Method for determining radial velocity of radar target

Abstract

The invention relates to a method for determining a radial velocity vr, og of a radar target (16) with respect to the ground by means of an angle-resolving radar sensor (12) mounted on a moving platform (10), comprising the following steps: creating a velocity profile of a fixed environment of the radar sensor (12) and determining parameters (C, S) of the velocity profile; measuring a radial relative speed vr, rel of the radar target (16); a correction amount is calculated for taking into account the own movement of the platform (10). According to the invention, the value of the radial speed vr, og with respect to the ground, which is corrected in accordance with the own movement, is calculated directly on the basis of angle information (theta) with respect to the radar target (16) and on the basis of the measured relative speed vr, rel and the parameters (C, S) of the speed profile.

Description

Technical Field

[0001] The invention relates to a method for determining the radial velocity of a radar target relative to the ground by means of an angle-resolving radar sensor, which is mounted on a moving platform, the method comprising the following steps:

[0002] - creating a velocity profile of the stationary environment of the radar sensor and determining parameters of the velocity profile,

[0003] - measure the radial relative velocity of radar targets, and

[0004] - Calculate corrections to account for the platform's own motion. Background Art

[0005] In highly automated vehicles, a large number of different types of sensors are used to create an image of the surrounding environment. This is called perception. The surrounding environment consists of the static world, i.e. fixed environment (infrastructure, vegetation, etc.) and moving objects (vehicles, pedestrians, etc.).

[0006] In order to achieve reliable perception, radar sensors are used. These radar sensors can be used advantageously, in particular, at night, in fog, in water and in the rain. They emit electromagnetic waves, which are reflected and received by the sensor again. As a result, positioning (reflection) is generated on static objects and moving targets. The measurement space is polar space, and the positioning essentially consists of distance, radial relative speed, azimuth and, in the case of new radar sensors, also the elevation angle of the target. In addition, the radar cross section that (simplified) describes the reflectivity of the target is evaluated.

[0007] In order to distinguish between static positioning and dynamic (i.e. moving) positioning, the radial velocity about the ground is calculated. For this purpose, in addition to the measured radial relative velocity of the target, the velocity of the radar sensor about the ground and the normalized direction vector from the sensor pointing positioning are also required. Traditionally, the odometer signal is used to describe the movement of the vehicle or more generally the platform on which the radar sensor is mounted. The radial velocity about the ground is finally used to track the movement of the target.

[0008] A disadvantage of the known methods is that, in addition to the radar, at least one further (odometer) sensor is also required. Furthermore, the odometer information must be sufficiently accurate and free of delay.

[0009] Therefore, various methods have been studied to determine the speed of the platform with respect to the ground based solely on radar positioning. In "Kellner et al: Instantaneous Ego-Motion Estimation using Doppler Radar", proceedings-IEEE International Conference on Robotics and Automation, October 2013, DOI: 10.1109 / ITSC.20136728341, a method of the type mentioned at the beginning is described, which makes full use of such a scenario, that is, for fixed radar targets, the measured radial relative velocity is related to the azimuth of the target in a representative manner. If the radial velocity is plotted against the azimuth for multiple fixed targets, a cosine curve is obtained, but if the optical axis of the radar sensor is not parallel to the direction of motion of the platform, the cosine curve must not necessarily have its maximum value at the azimuth of 0 °. If a larger population consisting of fixed targets is positioned, the radial velocity is on such a curve: the curve can be described as a linear combination of the cosine function and the sine function of the azimuth. The coefficients of this linear combination are then parameters that mark the curve, the so-called velocity profile.

[0010] These parameters can be determined by regression if the sampling is large enough. In this case, moving objects lead to non-embedding in this pattern and damage the positioning of the image. However, ambiguity only occurs if a larger number of movable objects all have the same velocity vector by chance, so that their radial velocities also lie on the cosine curve. However, such a form is extremely unlikely in practice and is also unstable in time.

[0011] However, when trying to determine the platform's own motion on the basis of the velocity profile, there is the following problem: the platform (when moving in a plane) has three degrees of freedom of motion, namely two translational degrees of freedom and one rotational degree of freedom, while the velocity profile is represented by only two parameters (coefficients). Therefore, additional information is required to unambiguously determine the velocity of the platform relative to the ground. If the case where the yaw velocity of the platform is zero cannot be ruled out, the velocity profile in the static world depends not only on the orientation of the optical axis of the radar sensor, but also on the location of the radar sensor on the platform, so that additional, possibly erroneous information is required.

