Method for classifying a relevance of an object

Abstract

The invention relates to a method for classifying a relevance of an object, which is located in the surroundings of a motor vehicle captured by a surrounding sensor, in respect of a collision with the motor vehicle, comprising the following steps: receiving measurement signals, which represent a radial distance dr of the object relative to the surroundings sensor, measured by the surroundings sensor, a radial relative velocity vr of the object relative to the surroundings sensor, measured by the surroundings sensor, and a measured own velocity vego of the motor vehicle; receiving dimension signals, which represent the dimensions of the motor vehicle, calculating whether the motor vehicle can collide with the object based on the measurement signals received and based on the dimension signals received; outputting a result signal which represents a result of the calculating of whether the motor vehicle can collide with the object, in order to classify the relevance of the object in respect of a collision with the motor vehicle. The invention furthermore relates to an apparatus, a computer program and a machine-readable storage medium.

Description

[0001] The invention relates to a method for classifying the relevance of an object. The invention further relates to a device configured to perform all steps of the method for classifying the relevance of an object. The invention further relates to a computer program. The invention also relates to a machine-readable storage medium. State of the art

[0002] Patent application US 2009 / 015462 A1 discloses a radar device for detecting an object in the vicinity of a vehicle.

[0003] The patent application US 2014 / 195132 A1 discloses a method and a device for monitoring objects in the vicinity of a vehicle.

[0004] Document DE102010054080A1 discloses, with the aim of determining the collision risk of an ego-vehicle with neighboring vehicles, a method for assessing the relevance of vehicles belonging to the assistance level of a Car2X network in relation to the ego-vehicle also belonging to this assistance level. A perimeter is placed around each vehicle, the radius of which is a function of the inaccuracy of the position determination for that vehicle. If the perimeter of a particular vehicle does not overlap with that of the ego-vehicle, that vehicle is classified as irrelevant.

[0005] Classifying the relevance of the stationary vehicle environment presents a challenge for certain types of environmental sensors (e.g., radar) with current technology. The aim of this classification is to differentiate between functionally relevant elements of the stationary vehicle environment (e.g., parked vehicles), which should be subject to control (e.g., braking), and irrelevant objects, such as (drive-under) overhead sign gantries or (drive-over) ground objects, which should not be subject to control.

[0006] The challenge with the classification under consideration lies particularly in sufficiently suppressing false-positive relevance reports while simultaneously maintaining a high level of performance regarding the positive reporting of relevant objects. Typically, complex classification approaches are used in a high-dimensional feature space, where the individual features often do not directly assess an object's relevance, but rather do so indirectly via derived variables. These derived variables necessitate complex assumptions about the nature or behavior of the respective objects, which limits the robustness of the classifier based on them. Disclosure of the invention

[0007] The object underlying the invention is to provide a concept for efficiently classifying the relevance of an object located in the vicinity of a motor vehicle comprising an environmental sensor with regard to a collision with the motor vehicle. This object is achieved by means of the respective subject matter of the independent claims.

[0008] Following a first aspect, a procedure for classifying the relevance of an object located in the vicinity of a motor vehicle comprising an environment sensor with regard to a collision with the motor vehicle is provided, comprising the following steps: Receiving measurement signals representing a radial distance dr of the object relative to the environmental sensor, a radial relative velocity vr of the object relative to the environmental sensor, and a measured self-velocity v ego of the vehicle; receiving dimension signals representing the dimensions of the vehicle; calculating whether the vehicle can collide with the object based on the received measurement signals and dimension signals; and outputting a result signal representing the result of calculating whether the vehicle can collide with the object in order to classify the relevance of the object with regard to a collision with the vehicle.

[0009] According to a second aspect, a device is provided which is set up to carry out all steps of the procedure according to the first aspect.

[0010] According to a third aspect, a computer program is provided which includes instructions that, when the computer program is executed by a computer, cause it to perform a procedure according to the first aspect.

[0011] According to a fourth aspect, a machine-readable storage medium is provided on which the computer program is stored according to the third aspect.

