Method for detecting obstacles by reflecting ultrasonic waves
By transforming the echo segment of the ultrasonic receiving signal from the time domain to the frequency domain, and using spectrum characteristics to distinguish obstacle types, the problem of inability to distinguish different obstacle types in the prior art is solved, and obstacles in the vehicle parking auxiliary system are realized Accuracy and reliability of object type identification.
Patent Information
- Application Number
- CN201910477232.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2018-06-05
- Filing Date
- 2019-06-03
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2039-06-03
AI Technical Summary
Existing reflective ultrasonic detection methods cannot effectively distinguish different types of obstacles, especially in vehicle parking assist applications, and it is impossible to identify different obstacle types through static measurements to distinguish the process of echo signals.
By transforming the echo segment of the ultrasonic received signal from the time domain to the frequency domain, checking the spectral characteristics in the spectrum, such as the spectrum width and the center of gravity, to distinguish different obstacle types and assigning the echo segment to a predetermined obstacle type . Use ultrasonic burst transmission signals and Fourier transform or other signal processing algorithms to determine obstacle types in combination with pattern recognition processes.
Accurate distinction between different obstacle types, such as the identification of stoppers, curbs, walls and benchmarks, improving the accuracy and reliability of vehicle parking assist systems.
Smart Images

Figure CN110568444B_ABST
Abstract
Description
[0001] The present invention claims priority from European Patent Application No. 18176088.5 filed on June 5, 2018, the contents of which are incorporated herein by reference. Technical Field
[0002] The invention relates to a method for detecting obstacles by reflecting ultrasound, wherein the method is used in particular as a parking aid for a vehicle. The invention relates in particular to a method for detecting obstacles, in particular in the vicinity of a vehicle. Background Art
[0003] Methods for detecting obstacles, especially those located near a vehicle, by analyzing received echo signals of previously emitted ultrasonic waves are generally known. Such methods have proven to be effective for automotive applications, for example as parking aids. So far, in most cases, the time course of the echo signal has been analyzed. Characteristic time courses of the time course of the echo signal, such as more or less distinct maximum values and the time course of the echo signal before and after the maximum value, are considered obstacles.
[0004] At present, it is not possible to distinguish between different types of obstacles by static measurements. The course of the echo signal can only be tracked for the echo signal segments that can be traced back to the obstacles. The fact that it is meaningful to distinguish between different types of obstacles is illustrated by the example of stoppers, which are protrusions that are particularly used in parking lots to signal to the user how far he / she should drive into the parking space when parking (that is, until the front or rear wheels touch the stoppers). During the parking process, such obstacles initially show a relatively large echo, but the size of the echo decreases strongly as you get closer. It is also possible to drive past protrusions that also include curbstones. However, the obstacle strength that can be traced back to higher obstacles such as walls or vehicles always remains essentially the same. This also applies to piles and poles.
[0005] Methods for using reflected ultrasound are known, for example, from DE-A-100 27 828 and DE-A-198 46241. Summary of the invention
[0006] The object of the present invention is to provide a method for detecting obstacles by reflecting ultrasound, by which method different obstacle types can be distinguished.
[0007] In order to solve this object, the present application proposes a method for detecting obstacles by reflecting ultrasonic waves, in particular as a parking aid for a vehicle, wherein the method comprises:
[0008] -Using an ultrasonic transmitter, an ultrasonic pulse train transmission signal is transmitted to the detection area to be observed;
[0009] - receiving, by means of an ultrasonic receiver, an ultrasonic signal reflected by an obstacle in the detection area as an ultrasonic reception signal;
[0010] - detecting at least one echo in the ultrasonic receive signal generated by an obstacle;
[0011] - transforming the echo segment of the ultrasound reception signal belonging to the echo from the time domain to the frequency domain;
[0012] - checking the frequency spectrum of the echo segment for the presence of at least one spectral feature of a plurality of predetermined spectral features, wherein each spectral feature is indicative of a predetermined obstacle type and possibly of at least one of a plurality of predetermined obstacle types; and
[0013] - Assigning echo segments to predefined obstacle types.
[0014] Accordingly, the invention proposes the use of the knowledge of acoustically imaging obstacles by sending ultrasonic waves and receiving reflections of these ultrasonic waves in order to be able to distinguish between different obstacles, in particular different obstacles that are located near the vehicle during the parking process. This is advantageous insofar as, for example, stoppers or other types of protrusions are to be distinguished from, for example, walls as obstacles, since, unlike walls, protrusions or curbs can be "driven over" during the parking process; this does not apply, for example, to nearby parked vehicles. The problem is that it is not possible or even possible to reliably distinguish between different obstacle types, for example during the parking process, by means of the time course of the echo signal or the envelope signal.
