DE102020007057A1 - Vehicle and method for determining a headroom - Google Patents

Abstract

The invention relates to a method for determining a drive-through height for a vehicle (5) by means of a camera (6), the camera (6) recording an image of the surroundings in front of the vehicle, the camera (6) determining from image data whether there is an object (2.1 to 2.6) in the vicinity and it is possible to drive through the object (2.1 to 2.6), with a long-range radar and two short-range radars with different positions and / or orientations on the Vehicle (5) can be used to detect objects (2.1 to 2.6) in the environment, with fields of view (LRR, MPC, SRR) of the radar sensors (1.1 to 1.3) and the camera (6) overlapping each other of the radar sensors (1.1 to 1.3) and the camera (6) an L-shape calculation is carried out in which, in addition to a recognized reflection point (3), a search is made for two other adjacent reflection points (3), the data from the radar sensors (1. 1 to 1.3) and the L-Shapes (4) determined by the camera (6) are compared with one another in order to generate a fused image with fused L-Shapes (4F), with a height (h) between fused L-Shapes to identify the passage height (4F) of an object (2.1 to 2.6) recognized as a bridge (11) or passage is determined.

Description

The invention relates to a method for determining a clearance height according to claim 1 and to a vehicle in which the method is implemented according to claim 7.

From the DE 10 2011 113 077 A1 a method for determining whether an object can be driven through by a vehicle by means of a 3D camera is known, with the 3D camera being used to record an image of the surroundings in front of the vehicle. A trajectory is determined on which the vehicle is likely to move, and image data from the 3D camera is used to determine whether there is an object located above the trajectory and whether it has one or more connections to the ground. This is used to determine whether it is possible to drive through the object, the dimensions and shape of an entry area between the object and the roadway being determined from the image data.

The invention is based on the object of specifying a method for determining a clearance height which is improved over the prior art.

The object is achieved according to the invention by a method which has the features specified in claim 1.

Advantageous embodiments of the invention are the subject of the subclaims.

In a method according to the invention for determining a clearance height for a vehicle by means of a camera, the camera records an image of the surroundings in front of the vehicle, with image data from the camera being used to determine whether an object is in the surroundings and whether the object is possible is. According to the invention, a radar sensor designed as a long-range radar and two radar sensors designed as short-range radar with different positions and / or orientations on the vehicle are used in order to detect objects in the vicinity, with the fields of view of the radar sensors and the camera overlapping one another , with the data from each of the radar sensors and the camera being used to perform an L-shape calculation in which, in addition to a recognized reflection point, a search is made for two other adjacent reflection points, the three reflection points being set in relation to one another and in a coordinate system, see above that a connection of these three points by means of lines results in an L-shape, whereby a closest point between the radar sensor or camera and object is defined as a reference point and adjacent reflection points belonging to the same object are virtually connected by means of a line, being at recognized ends of the object End reflection points are defined as the left extremum or right extremum, with an extrapolated reflection point being determined if no reflection point is detected that reveals a depth of the object, the L-shapes determined on the basis of the data from the radar sensors and the camera being compared with one another are used to generate a fused image with fused L-shapes, with a height between fused L-shapes of an object recognized as a bridge or a passage being determined in order to identify the passage height.

Vehicles, for example commercial vehicles, can be equipped with a radar and camera systems. This technology provides the foundation for a number of new innovative uses that can significantly improve the comfort and safety of vehicles. Instead of the known point scanning of the objects, an L-shape scanning is carried out. As a result, objects are recognized 3-dimensionally. In addition, the optical object data is merged with the radar data. This technology enables particularly reliable detection of objects. In particular, this technology can be used to implement a passage control for vehicles.

By means of the solution according to the invention, the risk of accidents can be reduced through reliable detection of possible accident risks and a basis can be created for autonomous driving.

Embodiments of the invention are explained in more detail below with reference to drawings.