[0012] The invention is based on the understanding that the above problem can be solved by using vehicle dimensions and measured values, which can be measured simply, efficiently, and accurately using a conventional environmental sensor, to classify the object's relevance. These measured values ​​are the radial distance of the object relative to the environmental sensor and the radial relative velocity of the object relative to the environmental sensor, i.e., relative to the vehicle, insofar as the environmental sensor is encompassed by or arranged on the vehicle.

[0013] Furthermore, the classification of an object's relevance is carried out using the measured speed of the motor vehicle, whereby such speed can also be measured simply, efficiently and accurately.

[0014] This results in the particular technical advantage that the relevance of the object can be classified using easily obtainable values, in this case the measurement signals and the dimension signals (the dimensions of the motor vehicle are known quantities).

[0015] In comparison to the aforementioned state of the art, no complex classification approaches in a high-dimensional feature space are required, meaning that no complex assumptions about the nature or behavior of the object need to be made.

[0016] Accordingly, the concept according to the invention is advantageously particularly robust compared to the aforementioned prior art.

[0017] Furthermore, the concept according to the invention has the technical advantage that the radial distance and the radial relative velocity can already be measured with a simple and cost-effectively manufactured environmental sensor.

[0018] In summary, the technical advantage is thus achieved by providing a concept for efficiently classifying the relevance of an object located in the vicinity of a motor vehicle comprising an environment sensor, with regard to a collision with the motor vehicle.

[0019] According to the invention, the environment sensor is configured to measure the radial distance of the object and its radial relative velocity to the environment sensor, i.e., to the motor vehicle.

[0020] In one embodiment, the environmental sensor is designed for time-of-flight measurement. This means that the environmental sensor is configured to perform a time-of-flight measurement. In this case, the environmental sensor can also be referred to as a time-of-flight measurement sensor.

[0021] The environmental sensor is, for example, a radar sensor, a lidar sensor, or an ultrasonic sensor.

[0022] In one embodiment, the environmental sensor is a video sensor.

[0023] According to the invention, calculating whether the motor vehicle can collide with the object comprises calculating an uncertainty location value based on the received measurement signals, wherein the uncertainty location value provides location information regarding possible locations of the object, calculating a threshold value based on the received dimension signals, and comparing the uncertainty location value with the threshold value, such that the result depends on the comparison.

[0024] This results, for example, in the technical advantage that calculating whether the vehicle can collide with the object can be carried out efficiently. Here, the comparison allows for a simple yes / no answer as to whether the vehicle can collide with the object.

[0025] The invention also provides that the calculation of whether the motor vehicle can collide with the object is performed based on the following assumptions: The object is stationary, the time derivative of the yaw rate Ψ of the motor vehicle is zero, and the time derivative of the pitch rate φ of the motor vehicle is zero. The assumption that the object is stationary is particularly uncritical if the measured radial velocity at the measured location of the object corresponds sufficiently closely to the vehicle's own speed. Sufficiently closely here means, for example, within an error tolerance of less than or equal to 10%, less than or equal to 5%, or less than or equal to 1%, relative to the vehicle's own speed.According to one embodiment, the time derivatives of the yaw rate and pitch rate of the vehicle are determined by one or more electronic stability programs of the vehicle, for example, with sufficient accuracy. The validity of the assumptions can therefore also be easily verified.

[0026] This results, for example, in the technical advantage that calculating whether the motor vehicle can collide with the object can be carried out efficiently. In particular, this results in the technical advantage that, when using at least one of these assumptions, the calculation can be performed using analytically solvable equations.

[0027] According to the invention, a position signal is received, wherein the position signal represents a position of the environment sensor on the motor vehicle, and the calculation of whether the motor vehicle can collide with the object is carried out based on the received position signal.

[0028] This results, for example, in the technical advantage that the calculation of whether the vehicle can collide with the object can be carried out more efficiently. In particular, taking into account the position of the environmental sensor on the vehicle allows for a more accurate prediction of whether the vehicle can collide with the object.

