DE102019107396A1 - Detection and classification of raised road markings using LIDAR - Google Patents

Abstract

The present invention relates to a method for the method of detecting and classifying raised lane markings (1) on a road (2) using a LIDAR detector with the following steps: - Detection of environmental information about the road by at least one LIDAR detector (100) wherein the LIDAR detector generates sampling points by processing the received light signals and converting the light signals into electrical signals which are the sampling points; - Contouring the segmented sampling points (300), which can be related to a raised road marking (1), into a feature shape, a bounding frame being generated based on the feature shape (330); and - classifying the bounding box (400) by comparing its segmented sample points with reference sample points of bounding boxes stored in memory.

Description

The present invention relates to a method for detecting and classifying raised lane markings on a road using a LIDAR detector of a vehicle.

The present invention also provides a computer program which contains instructions which, when the program are executed by a computer, cause the computer to carry out the steps of the method according to the invention or according to a modified embodiment of the invention.

A computer program is a collection of instructions for performing a particular task, designed to solve a particular class of problems. A program's instructions are designed to be carried out by a computer and it requires a computer to be able to run programs in order for it to function.

The present invention also provides a computer-readable storage medium which contains instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method according to the invention or according to a modified embodiment of the invention.

The present invention also provides a system which has a device for carrying out the steps of the method according to the invention or according to a modified embodiment of the invention.

The present invention also specifies a vehicle which has the system according to the invention or a system according to a modified embodiment of the invention.

The vehicle is preferably the driver's ego vehicle.

A raised lane marker is a safety device used on roads. These devices are typically made of plastic, ceramic, thermoplastic paint, glass, or occasionally metal, and come in a variety of shapes and colors. Raised reflective markings, e.g. made of plastic, ceramic or metal, have a lens or layer that improves their visibility by reflecting back car headlight light, while lane marker buttons made of glass collect car headlight light through a dome shape and reflect the light through a reflective layer disposed therein. Some other names for certain types of raised lane markings are convex vibration lines, botts points, delineator posts, cat's eyes, lane marker buttons, or street turtles. Sometimes they are simply referred to as "reflectors". Botts points are round, non-reflective raised road markings. Raised lane markings give drivers tactile and acoustic feedback when driving over designated lanes.

Detecting raised lane markings is more difficult for sensor systems than detecting colored lane markings because the raised lane markings are small, come in different shapes and are difficult to detect using a camera-based system because the recorded image of the road surface contains a lot of noise. Another problem is that a camera will perform poorly at night, heavy rain, snow, and fog. Furthermore, known image processing algorithms are prone to errors. This means that signals can be mistakenly interpreted as shadows and potholes. It may take several filter stages to detect the raised road markings as road boundaries.

LIDAR detectors are used to detect objects in an environment according to a so-called “transit time” principle. Individual laser pulses are emitted by a laser beam device, and reflections of these laser pulses on objects in the vicinity are received and evaluated as optical signals by a receiving unit. The distance to the respective object can be determined from a time difference between the emission of a laser pulse and the reception of its reflection on objects in the vicinity. In addition, an intensity and a position of the received reflection can also be determined in order to obtain additional information about the properties of the object. Typically, LIDAR uses ultraviolet, visible, or near-infrared light to image objects. LIDAR can be used in conjunction with a wide variety of materials, such as non-metallic objects, metallic objects, ceramic objects and chemical compounds. A narrow laser beam can have physical characteristics map very high resolutions, for example an aircraft can map terrain with a resolution of 30 cm or better. The wavelengths vary according to the target from about 10 µm to the UV range (about 250 nm). Unlike the pure reflection that occurs with a mirror, light is usually reflected by backscattering. Different types of scattering are used for different LIDAR applications: most commonly Rayleigh scattering, Mie scattering, Raman scattering, and fluorescence.

In the technical field of autonomous driving, cameras in combination with LIDAR detectors for lane detection are, for example, from the disclosure of the document US 9,530,062 B2 known. This document discloses the use of a camera and an image processing module in combination with a lidar detector and a lidar processing module. The camera and LIDAR detector work together to improve the performance of the lane boundary detection system. According to this document, the camera and the image processing module are able to detect both non-reflective raised lane markings and reflective raised lane markings over a certain area in front of the vehicle, for example 25 m or 30 m. Furthermore, the LIDAR detector, as configured in the raised lane markings detection system, is only able to detect reflective raised lane markings and is not able to detect non-reflective markings such as Bott points. This means that the system remains prone to errors if the raised road markings are more than 25 m or 30 m away. In addition, no further information is available about the properties of the raised pavement markings.