• 1 eine schematische Ansicht eines Szenarios mit einem Radarsensor eines Fahrzeugs und einer Anzahl von Objekten im Sichtfeld des Radarsensors,
• 2 schematische Ansichten eines Szenarios mit einem Radarsensor eines Fahrzeugs und einem Objekt im Sichtfeld des Radarsensors zur Verdeutlichung einer L-Shape-Berechnung,
• 3 schematische Ansichten eines Szenarios mit einem Radarsensor eines Fahrzeugs und einem Objekt im Sichtfeld des Radarsensors zur Verdeutlichung der L-Shape-Berechnung,
• 4 eine schematische Ansicht eines Szenarios mit einem Radarsensor eines Fahrzeugs und einer Anzahl von Objekten im Sichtfeld des Radarsensors sowie mit erkannten L-Shapes,
• 5 eine schematische Ansicht eines Fahrzeugs, an dem drei Radarsensoren und eine Kamera angeordnet sind,
• 6 eine schematische Ansicht einer elektronischen Architektur des Fahrzeugs,
• 7 eine schematische Ansicht von drei Objekten, von jeweils dazugehörigen L-Shapes, die jeweils anhand der Daten der Radarsensoren und der Kamera zu jedem Objekt ermittelt wurden, sowie durch Sensorfusionierung ermittelter fusionierter L-Shapes, und
• 8 eine schematische Ansicht der Brücke aus 5, wobei mittels des beschriebenen Verfahrens erkannte L-Shapes dargestellt sind, anhand deren eine Höhe der Durchfahrt bestimmt werden kann.

Show:

• 1 a schematic view of a scenario with a radar sensor of a vehicle and a number of objects in the field of view of the radar sensor,
• 2 schematic views of a scenario with a radar sensor of a vehicle and an object in the field of view of the radar sensor to illustrate an L-shape calculation,
• 3 schematic views of a scenario with a radar sensor of a vehicle and an object in the field of view of the radar sensor to illustrate the L-shape calculation,
• 4th a schematic view of a scenario with a radar sensor of a vehicle and a number of objects in the field of view of the radar sensor and with recognized L-shapes,
• 5 a schematic view of a vehicle on which three radar sensors and a camera are arranged,
• 6th a schematic view of an electronic architecture of the vehicle,
• 7th a schematic view of three objects, of associated L-shapes, which were each determined on the basis of the data from the radar sensors and the camera for each object, as well as fused L-shapes determined by sensor fusion, and
• 8th a schematic view of the bridge 5 , L-shapes recognized by means of the described method are shown, based on which a height of the passage can be determined.

Corresponding parts are provided with the same reference symbols in all figures.

Accidents occur again and again, the cause of which is that the height of passageways and bridges is incorrectly estimated and vehicles, for example trucks, have an accident.

Sometimes the driver forgets, especially when changing vehicles frequently, that he has to pay attention to the clearance height. In some cases the superstructure protrudes beyond the actual permissible height restriction, especially for special transports.

Known radar technology and thus possible object detection is based on the classification of a so-called RCS value. RCS stands for Radar Cross Section and indicates the reflection strength of a radar wave. This depends on the electrical conductivity, shape, distance and material of an object. Classification of the RCS value distinguishes how it reacts. For example, if the RCS value exceeds a specified limit, braking is carried out.

If the RCS value is not reached, a camera is prompted to check the established situation and to confirm (sensor fusion) what kind of object it is and only then is a reaction triggered, for example a warning message and / or braking.

In addition, the movement profile is analyzed and classified and an object classification is carried out, for example by determining a MicroDoppler effect. In this way it can be determined whether the object is a person, a vehicle or something else.

While distances and speeds of moving objects can be determined very well due to Doppler and Microdoppler effects, this is difficult for stationary objects.

It is therefore possible with the known technique to determine approximately where objects with a certain RCS value are located. However, a distance measurement of the objects themselves or a distance measurement of the objects to one another has not been possible so far in a reliable manner. The result is that an unequivocal detection of bridges and passages, i.e. a height control or a distance control, has not been possible so far.

1 Figure 3 is a schematic view of a scenario with a radar sensor 1 a vehicle and a number of objects 2.1 to 2.6 in the field of view of the radar sensor 1 . The radar sensor 1 recognizes reflection points 3 on some of the objects 2.1 to 2.6 . The object 2.4 is from the object 2.3 concealed and for the radar sensor 1 not visible and therefore receives the RCS value, for example 1 . The object 2.2 is for the radar sensor 1 with three reflection points 3 visible and receives, for example, the RCS value 2. The objects 2.1 and 2.6 are, for example, boundaries, especially guard rails.

Due to the limited scanning accuracy and limitation of the number of stationary objects 2.1 to 2.6 these can be identified and their approximate position determined. An exact distance measurement of the objects 2.1 to 2.6 each other is not possible.

Through the use of improved sensors with finer scanning combined with improved object recognition through several sensors, for example three radar sensors 1 With different positions on the vehicle and a high-resolution camera, as well as computing power increased by, for example, fifteen times, this problem is solved and thus opens up new application possibilities.

One of the radar sensors 1 is designed as a long-range radar, for example with a measuring range of up to about 250m. The range of vision of the camera can be up to 180m depending on the lighting conditions. Two more of the radar sensors are each designed as a short-range radar, for example with a measuring range of up to about 85 m.