[0012] In the method described in the above-mentioned source, the movement of the platform is described by a single track model, the so-called Ackermann model, which is based on the assumption that the wheels of the vehicle (or platform) do not have drift. In practice, this condition is often but not always met. Other possibilities for determining the platform's own movement are tracking of radar targets over multiple measurement cycles, which, however, may lead to greater delays in perception, or using multiple radar sensors on the same platform to locate the same target, which, however, requires additional sensors. Summary of the invention

[0013] The object of the present invention is to specify a method for determining the radial velocity of a radar target relative to the ground which requires less additional information or additional assumptions and is therefore less prone to errors.

[0014] According to the invention, this object is achieved in that a value of the radial velocity relative to the ground that is corrected for the own motion is directly calculated based on the angle information about the radar target and based on the measured relative velocity and velocity profile parameters.

[0015] The invention is based on the recognition that the parameters of the velocity profile make it possible to compensate for the effects of the platform's own motion without knowing all the degrees of freedom of this own motion. The method makes it possible to determine the radial velocity of a radar target with respect to the ground within a single measurement cycle of a single radar sensor without having to track the radar target over a longer period of time. In this case, no odometer data is required, so that costs for odometer sensors can be saved and possible sources of error are cut off. A further advantage is that no demanding algorithms for odometer evaluation are required and no delay problems arise. As long as an odometer sensor is present in the vehicle, an additional redundant path is established by the invention.

[0016] Advantageous embodiments and further developments of the invention result from the subject matter of the expanded claim.

[0017] In a simple implementation that does not consider the elevation angle of the radar target, the radial velocity of the moving target with respect to the ground can be calculated as follows: a velocity distribution map of the static world is created, and the parameters C and S are determined by regression, which are calculated according to the formula v r = C cos(θ) + S sin(θ) The distribution diagram is described as azimuth θ (v r ) and then from the measured radial relative velocity v r,rel Subtract this quantity C cos(θ)+S sin(θ) from θ.

[0018] If the radar sensor is also able to measure the elevation angle of the radar target, a corresponding velocity profile can also be created in the elevation angle and used to increase redundancy or also to compensate for the platform's own movements in the vertical direction (for example in hilly terrain). BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The embodiments are further explained below with reference to the accompanying drawings. The accompanying drawings show:

[0020] Figure 1-3 : A diagram to explain the velocity profiles for different attitudes of the radar sensor on the platform;

[0021] Figure 4 : Example of velocity profile;

[0022] Figure 5 and 6 : A diagram for deriving radial relative velocity for different target points of a rigid object implementing superimposed translational and rotational motions; and

[0023] Figure 7 : Similar to Figure 6 , however in this diagram the rigid object is formed by the entire static world and contains the position of the sensor on the platform. DETAILED DESCRIPTION

[0024] exist Figure 1 Schematically shown is a platform 10, for example a motor vehicle, whose velocity relative to the ground is given by the velocity vector v. A radar sensor 12 is mounted centrally on the front side of the platform 10, whose optical axis 14 extends in the direction of motion of the platform. The radar sensor locates a non-moving radar target 16 at an azimuth angle θ. The radar target 16 moves at a relative velocity v relative to the platform 10. rel = -v motion. The line of sight 18 from the radar sensor 12 to the radar target 16 and the velocity vector v rel The angle θ is also formed. Therefore, the radial component v of the relative velocity in the direction of the line of sight 18 is r By v r =|v rel |cos(θ) is given. If the relative speed of multiple stationary targets at different azimuth angles is measured, the velocity profile has the form of a cosine curve with a maximum value at an azimuth angle of 0°.

[0025] Figure 2 A similar diagram is shown for the case where the radar sensor 12 is mounted on the side of the platform 10 and its optical axis 14 extends at right angles to the direction of motion. The line of sight 18 again forms an angle θ with the optical axis 14, which is also reflected by the velocity vector v relThe vector decomposition of the right triangle appears. Therefore, in this case, the radial component of the relative velocity v r By v r =|v rel |sin(θ) is given.

