unit which is configured to determine a similarity between a signal coming from the signal transducer and an expected signal. The signal coming from the signal transducer is -as stated at the outset-an additive superposition of noise and possibly signals received from the environment. Depending on the operating instant, a signal impressed by means of the device may also be present at the transducer. Likewise depending on the operating instant, the expected signal therefore corresponds, for example, to the impressed signal and/or an echo signal received from the environment. The similarity can be determined, for example, as a cross-correlation.

Additionally, the evaluating unit is configured to determine a relative measure of the similarity by the similarity being referred to a predefined reference quantity for the similarity. The predefined reference quantity can be defined, for example, in dependence on the propagation time. The aforementioned method is therefore suitable for carrying out a norinalisation, for example, to a signal total energy for external interferer suppression and a weighting by means of a suitably defined threshold for reflex strength recognition (echo strength recognition) . Echoes in the transducer signal are therefore more reliably recognised and erroneous recognitions of echoes are avoided.

The subclaims show preferred developments of the invention.

The predefined reference quantity can vary over time, in particular in dependence on an instant within a working cycle of the device. A working cycle is understood as the period of time between an emitting of a test signal and a receiving of all corresponding echoes. This enables a normalisation of the transducer signal in dependence on the propagation time and thus a low error recognition rate.

The evaluating unit can further be configured to determine an instantaneous signal energy of the signal coming from the signal transducer and to take it into account in the definition of the predefined reference quantity. By entering the total signal energy of the signal received by the transducer into the predefined reference quantity for normalising the transducer signal, a division by particularly small values (e.g. by disregarding a noise component) is avoided. A normalisation to very small values, as has sometimes been performed in the prior art, results in an increased risk of a recognition of a signal without an environment echo actually being present in the signal. This risk is therefore reduced according to the invention.

To determine the predefined reference quantity, preferably a maximum-value function can be used. For example, the signal energy coming from the signal transducer and a signal energy of a reference echo signal can be entered into the maximum-value function. The reference echo signal can, for example, be stored and read out. The maximum-value function outputs, for each instant in the working cycle of the device, the greatest value resulting when considering in parallel the signal energy coming from the transducer and the signal energy of the reference echo signal. If the actual transducer signal at an instant considered has very low values, a norrnalisation to such a low value is thus avoided by normalising to the signal energy of the reference echo signal instead. An unexpectedly or unsuitably high output value of the normalisation, which can result in recognition of an echo not present, can thus be effectively prevented by the present invention.

In the maximum-value function, a noise signal existing between the transducer and the evaluating unit of the device according to the invention can further also be taken into account. This signal can be defined or measured in dependence on the propagation time, but preferably taken into account as constant in time. On adopting a noise signal constant in time, the taking account of the noise in the normalisation is computationally simplified.

To take into account a noise between signal transducer and evaluating unit, the evaluating unit can be further configured to determine a noise signal in time segments or operating states without internal excitation and without expected external excitation of the signal transducer (e.g. by echoes) . It should be possible to assume with high probability here that the device according to the invention is, so to speak, idling and no additional signals are superposed on the measured noise signal. In this way, a currently actual noise behaviour of the transmission strength can be optimally taken into account.

Preferably, the evaluating unit can be configured to determine a relative measure of the similarity over time by referring, i.e. normalising, the similarity to the predefined reference quantity. The reference quantity is obtained here with the aid of at least one previously determined reciprocal value of at least one of the quantities from the group "reference quantity" and "similarity" and "noise". A further processing of the at least one previously specified reciprocal value can be modified, for example, by a minimal function. Here, for example, the minimum of the aforementioned three reciprocal values to a reference quantity can be determined over time and subsequently the relative measure of the similarity can be determined as a product of the minimal function and the similarity. The reciprocal value determination here enables a division with markedly less complexity than an actual computational division. In particular, in the case of (constant) quantities used over a longer period of time, a

B

single reciprocal value determination enables performance savings. Of course, the relative measure of the similarity over time can also be determined solely from a reciprocal value of the predefined reference quantity which is multiplied by the similarity. Both quantities can be time-dependent, while the predefined reference quantity can also be assumed and defined as time-invariant. In this way, a determination of a minimum of different quantities can be omitted, which additionally reduces the computational complexity.

According to a second aspect of the present invention, there is proposed a means of locomotion having a device as has been described in detail above in connection with the first-mentioned aspect of the invention. Here, a signal transducer or a plurality of signal transducers can be formed as ultrasonic sensors and, for example, arranged in the front and/or rear aprons and/or in bumpers of the means of locomotion. Alternatively or additionally, the signal transducers can be accommodated in bumpers or other suitable body components. The evaluating unit can be realised as an electrical control device which can also be used for other purposes. For example, a microprocessor, a nanocontroller or the like can be configured to carry out the steps as carried out by the device presented above. The features, feature combination and the advantages resulting therefrom correspond in such an apparent manner to those stated in connection with the first-mentioned aspect of the invention that renewed discussion of the same can be dispensed with to avoid repetitions.