This results in a greater range of vision, better object recognition through higher scanning accuracy and improved camera resolution, and a larger angle of detection. This means that a higher number of objects can be recognized and tracked. The two additional radar sensors designed as short-range radar, In particular with extremely fine scanning and overlapping measuring range, an additional object inspection and confirmation (sensor fusion) can be carried out. Thus, improved edge detection and thus distance measurement are achieved through the use of L-shape technology.

2 shows schematic views of a scenario with a radar sensor 1 a vehicle and an object 2.1 in the field of view of the radar sensor 1 to clarify the L-shape calculation.

The L-shape calculation for radar detection helps in particular to achieve improved distance measurement. The L-shape calculation always looks for two further reflection points 3 next to a recognized reflection point 3 searched. Then an attempt is made to relate the three points and put them in a coordinate system. The connection of these three points with lines results in the shape of an L. In particular, the procedure is that the closest point between the radar sensor 1 and object 2.1 and used as a reference point 3 _Ref is defined. It is then checked whether there are adjacent reflection points 3 to the object 2.1 and these are virtually connected by a line. It is checked whether there is a relationship between these reflection points 3 consists. Ends the object 2.1 , then becomes the reflection point 3 before that as the left extremum 3L or right extremum 3R Are defined. Should be the three points of reflection 3 are not exactly on one line, an edge can be determined on which the kink of the L is placed.

3 shows schematic views of a scenario with a radar sensor 1 a vehicle and an object 2.1 in the field of view of the radar sensor 1 to clarify the L-shape calculation.

Will not be a reflection point 3 recognized the depth of the object 2.1 can be seen, it becomes a point of reflection 3L _e or 3R _e extrapolated. The distance to this is 0.3 m, for example. So should, which does not happen in practice, just a single point of reflection 3 be visible, the extrapolated L has a leg length of 0.3 m.

In the in 1 The scenario shown are therefore L-shapes 4th determined as in 3 shown. 4th Figure 3 is a schematic view of a scenario with a radar sensor 1 a vehicle and a number of objects 2.1 to 2.6 in the field of view of the radar sensor 1 as well as recognized L-shapes 4th .

Using the L-shape algorithm, the width and height of an object 2.1 to 2.6 and depending on the position, its depth can also be determined much more precisely.

In this way, the distance between the objects can also be adjusted accordingly 2.1 to 2.6 can be established among each other.

5 Fig. 3 is a schematic view of a vehicle 5 , in particular a commercial vehicle, on which three radar sensors 1 , 1.2 , 1.3 and a camera 6th are arranged, the radar sensor 1 is designed as a long-range radar that has a field of view LRR covering and being the radar sensors 1 , 1.3 are designed as short-range radar, each with a field of view SRR cover. The camera 6th covers a viewing area MPC from. In front of the vehicle 5 There is a bridge in the direction of travel 11 with an edge 12 . Not just a radar sensor on the vehicle 1 but three radar sensors 1 , 1.2 , 1.3 are used, which are positioned and / or aligned differently, the dimension and distance detection improves considerably. The camera recognizes this 6th the edges of objects 2.1 to 2.6 and also carries out a distance measurement using the L-shape calculation.

Thus, three radar sensors 1 , 1.2 , 1.3 , the long-range radar and the two short-range radars, as well as the camera 6th data supplied with overlapping viewing areas.

6th Figure 3 is a schematic view of an electronic architecture of the vehicle 5 , including the radar sensors 1 , 1.2 and the camera 6th , a video radar decision-making unit 7th , a central gateway 8th , an instrument cluster 9 for visual and acoustic warning of a driver as well as a control unit 10 of the vehicle 5 that can issue warnings to other road users, for example via a hazard warning light system or a horn.

A video radar decision-making unit 7th collects the data from the sensors, i.e. the three radar sensors 1 , 1.2 , 1.3 and the camera 6th , and compare them with each other. If extrapolated points are supplied by a sensor, they are compared with the data from another sensor and prioritized accordingly. In this way, a merged image is generated from the data supplied. 7th is a schematic view of three objects 2.1 to 2.3 , of the corresponding L-shapes 4L , 4S and 4M based on the data from the radar sensors 1 , 1.2 and the camera 6th to every object 2.1 to 2.3 as well as fused L-shapes determined by the sensor fusion 4F . The in 7th shown object 2.1 For example, the data from the radar sensor designed as a short-range radar would be 1 used because this detects more coherent reflections while the camera is 6th and the radar sensor designed as a long-range radar 1 in the right part of the Object 2.1 only provide extrapolated data. With the objects 2.2 and 2.3 it is the other way around.

