DE102018124979A1 - Driver assistance system for determining a distance between two vehicles with a camera - Google Patents

Abstract

The invention relates to a driver assistance system (1) having a camera (4) for determining a distance (13) between a first vehicle (2) and a second vehicle (3), the camera (4) having an image sensor (5) and Camera (4) a two-dimensional image with pixels on which the second vehicle (3) is at least partially formed generated, and the driver assistance system (1) with an object detection module (7) a part (8) of the second vehicle (3) based on Recognizes a vehicle model of the second vehicle (3) based on the image and the driver assistance system (1) a vehicle database (10) from which at least one dimension (11) of the part (8) in relation to the Vehicle model is readable, wherein the driver assistance system (1) for a dimension (11) corresponding reference dimension of the part (8) using the pixels of the image detected and the distance (13) based on a ratio between d he dimension (11) and the reference dimension determined.

Description

The invention relates to a driver assistance system for determining a distance between a vehicle and another vehicle.

In the DE 103 05 935 A1 such a driver assistance system is disclosed. The driver assistance system described therein has a radar sensor for determining distances between the vehicle and other objects. In addition, the driver assistance system has a camera sensor. With signals from the radar sensor and the camera sensor, information about the distances of the objects to the vehicle and the contours of the objects can be displayed to a driver of the vehicle.

From the DE 10 2011 086 025 A1 a method is known in which objects are detected in the environment of a vehicle with a camera. In addition, ultrasonic signals are transmitted with an ultrasonic sensor. At the same time as the ultrasound signals, light pulses are generated by a light source. Including echo propagation times of the ultrasonic signals transmitted and received by the ultrasonic sensor, a calculation of spatial coordinates of the objects in the environment is carried out as a function of the reflected light pulses.

Furthermore, the describes DE 102 41 464 A1 a device for monitoring the surroundings of a vehicle with a camera. The camera is used to monitor a remote area. A near zone is monitored by means of ultrasonic or microwave sensors, whereby a distance measurement takes place with these sensors.

These known driver assistance systems have the disadvantage that, in addition to the camera, a distance sensor is used to determine a distance of an object from the vehicle.

The object of the present invention is therefore to provide a driver assistance system which can determine a distance between an object detected by the camera and the vehicle with a lesser amount of hardware.

This object is achieved with a driver assistance system having the features of claim 1 and a method having the features of claim 4. Advantageous embodiments and further developments are the subject of the dependent claims.

To solve this problem, a driver assistance system for determining a distance between a first vehicle and a second vehicle is proposed. The driver assistance system has a camera with an image sensor. The camera produces a two-dimensional image with pixels on which the second vehicle is at least partially imaged. Furthermore, the driver assistance system with an object recognition module recognizes a part of the second vehicle based on the image and determines with an identification module a vehicle model of the second vehicle based on the image. Furthermore, the driver assistance system has a vehicle database, from which at least one dimension of the part of the second vehicle relative to the vehicle model can be read out. According to the invention, the driver assistance system detects a reference dimension corresponding to the dimension of the part with the aid of the pixels of the image and determines the distance between the first and the second vehicle based on a ratio between the dimension and the reference dimension.

The advantage of the proposed driver assistance system is that no additional distance sensor is necessary to determine the distance between the first and the second vehicle in addition to the camera. This saves costs in the production of the driver assistance system and space within the first vehicle, because no additional distance sensor must be mounted in a body of the vehicle. The camera can furthermore be designed as a monocamera in the proposed driver assistance system. As a result, compared to a use of a stereo camera to determine the distance also a reduction of material costs can be achieved.

Preferably, the driver assistance system displays the distance on a display of the vehicle, so that the driver can be supported when driving the vehicle. In addition, provision may be made for the distance to be used as an input variable for a control device of the vehicle for controlling the vehicle. For example, the distance may be an input variable of a speed controller of the control unit.

The invention is based on the idea that the camera can image reflecting points in a three-dimensional geometric environment of the first vehicle onto the two-dimensional image. The image sensor has pixels that correspond to the pixels of the image or can be assigned to them. The recognized part of the second vehicle is imaged on the image sensor, preferably with the aid of a lens of the camera. Depending on the distance of the part from the lens, an image of the part imaged on the image sensor extends over a certain number of pixels of the image. As the distance increases or decreases Part of the lens reduces or increases the number of pixels over which the image of the part extends.