Nevertheless, lane detection is a critical function in a transport vehicle that has an autonomous driving functionality, one example is a lane departure warning. In particular, raised lane markings and colored lane markings, mostly white or yellow, are known as lane boundaries. Depending on the shape of the raised road markings, various properties or information can be displayed. This could be evaluated and used by a suitably trained system of a vehicle. For example, raised lane markings can indicate uninterrupted sharp curves, sloping sections, access ramps or pedestrian crossings. It is very important for autonomous vehicles to recognize such displays and use them correctly in autonomous driving. A critical scenario would also be drifting into the wrong lane on a single lane.

The present invention is based on the object of increasing the safety of autonomous vehicles or of vehicles with a driving assistance system.

This problem is solved by the independent claims. Advantageous refinements are given in the dependent claims.

• Erfassen von Umgebungsinformation über die Straße durch mindestens einen LIDAR-Detektor, wobei erfasste optische Signale verarbeitet und in elektrische Signale umgewandelt werden, die als Abtastpunkte bereitgestellt werden;
• Segmentieren der Abtastpunkte in benachbarte, segmentierte Abtastpunkte, wobei mindestens einige der segmentierten Abtastpunkte aufgrund ihrer Erfassungsposition mit einer erhabenen Fahrbahnmarkierung in Beziehung gesetzt werden können;
• Konturieren der segmentierten Abtastpunkte, die mit einer erhabenen Fahrbahnmarkierung in Beziehung gesetzt werden können, in eine Merkmalsform, wobei basierend auf der Merkmalsform ein Begrenzungsrahmen erzeugt wird;
• Klassifizieren des Begrenzungsrahmens durch Vergleichen seiner segmentierten Abtastpunkte mit Referenzabtastpunkten von Begrenzungsrahmen, die in einem Speicher gespeichert sind.

In particular, the present invention specifies a method for detecting and classifying raised lane markings on a road using a LIDAR detector of a vehicle, comprising the steps:

• Detection of environmental information about the road by at least one LIDAR detector, the detected optical signals being processed and converted into electrical signals which are provided as sampling points;
• Segmenting the sampling points into adjacent, segmented sampling points, wherein at least some of the segmented sampling points can be related to a raised road marking based on their detection position;
• Contouring the segmented sampling points, which can be related to a raised lane marking, into a feature shape, a bounding box being generated based on the feature shape;
• Classifying the bounding box by comparing its segmented sample points with reference sample points of bounding boxes stored in memory.

The basic idea of the invention is not only to detect the raised lane markings but also to classify them so that any implicit information can be evaluated by the vehicle as a result of their design. The exclusive use of LIDAR has the advantage that it can be used to detect and classify the raised road markings through a more comprehensive evaluation and use of the sampling points. This classification can also be done in bad weather with a high efficiency, for example in weak light or fog. Furthermore, the Classification of the raised road markings in the range of> 100 meters with a high efficiency.

The difference between detecting and classifying lies in the fact that when detecting, the raised lane marking is only detected. On the other hand, the exact type is identified during classification so that information linked to the type can also be recognized. However, this classification requires an evaluation of the sampling points according to the invention in order to increase the safety of the driver.

The step of capturing environmental information about the road by at least one LIDAR detector, wherein captured optical signals are processed and converted into electrical signals that are provided as sampling points, is the first step of the method according to the invention. The term environmental information preferably includes mainly optical signals. Preferably, a unit of the LIDAR detector processes received light signals which are optical signals which are converted into electrical signals which form the sampling points. The processing of the optical signals can be carried out in the LIDAR detector or in another processing unit.

The segmentation of the sampling points into contiguous, segmented sampling points preferably means that the sampling points are related to one another based on their detection position. At least some of the segmented sampling points can be related to a raised road marking based on their detection position. This can be identified by checking the height of the segmented sample points, especially when the sample points are above or on a ground plane. In other words, it does this by differentiating sample points obtained from ground and non-ground sample points.