[0015] However, surprisingly, it is possible to distinguish obstacles if the echo segments of the ultrasound receive signal are examined in the frequency domain. As was determined during the development of the invention, the obstacle type can be verified by determining the spectral moments such as the spectral width and spectral center of gravity of the echo segment spectrum.
[0016] In the method according to the invention, an ultrasonic transmitter sends an ultrasonic pulse train transmission signal to a neighboring detection area. If an obstacle exists in the detection area, the ultrasonic receiver receives the reflected ultrasonic signal from the detection area as an ultrasonic reception signal. At least one echo segment that can be traced back to the obstacle can be detected by the ultrasonic reception signal through its time course. This is performed, for example, by comparing the signal size of the ultrasonic reception signal with a fixed or dynamically adjustable threshold. According to the invention, the detected echo segment is transformed from the time domain to the frequency domain. The frequency spectrum of the echo segment is checked for the presence of at least one spectral feature of a plurality of predetermined spectral features, wherein each spectral feature represents a predetermined obstacle type. However, it is also possible that each spectral feature represents more than a plurality of predetermined obstacle types. Then, the echo segment is assigned to a predetermined obstacle type, wherein the time point at which the echo segment appears in the ultrasonic reception signal represents the distance of the obstacle to the ultrasonic receiver. According to the invention, the ultrasonic transmitter and the ultrasonic receiver can be two different units or a common unit, namely a so-called ultrasonic transducer.
[0017] In an advantageous further embodiment of the invention, a degree of probability of the assignment of an echo segment to an obstacle type can be specified, or several degrees of probability of the assignment of an echo segment to different obstacle types can be specified. In this further embodiment of the invention, the additional information (i.e. the probability or probability density distribution) indicates the reliability with which the obstacle type or, if applicable, different obstacle types can be assigned to the detected echo segment.
[0018] It may further be advantageous that the detected obstacle is signaled optically and / or acoustically and / or tactilely (e.g., a seat or steering wheel of a vehicle), wherein the signal is different for different predetermined obstacle types. In this way, the driver parking his vehicle is signaled the type of obstacle.
[0019] As mentioned above, the spectrum characteristics include the spectrum width and the spectrum center of gravity of the echo section. These two spectrum characteristics can be determined in the same way according to the I signal part and the Q signal part of the I / Q modulation. In the process of the present invention, it has been shown that three different types of obstacles can be distinguished by the spectrum characteristics. Different types of distinguishable obstacles include curbs, stoppers or other protrusions as the first type, walls and in particular vehicles as the second type, and posts or piles such as for information and traffic signs, signal lights or street lights as the third type. It has been shown that the protrusions can be clearly distinguished from walls and posts by the spectrum center of gravity of the echo section. However, the distinction between the wall and the post by the spectrum center of gravity may not be clear. However, in contrast, if the echo section is checked for its spectrum width, the wall can be clearly distinguished from the protrusion and the post. However, it is almost impossible to distinguish protrusions and posts by the spectrum width. Therefore, if the spectral features are used to distinguish the above three obstacle types, this is highly likely because the three obstacle types "distribute" themselves in the 2-dimensional diagram in a way that the spectral width is plotted on one axis and the spectral center of gravity is plotted on the other axis over clearly separated areas.
[0020] Unfortunately, ultrasonic receivers used today in automotive applications are very temperature dependent. Therefore, if ultrasonic receivers are used whose sensitivity is relatively strongly dependent on temperature, it is advantageous according to a further advantageous embodiment of the invention if the detection of the echo segments and their transformation into the frequency domain are temperature compensated or temperature corrected or are otherwise effected by taking into account the current temperature of the ultrasonic receiver. It may be advantageous to determine the temperature of the ultrasonic receiver by measuring or by means of a temperature-dependent signal characteristic, to calibrate the ultrasonic receiver with respect to the temperature-dependent signal characteristic and to check the ultrasonic receive signal with respect to it. In this context, reference is made to German patent application DE102018102535.1 with a filing date of February 5, 2018, the disclosure of which is incorporated by reference.
[0021] It has proven to be particularly advantageous to use short pulse signals as ultrasonic transmission signals for implementing the invention. Thus, the largest spectral range is covered. However, the energy of a single pulse or a few pulses is too low to detect sufficiently reliable echo signal segments even at a distance of a few meters relevant for the parking process. Therefore, it is advantageous to use a pulse train transmission signal instead of a single pulse. Here, it has proven to be advantageous if the ultrasonic pulse train transmission signal has one to twenty or five to fifteen or eight to twelve pulses.