Brief description of the drawings

Exemplary embodiments of the invention are described in detail below with reference to the accompanying drawings, in which: Figure 1 is a basic circuit diagram of a first exemplary embodiment of a device according to the invention; Figure 2 is a basic circuit diagram of an alternative exemplary embodiment of a device according to the invention; Figure 3 is a basic circuit diagram of a further alternative exemplary embodiment of a device according to the invention; and Figure 4 is a schematic view of an exemplary embodiment of a means of locomotion according to the invention.

Embodiments of the invention Figure 1 shows a first exemplary embodiment of a device 1 according to the invention. Herein, an ultrasonic transducer 2 is configured as a signal transducer to pick up ultrasonic signals 4 from the environment and electrically convert them, in conjunction with expected signals emitted in particular by the device 1, in order to recognise environment objects. The output signal e(t) of the ultrasonic transducer 2 is supplied to a filter 5. The filter 5 additionally receives an input which corresponds to an expected signal s(i). In other words, s(i) can represent an echo expected at a specific instant, so that s(i) can be varied in dependence on the propagation time.

The filter 5 outputs a similarity X(r), which is supplied to a nornialising unit 9. In the device 1 there is additionally provided a minimum-dimension generator B which provides a predefined reference quantity X1 for the normalising unit 9. Within the normalising unit 9, the similarity X(i) is referred to the predefined reference quantity Xijm. For this purpose, firstly the reciprocal value of the predefined reference quantity Xnm can be obtained and the latter can subsequently be multiplied by the similarity X(t) . The output signal KLr) of the normalising unit 9 now represents a relative measure of the similarity X(r) . The output signal K(t) can be used for subsequent detection of echoes in the ultrasonic signal 4, it being possible to suppress interfering signals better than in the prior art and to avoid errors in the echo recognition.

Figure 2 shows an alternative exemplary embodiment of a device 1 according to the present invention. In this exemplary embodiment too, the ultrasonic transducer 2 is configured to receive an ultrasonic signal 4 from the environment and to supply an output signal e(t) to a filter 5, which additionally receives a signal s(i) as expected signal form. The output signal X(i) of the filter (the "similarity") is supplied to the normalising unit 9. In correspondence with the exemplary embodiment shown in Figure 1, the normalising unit 9 additionally receives the predefined reference quantity Xnm, which is provided by the minimum-dimension generator B. In addition to the variant shown in Figure 1, the energy EVt) of the output signal e(T) of the signal transducer 2 is determined in an energy determining unit 6 and combined with the predefined reference quantity X11 in a combining unit 7. The output signal ALt) of the combining unit 7 is taken into account as an additional input quantity within the normalising unit 9 when generating the relative measure K(t) of the similarity XLt) . The combining unit 7 can, for example, perform a maximal-value function, so that its output signal A(t) is in each case the, a considered instant, higher value of the functions E(t) and Xnm. This enables a greater adaptation of the normalisation to actual transducer signal energies.

Figure 3 shows a further alternative to the embodiments of the device 1 according to the invention which are shown in Figure 1 and Figure 2, respectively. In addition to the variant shown in Figure 2, a noise signal N0 is provided by a noise signal generator 10 for taking into account in the combining unit 7. The output signal ALt) of the latter thus now takes into account also the noise in the output signal e(i) of the ultrasonic transducer 2 (for example in the course of a maximum-value determination) -For this purpose, the noise signal generator 10 can furthermore be configured to receive the output signal e(T) of the ultrasonic transducer 2 as an input quantity. For example, in an operating state in which neither an internal nor an external excitation of the ultrasonic transducer 2 is to be expected ("idling"), a noise signal can be measured or the output signal e(t) of the ultrasonic transducer 2 can be assumed as a noise signal. The output signal N0 of the noise signal generator 10 can therefore exhibit both time dependence and time invariance.

Figure 4 shows a passenger car 12 as a means of locomotion, in the bumpers 13 of which respective ultrasonic transducers 2 are arranged. The ultrasonic transducers 2 are connected to a respective evaluating unit 3 via signal lines. In this way, the passenger car 12 is configured to realise the advantages of the present invention, for example, in association with a parking assistance system.

With the present invention, a dynamic reduction in the processing of environment signals can be achieved by normalisation in dependence on propagation time. Object information is in this case preserved, since not exclusively the signal energy is evaluated. Furthermore, the present invention provides external interferer suppression, which has the result that, in the event of a poor signal-to-noise ratio, the output signal of the algorithm according to the invention does not assume an excessive or computationally excessive value. Through the combination of the dynamic reduction with the external interferer suppression, the present invention avoids a division by 0 or by very small values oh the evaluation of environment echoes, and this increases the recognition rate and the error recognition rate.

Even though the inventive aspects and advantageous embodiments have been described in detail with the aid of the exemplary embodiments explained in conjunction with the appended drawing figures, modifications and combinations of features of the exemplary embodiments set out are possible for a person skilled in the art, without departing from the scope of the present invention, the scope of protection of which is defined by the appended claims.