The reference dimension is determined by the number of pixels the image of the part extends. The distance between the first and second vehicles, hereinafter referred to as object width, may be calculated from an image distance defined by a distance of the image sensor from the lens, the reference dimension, and the dimension.

Preferably, the object distance is calculated from a product having the image width as a first factor and a quotient as a second factor, wherein the quotient of the dimension is formed as a dividend and the reference dimension as a divisor. A calculation of the object width according to this method is particularly well suited for distances of the second vehicle from the first vehicle, which are in the order of more than 10 m.

The term "module" as used herein describes any known or later developed hardware, software, firmware, artificial intelligence, fuzzy logic or combination of hardware and software capable of interfacing with the particular module Perform associated functionality.

The object recognition module advantageously has filter banks and / or neural networks. Preferably, the object recognition module is designed such that the part of the second vehicle can be recognized on the basis of the pixels of the two-dimensional image. The part may be, for example, a tailgate, a rear of a truck container, a bumper, an entire rear of a body, a rear window, an automobile manufacturer's trademark, a taillight, a vent, a rear spoiler, or a C-pillar. These various embodiments of the part can be imaged in the object recognition module in the form of individual component classes. In this case, specific values of shape parameters can advantageously be stored in the object recognition module for each component class for the component class. Particularly advantageously, the object recognition module is designed such that upon recognition of the part, the part is assigned to a single component class.

The identification module is preferably constructed in such a way that the vehicle model of the second vehicle can be determined as the output variable of the identification module with the aid of the pixels of the two-dimensional image as an input variable of the identification module. In a preferred embodiment, it is possible, for example, that with the aid of the identification module at least a first value of a first shape parameter of the second vehicle is calculated with the aid of the pixels. In this case, the pixels can be used with which the part is displayed. In a further variant, pixels may also be used for calculating the first value of the first shape parameter, which map another part of the second vehicle. Under certain circumstances it is possible that the determination of the vehicle model with the aid of the further part is simpler than with the part recognized using the object recognition module.

Preferably, the first value is compared with values of shape parameters, which are preferably stored in a shape database. The values of the shape parameters stored in the shape database are each assigned to a vehicle model. It is possible that in the form database, several values of shape parameters are assigned to a single vehicle model. For example, for a single vehicle model of an automobile manufacturer from a particular year of manufacture, values of shape parameters for representing a curved edge curve of a rear window may be stored in the shape database. Advantageously, the values are suitable for constructing one or more splines.

In a special variant, the values of the shape parameters are stored in the form of stored weight values of individual weighted connections between neurons of a neural network. Particularly advantageously, the identification module has a neural network, which was preferably trained with values from the shape database. In this variant, the values of the shape parameters are represented by stored weight values of individual neurons of the neural network.

For reading out the dimension of the part, the vehicle database conveniently has one row for each vehicle model and one column for different parts of a vehicle model. Thus, it is possible that in each case a dimension of the respective part is stored in the vehicle database for different parts of the second vehicle. The various parts can preferably be classified in the different component classes of the object recognition module. Depending on which part of the second vehicle is detected more quickly with the aid of the object recognition module, this part can be used to determine the distance. This variant for determining the vehicle model can be compared to a variant in which only one dimension of a single part of a vehicle model in the vehicle database is deposited, requires less computing time or computing capacity.

In a particular embodiment, the dimension is a height of the part. This has the advantage that when a lane change of the second vehicle, the reference dimension, which in this case is a reference level, does not change. However, it is possible that the reference altitude changes as the second vehicle travels a higher or lower pitch than the first vehicle.

In order to overcome this disadvantage, the driver assistance system advantageously has a memory in which at least one gradient of a roadway traveled by the first vehicle is stored. The driver assistance system corrects the distance with the help of the slope progression. In a particular embodiment, the memory may be in the form of files stored in a cloud.

Advantageously, in a first step of correcting the distance by means of a first inclination angle of the roadway at a first distance determined according to the proposed method, the reference dimension can be corrected. The first inclination angle is preferably determined by means of a digital map. A corrected reference dimension can be obtained by dividing the reference dimension by the cosine of the first pitch angle. Using the corrected reference dimension, as described above, in a second step, the subject distance may be calculated according to the proposed method as a corrected distance between the first and second vehicles.