The next step is the contouring of the segmented sampling points, which can be related to a raised road marking, into a feature shape, a bounding frame being generated based on the feature shape. The contour is preferably built up on the basis of the weighted average of all sampling points with a specific angular sector. The segmented sampling points can be contoured taking into account the number of segments, the height, the width and the length. This data is used to generate other features, i.e. create a bounding box. The bounding box is a result of presumably logically related sampling points and can be examined in a next step in which the bounding box is classified in the further process.

In the step of classifying the bounding box, the dimensions of the object, such as the height, the length, the width, the echo pulse width and the state of movement of the object, can be taken into account. This data is used to classify the object as a raised lane marking, in particular the type of raised lane marking.

The idea makes it possible to detect and classify raised road markings from a distance of more than 100 meters using only the echo pulse width and the scanning point positions of LIDAR signals. The idea also enables the acquisition of patterns of Botts points and / or raised road markings for e.g. Speed bumps or downhill sections.

According to a modified embodiment of the invention, the method is characterized in that a field-programmable array unit reads the electrical signals and converts them into scanning points that contain information, in particular an echo pulse width, a position and / or an angle of the scanning points. This information can be passed on to the segmentation step. This has been found to be an efficient way of providing the sample points. This ensures fast data processing so that the driver's safety is not impaired by lengthy processing steps.

• - Schätzen einer Bodenebene von den Abtastpunkten;
• - Prüfen, ob Abtastpunkte auf oder über der Bodenebene angeordnet sind;
• - Erzeugen eines Bodenebenensegments, falls die Punkte auf der Bodenebene angeordnet sind;
• - Erzeugen eines Nicht-Bodenebenensegments und Schätzen seiner Höhe und Breite, falls die Punkte oberhalb der Bodenebene angeordnet sind;
• - Berechnen der mittleren Echopulsbreite jedes Nicht-Bodenebenensegments;
• - Prüfen, ob die mittlere Echopulsbreite eines Nicht-Bodenebenensegments eine definierte Höhe und eine definierte Breite überschreitet;
• - Bestimmen, dass das Nicht-Bodenebenensegments keine erhabene Fahrbahnmarkierung ist, falls die Echopulsbreite des Nicht-Bodenebenensegments die definierte Höhe und die definierte Breite nicht überschreitet;
• - Bestimmen des Nicht-Bodenebenensegments als wahrscheinliche erhabene Fahrbahnmarkierung, falls die Echopulsbreite des Nicht-Bodenebenensegments die definierte Höhe und die definierte Breite überschreitet. Diese Konkretisierung des Verfahrensschritts ermöglicht eine effiziente Beurteilung des Vorhandenseins einer erhabenen Fahrbahnmarkierung. Das Berechnen der mittleren Echopulsbreite jedes Nicht-Bodenebenensegments wird ausgeführt durch Bestimmen der Echopulsbreite jedes Abtastpunkts und Berechnen eines Mittelwerts für alle Abtastpunkte im Segment.

According to a modified embodiment of the invention, the method is characterized in that the segmentation of the sampling points into contiguous sampling points is carried out at least by:

• - estimating a ground plane from the sampling points;
• - Check whether sampling points are arranged on or above the ground plane;
• Generating a ground plane segment if the points are arranged on the ground plane;
• - Creating a non-ground plane segment and estimating its height and width if the points are above the ground plane;
• - calculating the mean echo pulse width of each non-ground plane segment;
• Checking whether the mean echo pulse width of a non-ground plane segment exceeds a defined height and a defined width;
• - Determining that the non-ground plane segment is not a raised road marking if the echo pulse width of the non-ground plane segment does not exceed the defined height and the defined width;
• Determination of the non-ground plane segment as a probable raised lane marking if the echo pulse width of the non-ground plane segment exceeds the defined height and the defined width. This specification of the method step enables an efficient assessment of the presence of a raised road marking. Calculating the mean echo pulse width of each non-ground plane segment is carried out by determining the echo pulse width of each sample point and calculating an average for all sample points in the segment.