The higher scanning performance, the higher number of sensors, the increased number of objects to be tracked 2.1 to 2.6 , the sensor fusion as a result of four different measuring sensors as well as the application of the described L-shape calculation enables an improved height and width control of passages for acoustic and / or optical warning of the driver.

In one embodiment, active intervention, for example braking, can be provided if a height that is too low and / or too wide has been detected.

8th is a schematic view of the bridge 11 out 5 , L-shapes recognized by means of the method described 4th are shown, based on which a height H the passage can be determined.

In one embodiment, a comparison with a traffic sign recognition is possible, provided that this indicates, for example, a construction site situation or a marked situation (in particular restricted clearance height).

List of reference symbols

1

Radar sensor

1.1 to 1.3

Radar sensor

2.1 to 2.6

object

3

Reflection point

3L

left extremum

3L _e

extrapolated reflection point, extrapolated left extremum

3R

right extremum

3 _Ref

Reference point

4th

L-shape

4F

fused L-shape

4L, 4M, 4S

L-shape

5

vehicle

6th

camera

7th

Video radar decision unit

8th

central gateway

9

Instrument cluster

10

Control unit

11

bridge

12

Edge

H

height

LRR

Viewing area

MPC

Viewing area

SRR

Viewing area

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

• DE 102011113077 A1 [0002]

Claims (7)

Method for determining a clearance height for a vehicle (5) by means of a camera (6), the camera (6) taking an image of the surroundings in front of the vehicle, with image data from the camera (6) being used to determine whether a Object (2.1 to 2.6) is in the vicinity and it is possible to drive through the object (2.1 to 2.6), characterized in that, in addition, a radar sensor (1.1) designed as a long-range radar and two radar sensors designed as short-range radar (1.2, 1.3) can be used with different positions and / or orientations on the vehicle (5) in order to detect objects (2.1 to 2.6) in the environment, with areas of view (LRR, MPC, SRR) of the radar sensors (1.1 to 1.3) and of the camera (6) overlap each other, with the data from each of the radar sensors (1.1 to 1.3) and the camera (6) being used to make an L-shape calculation in which, in addition to a recognized reflection point (3), after two other adjacent reflection points (3) is sought, the three reflection points (3) being set in relation to each other and in a coordinate system so that a connection of these three points by means of lines results in an L-shape (4), with a closest point between the radar sensor (1.1 to 1.3) or camera (6) and object (2.1 to 2.6) are defined as reference point (3 _Ref ) and adjacent reflection points (3) belonging to the same object (2.1) are virtually connected by means of a line, with recognized ends of the object (2.1 to 2.6) lying reflection points (3) can be defined as left extremum (3L) or right extremum (3R), whereby if no reflection point (3) is recognized that shows a depth of the object (2.1 to 2.6), an extrapolated one Reflection point (3L _e , 3R _e ) is determined, the L-shapes (4) determined on the basis of the data from the radar sensors (1.1 to 1.3) and the camera (6) being compared with one another to form a fused image with fused L-shapes (4F ) to generate, with a height (h) between fused L-shapes (4F) of an object (2.1 to 2.6) recognized as a bridge (11) or passage being determined to detect the passage height. Procedure according to Claim 1 , characterized in that when the three reflection points (3) do not lie exactly on one line, an edge is determined on which a kink of the L-shape (4) is placed. Procedure according to Claim 1 or 2 , characterized in that a drive-through width is determined in the same way. Method according to one of the preceding claims, characterized in that an acoustic and / or visual warning is output if the passage height and / or passage width detected is too low for the vehicle (5). Method according to one of the preceding claims, characterized in that if the passage height and / or passage width detected are too low for the vehicle (5), an active driving intervention takes place. Method according to one of the preceding claims, characterized in that a comparison with a traffic sign recognition also takes place. Vehicle (5), comprising an electronic architecture, comprising a camera (6), a central gateway (8), an instrument cluster (9) for the visual and acoustic warning of a driver and a control unit (10) of the vehicle (5), which is designed to issue warnings to other road users, characterized in that, in addition, a long-range radar designed as a radar sensor (1.1) and two short-range radar designed radar sensors (1.2, 1.3) with different positions and / or orientations on Vehicle (5) are arranged, with fields of view (LRR, MPC, SRR) of the radar sensors (1.1 to 1.3) and the camera (6) overlap each other, with a video radar decision unit (7) also being provided which is used for Implementation of the method according to one of the preceding claims is configured.

Images