In a third step, based on the corrected distance, a second inclination angle of the road surface corresponding to the corrected distance is read out using the digital map. The thus determined second tilt angle can be used for a new correction of the reference dimension. Steps one through three of correcting the distance may be performed several times in succession to iteratively calculate the object's distance.

In order to achieve the abovementioned object, a method for determining a distance between a first vehicle and a second vehicle by means of a driver assistance system is also proposed. The driver assistance system can be designed according to one of the variants mentioned above. In a first step, the two-dimensional image, on which the second vehicle is at least partially imaged, is generated with the camera. In a second step, the part of the second vehicle is detected by means of the object recognition module and the image. In a third step, the vehicle model of the second vehicle is determined based on the image and with the aid of the identification module.

In a fourth step, the dimension of the part of the second vehicle relative to the vehicle model is read from the vehicle database. In a fifth step, a reference dimension of the part corresponding to the dimension is determined with the aid of the pixels of the image. In a sixth step, a ratio between the dimension and the reference dimension is calculated. In a seventh step, the distance is determined by the ratio. The ratio may be formed, for example, as a quotient with the dimension as a dividend and the reference dimension as a divisor.

In a particular embodiment of the method, it is provided that an image width, which is defined by a distance of the image sensor from a lens of the camera, is changed for focusing the second vehicle and the distance is determined on the basis of the ratio and the image width.

• 1 eine schematische Ansicht eines Fahrerassistenzsystems für ein erstes Fahrzeug zur Bestimmung einer Entfernung zwischen dem ersten Fahrzeug und einem zweiten Fahrzeug,
• 2 Schritte eines Verfahrens zur Bestimmung der Entfernung zwischen dem ersten Fahrzeug und dem zweiten Fahrzeug,
• 3 eine Schnittansicht einer Linse, eines Bildsensors mit Bildpunkten, von zwei Objekten und von den Objekten reflektierte auf die Bildpunkte auftreffende Lichtstrahlen,
• 4 eine Draufsicht der Linse aus 3.

Further advantages, features and details of the invention will become apparent from the following description of at least one preferred embodiment and with reference to the figures. These show in:

• 1 1 is a schematic view of a driver assistance system for a first vehicle for determining a distance between the first vehicle and a second vehicle;
• 2 Steps of a method for determining the distance between the first vehicle and the second vehicle,
• 3 a sectional view of a lens, an image sensor with pixels, of two objects and reflected by the objects incident on the pixels light beams,
• 4 a plan view of the lens 3 ,

1 shows a driver assistance system 1 for a first vehicle 2 for determining a distance 13 between the first vehicle 2 and a second vehicle 3 , The driver assistance system 1 has a camera 4 , an object recognition module 7 , an identification module 9 and a vehicle database 10 on. The camera 4 has an image sensor 5 and a lens 6 , where the image sensor 5 preferably in a focal plane of the lens 6 is arranged. With the camera 4 is a two-dimensional image can be generated with pixels on which the second vehicle 3 at least partially depicted. The Object recognition module 7 is designed such that with this a part 8th of the second vehicle can be recognized by the image.

The identification module 9 is set up such that a vehicle model of the second vehicle can be determined based on the image. From the vehicle database 10 is at least one dimension 11 of the part 8th of the second vehicle 3 readable in relation to the vehicle model. With the driver assistance system 1 is one to the dimension 11 corresponding reference dimension 12 of the part 8th detectable with the help of the pixels of the image. Furthermore, with the driver assistance system 1 the distance 13 by a ratio between the dimensions 11 and the reference dimension 12 determinable.

2 shows a method for determining the distance 13 between the first vehicle 2 and the second vehicle 3 with the help of the driver assistance system 1 , In a first step 21 becomes the two-dimensional image with the camera 4 generated on which the second vehicle 3 at least partially depicted. In a second step 22 is done with the help of the object recognition module 7 and the picture the part 8th of the second vehicle 3 recognized. In a third step 23 becomes the vehicle model of the second vehicle 3 based on the image and with the help of the identification module 9 certainly. In a fourth step 24 becomes the dimension 11 of the part 8th of the second vehicle 3 in relation to the vehicle model from the vehicle database 10 read. In a fifth step 25 becomes one to the dimension 11 corresponding reference dimension 12 of the part 8th determined with the help of the pixels of the image. In a sixth step 26 will be a ratio between the dimensions 11 and the reference dimension 12 calculated. In a seventh step 27 will the distance 13 determined by the ratio.