[0026] If a velocity profile of a plurality of stationary objects is recorded in such an arrangement, the profile takes the form of a sinusoidal curve.

[0027] Figure 3 The following situation is illustrated: the radar sensor 12 is mounted at a corner of the platform 10 and its optical axis 14 forms an angle φ with the direction of motion of the platform. The line of sight 18 to the radar target 16 again forms an angle θ with the optical axis 14. Then the relative velocity vector v rel Angle with line of sight 18 So for the radial component of the relative velocity v r get:

[0028]

[0029] By applying the addition theorem we obtain:

[0030]

[0031] use and Get the velocity profile:

[0032] v r =|v rel |((C cos(θ)–S sin(θ)),

[0033] The velocity profile is characterized by the parameters C and S.

[0034] If the relative speeds of a plurality of radar targets are measured at different azimuth angles θ in such an arrangement, the speed profile is a cosine curve whose maximum value is offset from the angle θ=0°.

[0035] exist Figure 4In FIG. 2 , curve 20 is an example for such a velocity profile. In addition, some measuring points 22 and 24 are shown here. Measuring points 22 represent static targets and are concentrated on curve 20, while measuring points 24 represent moving targets, which are more strongly divergent and are mostly located next to curve 20. Since the number of static targets is usually significantly larger than the number of moving targets, measuring points 22 can already identify the trend of curve 20 very well. At this time, parameters C and S can be selected for curve 20 so that the best possible agreement with the position of measuring points 22 is obtained, for example according to the minimum distance square method. In this way, the parameters C and S of the velocity profile can be determined only by using radar measurements performed by a single radar sensor in a single measurement cycle.

[0036] The physical meanings of the parameters C and S are explained below.

[0037] exist Figure 5 Schematically shows a rigid object 26, which has a plurality of target points R and P that can be located by the radar sensor 12. It is assumed here that the object 26 moves with a vector velocity v relative to the ground. R Any arbitrary target point can be selected as a reference point for tracking the translational movement of the object 26. In the example shown here, the target point R is selected. As long as the object 26 does not perform a rotational movement (about a vertical axis, at right angles to the xy plane), all target points, i.e. also the points P, move with the same velocity vector v R Assume that the radar sensor 12 is very far from the rigid object 26, so that the radar sensor Figure 5 However, a line of sight 28 is visible, which leads from the radar sensor to the target point R and forms an angle θ with the x-axis (in the direction of motion of the platform 10). The velocity vector v R can be decomposed into its x components v by a vector Rx and its y component v Ry Each of these components can be further vector-decomposed into a component parallel to the line of sight 28 and a component at right angles to the line of sight. The velocity v of the target point R about the ground R,r The sum of the components parallel to the line of sight 28 gives:

[0038] v R,r = v Rx cos(θ) + v Ry sin(θ) (1)

[0039] If we first assume that the platform 10 and the radar sensor 12 are stationary, then v R,r At the same time, it is the radial relative velocity v of the target point R relative to the radar sensor R,rel .

[0040] exist Figure 6 The following is illustrated in FIG: The translational movement of the object 26 - and therefore the translational movement of the target point R - is superimposed on a rotational movement about a vertical axis through the target point R. The target point P moves relative to the target point R at a speed v P Movement. If p is the position vector from target point R to target point P, then v P The vector product of the (vertical) angular velocity vector ω and the position vector p gives:

[0041] v P = ω × p = (ωp y , -ωp x ) (2)

[0042] exist Figure 6 In the example in , the rotation is in the mathematically positive direction (counterclockwise), so that the z component of ω is positive. In this example, the y component of p is negative, and so v P The x component of p is also negative, as can be seen in the figure. The x component of p is positive, and therefore v P The y component of the vector v is negative ( P pointing to the left).