[0029] According to the invention, the calculation of whether the motor vehicle can collide with the object is carried out based on all assumptions, wherein the dimensions of the motor vehicle include a width B and a height H, wherein the position of the environment sensor is defined by a height h above ground and a distance b off-center to a longitudinal axis of the motor vehicle, wherein the uncertainty location value according to d r 2 1 − v r v ego 2 is calculated, whereby the threshold is set according to max h 2 H − h 2 + max B 2 + b 2 B 2 − b 2 ) is calculated, whereby if the uncertainty location value is less than or less than equal to the threshold, the result is calculated to be that the motor vehicle cannot collide with the object, and if the uncertainty location value is greater than the threshold, the result is calculated to be that the motor vehicle can collide with the object.

[0030] This results, for example, in the technical advantage that the calculation of whether the vehicle can collide with the object can be carried out efficiently. According to this embodiment, possible locations of the object lie on a circle with a radius equal to the square root of the uncertainty value, the circle having a center located radially away from the installation location of the environmental sensor.

[0031] In another embodiment , which lies outside the claimed subject matter, it is provided that the measurement signals are a lateral displacement dy of the object measured by means of the environmental sensor. represent, wherein the calculation of whether the motor vehicle can collide with the object is carried out based on all assumptions, wherein the dimensions of the motor vehicle include a height H, wherein the position of the environment sensor is specified by a height h above ground, wherein the uncertainty location value according to d r 2 1 − v r v ego 2 − d y 2 is calculated, whereby the threshold is calculated according to max( h 2< , ( H - h ) 2< ) is calculated, whereby if the uncertainty location value is less than or less than equal to the threshold, the result is calculated to be that the motor vehicle cannot collide with the object, and if the uncertainty location value is greater than the threshold, the result is calculated to be that the motor vehicle can collide with the object.

[0032] This results, for example, in the technical advantage that calculating whether the motor vehicle can collide with the object can be carried out efficiently. According to this embodiment, the circle mentioned in connection with the preceding embodiment is reduced to two possible locations of the object. Thus, a more refined assessment of the object's relevance can advantageously be made.

[0033] Here, it is implicitly assumed that the relevance has already been determined via the lateral position and the formula is only applied if an object from the lateral position is already relevant and only the unknown elevation needs to be decided.

[0034] In one embodiment, the measurement signals represent a measured elevation offset with an error value, and the measured elevation offset is corrected based on the measured elevation offset and the uncertainty position value.

[0035] This results, for example, in the technical advantage that the estimation of elevation drift can be improved efficiently and significantly.

[0036] According to one embodiment, the environmental sensor is designed to map the environment or surroundings of the motor vehicle.

[0037] The environmental sensor is capable of measuring not only the (radial) distance of an object but also its radial relative velocity. Therefore, the following relationship exists between the measured quantities (radial distance and radial relative velocity), with the subsequent calculations performed in three-dimensional Cartesian space. In this respect, an x, y, z coordinate system is defined as follows: The x-axis of the coordinate system runs parallel to the longitudinal axis of the vehicle, the y-axis of the coordinate system runs perpendicular to the vehicle, the z-axis of the coordinate system runs perpendicular to the x- and y-axes, and the center of the coordinate system lies at the center of the environmental sensor.

[0038] The radial relative velocity vr of the object is therefore the scalar product of the object's relative position p in Cartesian coordinates (dx, dy, dz, coordinate origin at the sensor location) and the object's relative velocity vr in the same coordinate system (vx, vy, vz), normalized to the object's Cartesian (radial) distance dr. dx subsequently denotes the object's longitudinal position. dy subsequently denotes the object's transverse position. dz subsequently denotes the object's elevation position.

[0039] In the following, it is assumed that the object is a stationary object.

[0040] This assumption is justified, for example, for object speeds that are negligibly small relative to the speed of the motor vehicle.

[0041] According to the invention, the object is a stationary object.