To check whether the mean echo pulse width of a non-ground plane segment exceeds a defined height and a defined width, a database of echo pulse width measured values for raised road markings, in particular Botts points and other objects, can be used. The other objects can be cars, trucks, motorcycles, tires, reflective posts, and / or pedestrians. The limit value for a maximum height has proven to be 5 centimeters and the limit value for a maximum width 12 centimeters including as suitable values for checking whether the mean echo pulse width of a non-ground plane segment exceeds a defined height and a defined width. If the result of the check as to whether the mean echo pulse width of the non-ground plane segment exceeds a defined height and a defined width shows that the non-ground plane segment is not a raised road marking, the method is ended, at least according to the object of the invention, and the mean one appears Echo pulse width unrelated to a raised lane marking.

• - Extrahieren von Konturpunkten von den segmentierten Abtastpunkten des Nicht-Bodenebenensegments;
• - Erzeugen einer Merkmalsform von den Konturpunkten.

According to a modified embodiment of the invention, the method is characterized in that the contouring of the segmented sampling points is carried out at least by

• - Extracting contour points from the segmented sampling points of the non-ground plane segment;
• - Generating a feature shape from the contour points.

When extracting contour points from the segmented sampling points of the non-ground plane segment, it should be taken into account that the number of segmented sampling points can be very high, so that a lot of computing power is required for this. In order to reduce the computing power, the most likely sampling points of the segmented sampling points are selected and used as contour points.

According to a modified embodiment of the invention, the method is characterized in that, after the contouring of the segmented scanning points, the boundary frame is tracked for validating the detected and / or classified raised lane marking. This increases the quality of the detection and classification of the raised road markings. For example, different models of raised pavement markings may have the same surface texture and height, but the lateral contour may be different. In this case, tracking helps ensure that no misclassification occurs. This ensures that the raised lane markings, in particular the Botts points, are classified until the vehicle has passed them.

According to a modified embodiment of the invention, the method is characterized in that the classification of the boundary frame is carried out by comparing its segmented sample points with reference sample points of boundary frames by a vehicle-internal or an external classification unit. The classification of the bounding box is preferably carried out in the vehicle. The internal classification unit enables a quick analysis of sampling points even with bad connections in respective distant areas. The advantage of the external classification unit, for example on an external server, and when evaluating the data externally is that external data can be kept up-to-date across all vehicles. For example, this saves storage capacity in the vehicle that is seldom required when abroad, ie in areas in which the driver normally drives infrequently.

According to a modified embodiment of the invention, the method is characterized in that the classification of the bounding frame by comparing its segmented sampling points with the reference sampling points of bounding frames using dimensions of the bounding frame, in particular the height, the length and / or the width, the echo pulse width and / or a moving state of the bounding box. It has been shown that this data is very useful for a good and quick collection and classification.

According to a modified embodiment of the invention, the method is characterized in that in the event that a defined pattern of raised road markings is identified, a driver receives a defined communication, in particular a warning message. Therefore, by classifying the raised lane markings together with a pattern, the warning level can be higher than with detection alone, thereby increasing the safety for the driver.

These and other aspects of the invention will be apparent and illustrated in the embodiments described below. Individual features that are shown in the embodiments can form an aspect of the present invention on their own or in combination. Features of the various embodiments can be transferred from one embodiment to another embodiment.

• 1 ein Ablaufdiagramm eines erfindungsgemäßen Verfahrens;
• 2 ein Ablaufdiagramm des Verfahrens gemäß 1 mit mehr Merkmalen;
• 3 ein Diagramm erfasster Ziele mit unterschiedlichen Echopulsbreiten; und
• 4 eine perspektivische Ansicht einer erhabenen Fahrbahnmarkierung, die in der Nähe einer Fahrspurmarkierung auf einer Bodenebene angeordnet ist.

Show it:

• 1 a flow chart of a method according to the invention;
• 2 a flow chart of the method according to 1 with more features;
• 3 a diagram of detected targets with different echo pulse widths; and
• 4th a perspective view of a raised lane marking, which is arranged in the vicinity of a lane marking on a ground level.