[0022] In a further advantageous embodiment of the invention, it can be provided that the transformation into the frequency domain is carried out by means of a Fourier transformation, for example an FFT, or without a Fourier transformation by means of an algorithm known in particular from signal processing of weather radars (see for example Doviak, Richard J. / S. (1993): "Doppler Radar and Weather Observations" (2nd ed.), pp. 122-159), such as autocovariance processes.
[0023] As mentioned above, different obstacle types are identified by determining spectral features. By known classifiers, such as Gaussian naive Bayes classifier, nearest centroid classifier, k nearest neighbor classifier (see Duda, RO / Hart, PE / Stork, DG (2001): "Pattern Classification", Wiley, 654 pages), or other known pattern recognition processes (Haykin, S. (1998): "Neural Networks: A Comprehensive Foundation", Prentice Hall, 842 pages; Hastie, T. / Tibshirani, R. / Friedman, J. (2001): "The Elements of Statistical Learning", Springer, 533 pages; Bishop, CM (2005): "Pattern Recognition and Machine Learning", Springer, 738 pages; Rojas, R. (1996): "Theorie der NeuronalenNetze (neural network theory), Springer, about 300 pages), can ultimately determine the obstacle type and accordingly specify the probability that the obstacle can be assigned to one of a plurality of predetermined obstacle types. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] Hereinafter, the present invention will be described in detail with reference to the accompanying drawings. The accompanying drawings are shown as follows:
[0025] Figure 1 The time course of the ultrasonic reception signal is shown, wherein there are different echo sections due to a protrusion about 80 cm (centimeter) away from the receiver, different echo sections due to a pile about 120 cm away from the ultrasonic receiver, and different echo sections due to a wall about 150 cm away from the ultrasonic receiver;
[0026] Figure 2 The time course of the extracted echo segments is shown;
[0027] Figure 3 is a diagram in which the spectral centroids of the extracted echo segments are plotted for the obstacles "bump", "wall" and "pile" when measuring obstacles located at different distances from the ultrasound receiver;
[0028] Figure 4 is a graph in which the spectrum widths of the extracted echo segments are plotted for obstacles "bump", "wall" and "pile" when measuring obstacles located at different distances from an ultrasonic receiver; and
[0029] Figure 5 Yes Figure 3 and Figure 4 The figure of FIG serves as a summary of the 2D maps of the above obstacles to be tested during the test, wherein it is clear that each obstacle type can be clearly distinguished by the position of the two above spectral features in the 2D map. DETAILED DESCRIPTION
[0030] As mentioned above, laboratory tests were performed within the scope of the present invention, in which an ultrasonic pulse train transmission signal, for example with 8 pulses, was sent by means of an ultrasonic transducer into a detection area in which a pile (a 75 mm (millimeter) pipe with a height of 1 m (meter)), a simulated standard curbstone with a length of 1 m and aligned at right angles to the propagation direction of the ultrasonic wave, and a wall were arranged. Measurements were made for this scenario, whereby, for example, the time course of the ultrasonic reception signal was calculated according to Figure 1 To determine. Figure 1 In FIG, the continuous line represents the ultrasonic receiving signal, and the dotted line represents the threshold signal. Three characteristic echo sections can be seen, namely, for the protrusion (standard curb), for the pile and for the wall.
[0031] Figure 2The echo section extracted from the ultrasonic receiving signal is shown as an example. In terms of using eight pulses as the ultrasonic pulse train transmission signal for testing, it has been proved that this is a good compromise. The echo area is detected by the echo maximum. Multiple samples are selected as the left and right sides of the maximum value (the exemplary 8 samples after the echo maximum and the exemplary 8 samples before the echo maximum). Then, the spectral center of gravity and the spectral width of the extracted echo area are determined. According to the literature (for example, as cited in: Keeler, RJ / Passarelli, RE (1990): Signal Processing for Atmospheric Radars (signal processing for atmospheric radar). Cited from: Atlas, D. (eds.): "Radar in Meteorology (radar in meteorology)", American Meteorological Society (American Meteorological Society), Boston, Massachusetts), the formula correlation for the spectral center of gravity and spectral width is known as follows:
[0032] 1. Classification number (spectrum center of gravity)
[0033]
[0034]
[0035] 2. Classification number (spectrum width)
[0036]
[0037] Figure 3 and Figure 4 Shown are the "distributions" of different measurement results after calculating the spectral centroid and spectral width of the echo signal segments for stoppers, piles and walls. Figure 3 It is shown that the stoppers can be clearly distinguished from walls and piles by the spectral centroid, whereas the two obstacle types walls and piles cannot be clearly distinguished from each other. Figure 4 It is shown that the spectrum width can be used to clearly distinguish the wall as an obstacle from the stoppers and the piles, while the stoppers and the piles cannot be clearly distinguished from each other. Figure 5 Finally it is shown that all three obstacle types can be well distinguished from each other when both spectral parameters are considered.