[0042] Assuming that the object is stationary (its identification / filtering is possible with simple methods according to the current state of the art), the components of the relative velocity (vx, vy, vz) are completely determined or calculated by the movement of the motor vehicle, using, for example, the following approximations or...Assumptions are used, noting that the use of the approximation serves only to simplify the following steps: The relative velocity vr in the longitudinal direction (vx) corresponds to the negative self-velocity v ego of the vehicle, while the other two velocity components (vy, vz) result approximately from the negative rotation rates of the vehicle about its vertical axis (yaw rate, φ) and about its transverse axis (pitch rate, ω), which are converted from an angular velocity to a Cartesian velocity by multiplying them by the radial distance dr. Using a simple transformation for the radial distance (dr) and the longitudinal distance (dx), the following approximation applies for the radial relative velocity: . <menclose notation="topleft"> v r = 1 p v t p = : 1 d r d x v x + d y v y + d z v z < / menclose> v x ≈ − v ego v y ≈ − d r ∂ φ ∂ t v z ≈ − d r ∂ ω ∂ t d x = d r 2 − d y 2 − d z 2 = > d x d r = 1 − d y 2 + d z 2 d r 2 − v r ≈ 1 − d y 2 + d z 2 d r 2 v ego + d y ∂ φ ∂ t + d z ∂ ω ∂ t

[0043] To further simplify the presentation, only situations in which the vehicle moves at an approximately constant speed and without significant rotation rates will be considered in the following (this condition can be directly determined, for example, using signals from an ESP sensor in the vehicle). This means that it is assumed that the time derivatives of the vehicle's yaw rate, pitch rate, and velocity are all zero. In particular, the following applies: ∂ φ ∂ t = ∂ ω ∂ t = 0 Thus, the simple approximation follows: <menclose notation="topleft"> D 2 : = d y 2 + d z 2 ≈ d r 2 1 − v r v ego 2 < / menclose>

[0044] Accordingly, a mixed form D of the lateral displacement dy and the elevation displacement dz of the measured object (each relative to the position of the surrounding sensor) can be deduced solely from the measurement of the radial distance dr, the radial relative velocity vr, and the vehicle's own velocity v ego. D 2< here denotes the uncertainty position value described above or below.

[0045] Based on such a single measurement, the possible combinations of dy and dz in Cartesian space describe a circle (hereinafter also referred to as the uncertainty circle) with radius D and distance dr from the installation location of the environmental sensor. The required measured quantities can, for example, be determined using inexpensive radar sensors of minimal size, since no complex antenna structures are needed to determine the angle of incidence of the reflected signals.

[0046] In addition, the value of the relative velocity is independent of any possible rotation or misalignment of the environmental sensor, which further increases robustness.

[0047] Based on the value of D 2<, a very simple assessment of the object's relevance can already be made using the above spatial illustration.

[0048] A necessary condition for the collision with the object at a future time is that both its lateral displacement dy and elevation displacement dz lie within an area that corresponds to the dimensions of the motor vehicle (with width B, height H).

[0049] Assuming that the environmental sensor on the vehicle is mounted off-center by a distance b and at a height h above the ground, the vehicle can only collide with an object whose value for D 2< is less than or less than or equal to the following threshold: <menclose notation="bottomleft"> max h 2 H − h 2 + max B 2 + b 2 B 2 − b 2 < / menclose>

[0050] Conversely, if the value D 2< is greater than the aforementioned threshold, the motor vehicle cannot collide with the stationary object under consideration, which is why it does not need to be taken into account for the control of a lateral and / or longitudinal guidance of the motor vehicle.

[0051] This enables a robust pre-selection of relevant objects at the level of environmental sensor measurements, which, for example, reduces the computing time required in subsequent data processing layers of the system and allows for higher update rates or more cost-effective computing units.

[0052] If the environmental sensor can directly determine the lateral displacement dy of an object (in addition to the radial distance dr and the radial relative velocity vr), which is almost always the case for modern radar sensors, then the above approximation can be refined into an estimate for the (absolute) elevation displacement dz of the corresponding object. The circle of uncertainty in space from the above example now reduces to two possible point-like locations of the object.

[0053] The uncertainty value is calculated as follows: <menclose notation="topleft"> d z 2 ≈ d r 2 1 − v r v ego 2 − d y 2 < / menclose>

[0054] The threshold is calculated as follows: <menclose notation="topleft"> max h 2 H − h 2 < / menclose>

[0055] If the uncertainty value is less than or equal to the threshold, the result is calculated to be that the vehicle cannot collide with the object. If the uncertainty value is greater than the threshold, the result is calculated to be that the vehicle can collide with the object.