• - Erfassen von Umgebungsinformation über die Straße durch mindestens einen LIDAR-Detektor 100, wobei erfasste optische Signale verarbeitet und in elektrische Signale umgewandelt werden, die als Abtastpunkte bereitgestellt werden;
• - Segmentieren der Abtastpunkte 200 in zusammenhängende Abtastpunkte, wobei mindestens einige der segmentierten Abtastpunkte aufgrund ihrer Erfassungsposition mit einer erhabenen Fahrbahnmarkierung 1 in Beziehung gesetzt werden können;
• - Konturieren der segmentierten Abtastpunkte 300, die mit einer erhabenen Fahrbahnmarkierung 1 in Beziehung gesetzt werden können, in eine Merkmalsform, wobei basierend auf der Merkmalsform 330 ein Begrenzungsrahmen erzeugt wird;
• - Klassifizieren des Begrenzungsrahmens 400 durch Vergleichen seiner segmentierten Abtastpunkte mit Referenzabtastpunkten von Begrenzungsrahmen, die in einem Speicher gespeichert sind.

1 shows a flow chart of a method according to the invention with its main features. This method is used to detect and classify raised road markings 1 on a street 2 using a vehicle lidar detector and has the following steps:

• - Acquisition of environmental information about the road by at least one LIDAR detector 100 wherein detected optical signals are processed and converted into electrical signals that are provided as sampling points;
• - segmenting the sampling points 200 into contiguous sampling points, with at least some of the segmented sampling points having a raised lane marking due to their detection position 1 can be related;
• - Contouring of the segmented sampling points 300 that have a raised lane marking 1 can be related into a feature form, based on the feature form 330 creating a bounding box;
• - Classifying the bounding box 400 by comparing its segmented sample points with reference sample points of bounding frames stored in memory.

According to a preferred embodiment of the invention, a field-programmable gate array unit reads the electrical signals and converts them into scanning points which contain information, in particular an echo pulse width, a position and / or an angle of the scanning points.

• Das Verfahren beginnt mit dem Erfassen von Umgebungsinformation über die Straße durch mindestens einen LIDAR-Detektor 100, wobei der LIDAR-Detektor Abtastpunkte durch Verarbeiten erfasster Lichtsignale und Umwandeln der Lichtsignale in elektrische Signale erzeugt, die die Abtastpunkte darstellen.

2 FIG. 11 shows a flow chart of a method according to a preferred embodiment with the method of FIG 1 . This method is used to detect and classify the raised road markings 1 in the street 2 using the vehicle's lidar detector and has the following steps:

• The method begins with the detection of environmental information about the road by at least one LIDAR detector 100 wherein the lidar detector generates sample points by processing detected light signals and converting the light signals into electrical signals representative of the sample points.

• - Schätzen einer Bodenebene von den Abtastpunkten (210);
• - Prüfen, ob Abtastpunkte auf oder über der Bodenebene angeordnet sind (220);
• - Erzeugen eines Bodenebenensegments, falls die Punkte auf der Bodenebene angeordnet sind (235);
• - Erzeugen eines Nicht-Bodenebenensegments und Schätzen seiner Höhe und Breite, falls die Punkte oberhalb der Bodenebene angeordnet sind (230);
• - Berechnen einer mittleren Echopulsbreite jedes Nicht-Bodenebenensegments (240);
• - Prüfen, ob die mittlere Echopulsbreite eines Nicht-Bodenebenensegments eine definierte Höhe und eine definierte Breite überschreitet (250);
• - Bestimmen, dass das Nicht-Bodenebenensegments keine erhabene Fahrbahnmarkierung 1 ist, falls die Echopulsbreite des Nicht-Bodenebenensegments die definierte Höhe und die definierte Breite nicht überschreitet (265);
• - Bestimmen, dass das Bodenebenensegment wahrscheinlich eine erhabene Fahrbahnmarkierung 1 ist, falls die Echopulsbreite des Nicht-Bodenebenensegments die definierte Höhe und die definierte Breite überschreitet (260). Wenn das Ergebnis von Schritt 250 anzeigt, dass das Bodenebenensegment keine erhabene Fahrbahnmarkierung 1 ist (265), ist das Verfahren beendet, zumindest gemäß der Aufgabe der Erfindung, und scheint die mittlere Echoimpulsbreite nicht mit einer erhabenen Fahrbahnmarkierung 1 in Beziehung zu stehen.