3 shows an exemplary representation of how a change in a distance of an object from the lens 6 to a number of pixels over which an image of the object extends. As in 1 Shown to be the second vehicle 3 in this example the same distance 13 from the first vehicle 2 exhibit. Take the second vehicle 3 the distance 13 from the first vehicle 2 a, so extends the image of the recognized part 8th over two of the pixels 31 of the image sensor 5 ,

The distance 13 is from a trailing edge of the part 8th , which is a complete rear end of the second vehicle 3 can be up to a central axis 32 the lens 6 measured. In addition to the second vehicle 3 should a third vehicle 33 which is a distance 36 from the lens 6 has to be considered. The third vehicle 33 has a second part 34 on, which is also an entire rear area of the third vehicle 33 can be. The entire rear area 34 has a second dimension 35 which is equal to the dimension 11 of the part 8th of the second vehicle 3 is.

The entire rear area 34 of the third vehicle 33 is in 3 represented larger than the entire rear area 8th of the second vehicle 3 , This is justified by the fact that 3 is merely a schematic representation of an arrangement in which the second vehicle 3 significantly further away from the lens 6 as the third vehicle 33 is. For this reason, the second part captures 34 of the third vehicle 33 a larger area of an in 3 not shown light cone of the lens 6 as the part 8th of the second vehicle 3 , In 1 . 3 and 4 are each shown for clarity only the light rays passing through the center of the lens 6 walk.

By doing that part 34 of the third vehicle 33 detects a larger area of the cone of light that hits the part 34 of the third vehicle 33 emitted light on more pixels 31 of the image sensor 35 as that of the part 8th of the second vehicle 3 emitted light. In the in 3 the example shown extends the image of the part 34 of the third vehicle 33 over seven pixels 31 , With the help of the image sensor 5 can be a corresponding number of the part 8th of the second vehicle 3 or the part 34 of the third vehicle 33 exposed pixels 31 be recorded.

The image sensor 5 is how in 1 shown with the help of an evaluation unit 14 connected. The evaluation unit 14 may be a microprocessor on which the object recognition module 7 , the identification module 9 and the vehicle database 10 in the form of electronic components, such. B. transistors or memory elements are arranged. In any case, the image sensor 5 via an electronic way with the evaluation unit 14 connected. Preferably, the identification module 9 with the vehicle database 10 as well as the object recognition module 7 directly with the image sensor 5 electronically connected.

4 shows a plan view of the lens 6 and the image sensor 5 , the second vehicle 3 and a fourth vehicle 41 , 4 shows that it is with the proposed driver assistance system 1 is possible, at least a first pixel 43 a first volume element 44 with a position in one with the camera 4 detectable volume 45 assigned. The second vehicle 3 is here at least partially in the first volume element 44 , one of the second vehicle 3 reflected light steel with the help of the lens 6 at the first pixel 43 is diverted.

In the same way, a second pixel 46 the first volume element 44 be assigned. It is not necessary that the entire second pixel 46 from the second vehicle 3 outgoing light rays is detected. Becomes the second pixel 46 only partially from that of the second vehicle 3 detected outgoing light rays, it reduces only one with the second pixel 46 recorded light intensity.

In the same way, a third pixel 47 and a fourth pixel 48 a second volume element 49 with a second position in the volume detectable by the lens 45 be assigned. Preferably, a respective light intensity of the pixels 31 , which include the pixels 43 . 46 . 47 and 48 belong, with the help of the evaluation unit 14 be recorded.

Furthermore, it is possible that with the help of the evaluation unit 14 the first pixel 43 the second vehicle 3 is assigned when the part 8th of the second vehicle 3 using the object recognition module 7 is recognized. An azimuth angle is preferred 50 between a connecting line 51 that between the second vehicle 3 and the first vehicle 2 runs, and an optical axis 52 the lens 6 calculated.