[0043] In order to obtain the velocity v of the target point P about the ground P,og , we must transform the vector v R ( Figure 5 ) and v P ( Figure 6 ) add:

[0044] v P,og = v R + ω × v P = (v Rx +ωp y , v Ry -ωp x ) (3)

[0045] In order to obtain the radial component v of the motion of the target point P along the line of sight from the radar sensor 12 to the target point P P,r , must be for the vector v P,og Executed in Figure 5 For vector v R The vector decomposition shown (see equation (1)):

[0046] v P,r = (v Rx +ωp Rx )cos(θ) + (v Rx -ωp Rx )sin(θ) (4)

[0047] If it is assumed that radar sensor 12 is stationary, then the velocity v with respect to the ground P,r The relative speed v of the target point P relative to the radar sensor 12 P,rel This relative speed can be calculated based on Figure 3 is described as a linear combination of cosine and sine functions:

[0048] v P,rel = (C cos(θ) + S sin(θ)) (5)

[0049] Comparing with the coefficients of equation (2), we can obtain:

[0050] C = v Px +ωp y S = v Py -ωp x (6)

[0051] If we understand C and S as vectors v with z component 0 rel The x-component and y-component of v show the physical meaning of these parameters. Comparison with equation (3) shows that v rel can be described as the velocity vector v P And the sum of the "cross product of the angular velocity vector ω and the position vector p":

[0052] v rel = v P + ω × p (7)

[0053] This means that C and S are the velocity vector v rel , which describes the velocity of target point P at the location of radar sensor 12 in the coordinate system of (stationary) platform 10 .

[0054] exist Figure 7 middle, Figure 5 and Figure 6 The rigid object 26 shown in is expanded to a fictitious rigid object 26', which consists of the entire static world and therefore also includes the platform 10 and the sensor 12. It is now assumed that the rigid object 26' is stationary and the platform 10 is moving. Here, the point that is instantaneously located at the position of the radar sensor 12 is considered as the target point P. The velocity of this point relative to the platform is given by equation (7). The parameters C and S of the velocity profile are thus directly explained as follows

[0055] Components of the velocity vector: The target point P at the position of the radar sensor 12 relative to the platform 10

[0056] Move with this velocity vector.

[0057] Then, in the coordinate system of the static world, radar sensor 12 has a velocity v relative to the (stationary) target point P, ie, relative to the ground. S,og =-v rel =(−C, −S, 0) That is, in order to determine the self-motion of the radar sensor 12, it is not necessary to know the deflection velocity ω nor the position of the radar sensor 12 on the platform 10, but it is sufficient to know the parameters C and S.

[0058] In the xy plane, radar targets, such as Figure 3 The velocity of the radar target 16 with respect to the ground and the relative motion v of the radar target with respect to the platform 10 rel The sum of the radar sensor's own motion (-C, -S) is equal. Then for the radial component of the radar target, that is, the radial velocity v r,og Applicable:

[0059] v r,og = v r,rel - C cos(θ) - S sin(θ) (8)

[0060] This equation allows the radial velocity of the radar target with respect to the ground to be determined solely from the parameters C and S and the measured relative velocity and the measured azimuth.

[0061] If the radar sensor 12 is also angle-resolving in elevation, an additional velocity profile can be created for the elevation angles. As long as the radar target does not make a significant vertical movement, the radar sensor's own movement can be determined in an independent manner based on this profile and aligned with the own movement determined based on the azimuth. If the vertical movement of the platform and the radar target is not negligible in hilly terrain, the vertical movement of the radar sensor can also be determined similarly to the method described here, so that the influence of this vertical movement can also be taken into account when determining the radial velocity of the radar target relative to the ground.

Claims

1. For determining the radial velocity v of a radar target (16) relative to the ground by means of an angle-resolving radar sensor (12) mounted on a moving platform (10) r,og The method comprises the following steps: - creating a velocity profile of the stationary environment of the radar sensor (12) and determining parameters (C, S) of the velocity profile, - measuring the radial relative velocity v of the radar target (16) r,rel as well as - calculation of corrections for taking into account the intrinsic motion of the platform (10), It is characterized in that Based on the angle information (θ) about the radar target (16) and based on the measured relative speed v r,rel The radial velocity v with respect to the ground, which is corrected for the own motion, is directly calculated from the parameters (C, S) of the velocity profile. r,og The value of .

2. Used to determine the radial velocity v about the ground r,og The velocity profile is given by a function v of the azimuth angle θ. r =C cos(θ)+S sin(θ), where, The coefficients C and S are parameters of the velocity profile, and in this method, the radial velocity v of the radar target with respect to the ground r,og Calculated according to the following formula: v r,og =v r,rel -C cos(θ)-S sin(θ).

Images