The next step in the process is to segment the sample points 200 into contiguous sampling points, with at least some of the segmented sampling points based on their detection position with raised road markings 1 can be related. The segmentation of the sampling points 200 in contiguous sampling points is carried out at least by

• - Estimating a ground plane from the sampling points ( 210 );
• - Check whether sampling points are arranged on or above the ground level ( 220 );
• - Create a ground plane segment if the points are arranged on the ground plane ( 235 );
• - Create a non-ground plane segment and estimate its height and width if the points are above the ground plane ( 230 );
• - Computing an average echo pulse width of each non-ground plane segment ( 240 );
• - Check whether the mean echo pulse width of a non-ground plane segment exceeds a defined height and a defined width ( 250 );
• - Determine that the non-floor level segment is not a raised pavement marking 1 is if the echo pulse width of the non-ground plane segment does not exceed the defined height and the defined width ( 265 );
• - Determine that the ground level segment is likely to be a raised pavement marker 1 is if the echo pulse width of the non-ground plane segment exceeds the defined height and the defined width ( 260 ). If the result of step 250 indicates that the ground level segment is not a raised pavement marking 1 is ( 265 ), the method is ended, at least according to the object of the invention, and the mean echo pulse width does not appear with a raised road marking 1 to be related.

• - Extrahieren von Konturpunkten von den Abtastpunkten des Nicht-Bodenebenensegments 310;
• - Erzeugen einer Merkmalsform von den Konturpunkten 320.

The next step after segmenting the sample points 200 In contiguous sampling points is the contouring of the segmented sampling points 300 that have a raised lane marking 1 can be related in a feature form. The contouring of the segmented sampling points 300 is carried out at least by:

• - Extracting contour points from the sampling points of the non-ground plane segment 310 ;
• - Generating a feature shape from the contour points 320 .

The feature shape is used to generate a bounding box based on the feature shape 330 used.

After contouring the segmented sampling points 300 becomes the bounding box 340 for a validation of the recorded and / or classified raised road markings 1 tracked.

Then the bounding box 400 classified by comparing its segmented sample points with reference sample points of bounding frames stored in memory. This is the step of classifying the raised pavement marking based on the identified properties of the classified boundary box. The bounding box is preferably classified 400 by comparing its segmented sample points with reference sample points of bounding frames by an in-vehicle or an external classification unit. Hence, the classification of the bounding box 400 carried out in the vehicle. The step 400 executed using dimensions of the bounding frame, in particular the height, the length and / or the width, the echo pulse width and / or a state of motion of the bounding frame.

If so, a defined pattern of preferably classified raised road markings 1 is identified, a driver receives a defined communication, in particular a warning message.

3 shows a diagram of detected targets with different echo pulse widths. The object of the invention can be achieved by observing a plurality of echo pulse width values. The diagram shows stress amplitudes U over time t , the threshold values for the detection of raised road markings 1 exceed. The intersections of the line curves with the relevant threshold value are marked by a cross. Usual raised road markings 1 often contain retroreflective material that is highly reflective. Based on recordings by a LIDAR detector of 16 layers (10 °) vertical field of view VFOV and with a horizontal field of view HFOV of 145 ° (total horizontal resolution of 0.25 ° and 0.125 ° only in the range ± 15 °), a mean echo pulse width value of 1.8 m observed as follows. Echo pulse width [m] Distance [m] Detected height of a raised road marking 1.8 36 1.5875 1.8 50.5 1.5875 1.79 60 1.5875

This shows the high precision of the LIDAR detector and that a LIDAR detector for detecting raised road markings e.g. by evaluating the detected height of the raised lane marking is suitable. The risk of any errors can be reduced by, for example, tracking the object so that a high measurement quality can be guaranteed.

The echo pulse width is the indicator of the strength of the backscattered signal, taking into account the distance. 3 shows the theoretical differences in echo pulse widths based on different target types. It is observed that an object with retroreflective material compared to a target with black color and its echo pulse width B. or with white color and its echo pulse width W. , a larger echo pulse width R. Has. This means that the echo pulse widths together with the recorded heights, which are present in the sampling point properties, are not only used for recording, but also for classifying the raised road markings 1 can be used with high precision.

4th Figure 13 shows an example of a perspective view of a raised pavement marking that is in the vicinity of a lane marking 3 is arranged, both of which are on a road that is a ground level 2 represents.