With the help of the azimuth angle 50 and the distance 13 can be a position of the second vehicle 3 in a horizontal environment of the first vehicle 2 be determined. In the same way as for in 4 shown horizontal plane can in a vertical plane, an elevation angle of the first volume element 44 determined and in connection with the distance 13 a position of the second vehicle 3 in a vertical environment of the first vehicle 2 be determined.

By doing that, a total number of pixels 31 of the image sensor 5 and one with the lens 6 detectable total volume 45 are known, each pixel of the image sensor 5 a defined volume element of the entire volume 45 be assigned. This becomes the three-dimensional volume 45 on a two-dimensional surface of the pixels 31 of the image sensor 5 displayed. Is a single pixel of a ray of light from the second vehicle 3 is reflected, illuminated, so this single pixel and thus the second vehicle 3 on the one hand a certain volume element of the volume 45 as well as a corresponding elevation angle and a corresponding azimuth angle are assigned.

A particular variant of the method may provide for an image distance defined by a distance 37 of the image sensor from the lens 6 is defined, for focusing the second vehicle is changed and the distance 13 determined on the basis of the ratio and the image width.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been 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.

Cited patent literature

• DE 10305935 A1 [0002]
• DE 102011086025 A1 [0003]
• DE 10241464 A1 [0004]

Claims (10)

Driver assistance system (1) having a camera (4) for determining a distance (13) between a first vehicle (2) and a second vehicle (3), the camera (4) having an image sensor (5) and the camera (4) a two-dimensional image with pixels on which the second vehicle (3) is at least partially displayed generated, and the driver assistance system (1) with an object recognition module (7) recognizes a part (8) of the second vehicle (3) based on the image and with an identification module (9) determines a vehicle model of the second vehicle (3) on the basis of the image and the driver assistance system (1) a vehicle database (10) from which at least one dimension (11) of the part (8) can be read in relation to the vehicle model, wherein the driver assistance system (1) detects a reference dimension of the part (8) corresponding to the dimension (11) using the picture elements of the image and the distance (13) based on a ratio between the dimension (11) and the R Reference dimension determined. Driver assistance system (1) after Claim 1 , characterized in that the dimension (11) is a height of the part (8). Driver assistance system (1) after Claim 1 or 2 , characterized in that the driver assistance system (1) has a memory in which at least one gradient of a roadway traveled by the first vehicle is stored, and the driver assistance system (1) corrects the distance (13) by means of the gradient course. Method for determining a distance (13) between a first vehicle (2) and a second vehicle (3) by means of a driver assistance system (1) according to one of Claims 1 to 3 comprising the following steps: - generating the two-dimensional image with the camera (4) on which the second vehicle (3) is at least partially imaged, - recognizing the part (8) of the second vehicle (3) using the object recognition module (7) and the image, - determining the vehicle model of the second vehicle (3) based on the image and using the identification module (9), - reading the dimension (11) of the part (8) of the second vehicle (3) from the vehicle database (10) Relating to the vehicle model, - determining the reference dimension of the part (8) corresponding to the dimension (11) using the pixels of the image, - calculating the ratio between the dimension (11) and the reference dimension, - determining the distance (13) from the ratio , Method according to Claim 4 , Characterized in that to at least a first volume element (44) is associated with a position in one with the camera (4) detectable volume of a first pixel of the pixels, wherein the second vehicle (3) at least partially (in the first volume element 44 ) and the position is assigned to the second vehicle (3). Method according to Claim 4 or 5 characterized in that an azimuth angle (50) between a connecting line (51) from the second vehicle (3) to the first vehicle (2) and an optical axis (52) of a lens (6) of the camera (4) is calculated. Method according to Claim 6 , characterized in that by means of the azimuth angle (50) and the distance (13) a position of the second vehicle (3) in an environment of the first vehicle (2) is determined. Method according to one of Claims 4 to 7 , characterized in that a height of the part (8) is used as the dimension (11). Method according to one of Claims 4 to 8th , characterized in that with a digital map determines a slope course of a of the first vehicle (2) traveled roadway and the distance (13) is corrected based on the inclination curve. Method according to one of Claims 4 to 9 characterized in that an image distance defined by a distance (37) of the image sensor from a lens (6) of the camera (4) is changed to focus the second vehicle, and the distance (13) is changed based on the ratio and the image width is determined.

Images