[0056] Even though the environmental sensor, due to its antenna structure, is unable to measure the elevation of an object, the method described here allows for a direct estimation of the absolute elevation deviation, at least for a stationary object. Based on this estimate, a more refined assessment of the relevance of the stationary object in question can be made, analogous to the procedure described above.

[0057] If the antenna structures of the environment sensor eventually also allow a measurement of the elevation offset, the described method can still significantly improve the estimation of this elevation, since the quantities required for this are often available with higher accuracy than the direct measurement of the elevation via the antenna structure of the radar sensor allows.

[0058] According to one embodiment, the method according to the first aspect is carried out or performed by means of the device according to the second aspect.

[0059] Process characteristics result directly from corresponding device characteristics and vice versa.

[0060] This means, in particular, that technical functionalities of the device result analogously from corresponding technical functionalities of the process and vice versa.

[0061] The invention is explained in more detail below with reference to preferred embodiments. These include: Fig. 1 a flowchart of a procedure for classifying the relevance of an object, Fig. 2 a device that is set up to carry out a method for classifying the relevance of an object, Fig. 3 a machine-readable storage medium and Fig. 4 a motor vehicle.

[0062] Fig. 1 shows a flowchart of a procedure for classifying the relevance of an object located in the vicinity of a motor vehicle comprising an environment sensor with regard to a collision with the motor vehicle, comprising the following steps: Receiving 101 measurement signals representing a radial distance dr of the object relative to the environmental sensor measured by the environmental sensor, a radial relative velocity vr of the object relative to the environmental sensor measured by the environmental sensor, and a measured own velocity v ego of the motor vehicle; Receiving 103 dimension signals representing dimensions of the motor vehicle; Calculating 105 whether the motor vehicle can collide with the object based on the received measurement signals and based on the received dimension signals; Outputting 107 a result signal representing a result of the calculation of whether the motor vehicle can collide with the object in order to classify the relevance of the object with regard to a collision with the motor vehicle.

[0063] Fig. 2Figure 201 shows a device that is set up to perform all the steps of a procedure for classifying the relevance of an object.

[0064] For example, device 201 is designed to be in Fig. 1 to perform the procedures shown.

[0065] The device 201 comprises an input 203 for receiving measurement signals representing a radial distance dr of the object relative to the environmental sensor measured by means of the environmental sensor, a radial relative velocity vr of the object relative to the environmental sensor measured by means of the environmental sensor and a measured intrinsic velocity v ego of the motor vehicle, and for receiving dimension signals representing dimensions of the motor vehicle.

[0066] The device 201 includes a processor 205 for calculating whether the motor vehicle can collide with the object, based on the received measurement signals and based on the received dimension signals.

[0067] The device 201 further includes an output 207 for outputting a result signal, which represents a result of calculating whether the motor vehicle can collide with the object, in order to classify the relevance of the object with regard to a collision with the motor vehicle.

[0068] According to one embodiment, several processors are provided to calculate whether the motor vehicle can collide with the object.

[0069] Fig. 3 Figure 3 shows a machine-readable storage medium 303 on which a computer program 303 is stored, wherein the computer program 303 comprises instructions which, when the computer program is executed by a computer, for example by the device 201 of the Fig. 2, cause this, all steps of a procedure for classifying the relevance of an object, for example, all steps of the in Fig. 1 to carry out the procedure shown.

[0070] Fig. 4 shows a motor vehicle 401.

[0071] The motor vehicle 401 includes an environment sensor 403. The environment sensor 403 is, for example, a radar sensor or a lidar sensor.

[0072] The motor vehicle 401 also includes the device 201 according to Fig. 2 .

[0073] The environment sensor 403 detects, for example, the area surrounding the motor vehicle. When an object is detected in the detected area, the environment sensor 403 can measure the radial distance of the object to the environment sensor 403 as well as the radial relative velocity of the object relative to the environment sensor, i.e., to the motor vehicle 401, provided that the environment sensor 403 is located on the motor vehicle 401.