List of reference symbols

1

raised road markings

2

Ground level

3

Lane marking

100

Acquisition of environmental information

200

Segment the sample points

210

Estimate a ground plane from the sample points

220

Verify that sampling points are on or above the ground level

230

Create a non-ground plane segment and estimate its height and width

235

Create a ground plane segment

240

Compute the mean echo pulse width of each non-ground plane segment

250

Check whether the mean echo pulse width of a non-ground plane segment exceeds a defined height and a defined width

260

Determine the ground plane segment as the probable raised pavement marking

265

Determine the non-ground plane segment to be no raised pavement marking

300

Contouring of the segmented sampling points

310

Extract contour points from the sample points of the non-ground plane segment

320

Generating a feature shape from the contour points

330

Generate a bounding box based on the feature shape

340

Track the bounding box

400

Classify the bounding box

U

Stress amplitude

t

time

B.

Echo pulse width of a black target

W.

Echo pulse width of a white target

R.

Echo pulse width of a reflector target

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

• US 9530062 B2 [0011]

Claims (14)

Method for detecting and classifying raised lane markings (1) on a road (2) using a LIDAR detector of a vehicle, with the following steps: - Detection of environmental information about the road by at least one LIDAR detector (100), wherein detected optical signals are processed and converted into electrical signals which are provided as sampling points; - segmenting the sampling points (200) into contiguous, segmented sampling points, at least some of the segmented sampling points being able to be related to a raised road marking (1) on the basis of their detection position; - Contouring the segmented scanning points (300), which can be related to a raised road marking (1), into a feature shape, a bounding frame being generated (330) based on the feature shape; and Classifying the bounding box (400) by comparing its segmented sample points with reference sample points of bounding boxes stored in a memory. Procedure according to Claim 1 , characterized in that a field-programmable gate array unit reads the electrical signals and converts them into scanning points which contain information, in particular an echo pulse width, a position and / or an angle of the scanning points. Procedure according to Claim 1 or 2 , characterized in that the segmentation of the sampling points (200) into contiguous sampling points is carried out at least by: - estimating a ground plane from the sampling points (210); - Checking whether sampling points are arranged on or above the ground plane (220); - generating a ground plane segment if the points are arranged on the ground plane (235); - creating a non-ground plane segment and estimating its height and width if the points are located above the ground plane (230); - calculating an average echo pulse width of each non-ground plane segment (240); - Checking whether the mean echo pulse width of a non-ground plane segment exceeds a defined height and a defined width (250); - Determining that the non-ground plane segment is not a raised lane marking (1) if the echo pulse width of the non-ground plane segment does not exceed the defined height and the defined width (265); and - determining that the ground plane segment is likely to be a raised road marking (1) if the echo pulse width of the non-ground plane segment exceeds the defined height and the defined width (260). Method according to one of the preceding claims, characterized in that the contouring of the segmented sampling points (300) is carried out at least by - extracting contour points from the sampling points of the non-ground plane segment (310); and - generating a feature shape from the contour points (320). Method according to one of the preceding claims, characterized in that after the contouring of the segmented scanning points (300), the bounding frame is tracked (340) for validating the detected and / or classified raised lane marking (1). Method according to one of the preceding claims, characterized in that the classification of the boundary frame (400) is carried out by comparing its segmented scanning points with reference scanning points of boundary frames by an in-vehicle or an external classification unit. Method according to one of the preceding claims, characterized in the classification of the bounding frame (400) by comparing its segmented scanning points with reference scanning points of boundary fields using dimensions of the boundary frame, in particular the height, the length and / or the width, the echo pulse width and / or the Moving state of the bounding box is in progress. Method according to one of the preceding claims, characterized in that if a defined pattern of raised lane markings (1) is identified, a driver receives a defined communication, in particular a warning message. Computer program with instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method according to one of the Claims 1 to 8th execute. Computer readable storage medium with instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method according to one of the Claims 1 to 8th execute. System with a device for carrying out the steps of the method according to one of the Claims 1 to 8th . Internal system of the vehicle (3) with a device for carrying out steps of the method according to one of the Claims 1 to 8th . Vehicle with the internal system Claim 12 . Signal for transmission in one of the steps of the method according to one of the Claims 1 to 8th .

Images