[0038] Thus, once these preliminary checks have been carried out and field trials have subsequently been carried out, it can be seen that determining the two spectral features "centre of gravity" and "width" makes it possible to assign an obstacle to one of the three above-mentioned types. The actual measurement points of the spectral centroid and the spectral width from the echo segment to be measured are finally assigned to one of the three obstacle categories by means of classifiers known in the prior art. It is then also possible to specify the probability with which the currently measured obstacle can be assigned to one of the three categories. Alternatively, it is also possible to specify several probabilities in order to specify the probability of the assignment of an obstacle to several categories.
Claims
1. A method for detecting obstacles by reflecting ultrasonic waves, the method comprising: - transmitting an ultrasonic pulse train signal to a detection area to be observed through an ultrasonic transmitter; - receiving, by an ultrasonic receiver, an ultrasonic signal reflected by an obstacle in the detection area as an ultrasonic reception signal; - detecting at least one echo signal segment in the ultrasound reception signal generated by the obstacle which can be traced back to the obstacle by comparing the signal amplitude with a threshold value, wherein the threshold value varies over the time course of the ultrasound reception signal; - extracting the detected echo signal segment from the ultrasound reception signal; - transforming the extracted echo signal segments of the ultrasound reception signal belonging to the echo from the time domain to the frequency domain to generate a frequency spectrum of the detected echo signal segments; - checking the frequency spectrum of the extracted echo signal segment for the presence of at least two spectral features out of a plurality of predetermined spectral features, wherein each spectral feature is indicative of a predetermined obstacle type; as well as - assigning the extracted echo signal segment of the ultrasound reception signal to one of three predetermined obstacle types based on an inspection of the spectrum of the extracted echo signal segment, wherein the at least two spectrum features include a spectrum centroid and a spectrum width of the extracted echo signal segment.
2. The method according to claim 1, characterized in that The method is used as a parking aid for a vehicle.
3. The method according to claim 1, characterized in that Each spectral signature represents at least one obstacle type from among a plurality of predetermined obstacle types.
4. The method according to claim 1, characterized in that: The probability that the extracted echo signal segment can be assigned to an obstacle type is specified, or a plurality of probabilities that the extracted echo signal segment can be assigned to different obstacle types are specified.
5. The method according to claim 1, characterized in that The detected obstacles are signaled optically and / or acoustically and / or haptically, and in that the signal is different for predetermined obstacle types.
6. The method according to any one of claims 1 to 5, characterized in that The spectrum features include the spectrum center and spectrum width of the spectrum of the extracted echo signal segment.
7. The method according to any one of claims 1 to 5, characterized in that The different types of obstacles that can be distinguished include curbs and bumps as a first type, walls and vehicles as a second type, and poles and stakes as a third type.
8. The method according to claim 7, characterized in that The posts are used for information and traffic signs, traffic lights or street lamps.
9. The method according to any one of claims 1 to 5, characterized in that The sensitivity of the ultrasound receiver is temperature dependent and in that the detection of the extracted echo signal segments and the subsequent transformation of the extracted echo signal segments into the frequency domain are temperature compensated or temperature corrected or are achieved by taking into account the current temperature of the ultrasound receiver.
10. The method according to claim 9, characterized in that The temperature of the ultrasound receiver is determined by measurement or by a temperature-dependent signal characteristic, with respect to which the ultrasound receiver is calibrated and with respect to which the ultrasound receive signal is checked.
11. The method according to any one of claims 1 to 5, characterized in that The ultrasonic pulse train transmission signal includes one to twenty or five to fifteen or eight to twelve pulses.
12. The method according to any one of claims 1 to 5, characterized in that The transformation into the frequency domain is performed by a Fourier transform or, without performing a Fourier transform, by an algorithm known from signal processing of weather radars.
13. The method according to claim 12, characterized in that The Fourier transform includes FFT.
14. The method according to claim 12, characterized in that Algorithms known from signal processing of weather radars include autocovariance processes.
Citation Information
Patent Citations
Temperature measurement using the impedance of an ultrasonic transducer
DE102018102535B3
Method and device for moisture measurement in masonry materials with ultrasonic signal analysis
DE19846241A1
Method for recognition of an obstacle with the aid of reflected ultrasonic waves
EP3579020A1
Method and apparatus for identifying and classifying objects
CN101189533B
Distance sensor and method for determining a distance
CN101779142A