[0074] Measurement signals corresponding to this measurement are sent to input 203 of the device 201. Input 203 receives these measurement signals and also receives further measurement signals representing a measured self-speed of the motor vehicle 401. These measurement signals and the further measurement signals are therefore measurement signals received by the processor 205 and represent the radial distance dr of the object relative to the environmental sensor 403, measured by the environmental sensor 403; the radial relative velocity vr of the object relative to the environmental sensor 403, measured by the environmental sensor 403; and the measured self-speed v ego of the motor vehicle 401.

[0075] The processor 205 calculates, as described above and / or below, whether the motor vehicle can collide with the object.

[0076] The processor 205 generates a result signal from the calculation, which is output via output 207.

[0077] For example, the result signal is output to a control unit 405 of the motor vehicle 401.

[0078] According to one embodiment, the control device 405 is designed to control the lateral and / or longitudinal guidance of the motor vehicle 401 based on the output result signal.

[0079] In summary, a concept is provided for efficiently classifying the relevance of an object located in the vicinity of a motor vehicle equipped with an environmental sensor, with regard to a collision with the motor vehicle. The concept according to the invention is not based, among other things, on temporal filtering of measured quantities or hypotheses regarding the formation of objects, but rather, in particular, directly on basic measured quantities of an environmental sensor, which guarantees a high degree of general applicability and robustness. The basic measured quantities used, i.e., the measured radial distance and the measured radial relative velocity, can advantageously already be provided by an environmental sensor that can be designed very cost-effectively.

Claims

1. Method for classifying a relevance of an object, which is located in an environment of a motor vehicle (401) comprising an environment sensor (403), with regard to a collision with the motor vehicle (401), comprising the following steps: receiving (101) measurement signals which represent a radial distance dr of the object relative to the environment sensor (403), as measured by means of the environment sensor (403), a radial relative speed vr of the object relative to the environment sensor (403), as measured by means of the environment sensor (403), and a measured ego speed vego of the motor vehicle (401), receiving (103) dimension signals representing dimensions of the motor vehicle (401), calculating (105) whether the motor vehicle (401) can collide with the object, based on the received measurement signals and based on the received dimension signals, outputting (107) a result signal representing a result of calculating whether the motor vehicle (401) can collide with the object, in order to classify the relevance of the object with respect to a collision with the motor vehicle (401), characterized in that the calculation (105) of whether the motor vehicle (401) can collide with the object comprises calculating an uncertainty location value based on the received measurement signals, wherein the uncertainty location value indicates location information relating to possible locations of the object, calculating a threshold value based on the received dimension signals and comparing the uncertainty location value with the threshold value, such that the result depends on the comparison, wherein the calculation (105) of whether the motor vehicle (401) can collide with the object is carried out on the basis of the following assumptions: the object is a stationary object, a time derivative of a yaw rate Ψ of the motor vehicle (401) is zero, a time derivative of a pitch rate φ of the motor vehicle (401) is zero, and a time derivative of the ego speed of the motor vehicle is zero, wherein a position signal is received, wherein the position signal represents a position of the environment sensor (403) on the motor vehicle (401), wherein the calculation (105) of whether the motor vehicle (401) can collide with the object is carried out based on the received position signal, wherein the dimensions of the motor vehicle (401) comprise a width B and a height H, wherein the position of the environment sensor (403) is specified by a height h above ground and a distance b off-centre with respect to a motor vehicle longitudinal axis, wherein the uncertainty location value is calculated according to d r 2 1 − v r v ego 2 , wherein the threshold value is calculated according to max h 2 H − h 2 + max B 2 + b 2 B 2 − b 2 , wherein, if the uncertainty location value is less than or less than or equal to the threshold value, it is calculated as the result that the motor vehicle (401) cannot collide with the object, wherein, if the uncertainty location value is greater than the threshold value, it is calculated as the result that the motor vehicle (401) may collide with the object.

2. Device (201) which is configured to carry out all steps of the method according to Claim 1.

3. Computer program (303) comprising instructions which, when the computer program (303) is executed by a computer, cause the latter to perform a method according to Claim 1.

4. Machine-readable storage medium (301) on which the computer program (303) according to Claim 3 is stored.

Images