DE102017215056A1 - Method for detecting crossing pedestrians, driver assistance system and vehicle - Google Patents

Abstract

The invention relates to a method for detecting pedestrians (3) crossing in front of a vehicle (2), comprising the steps of: determining (S1) a microdoppler signal; and performing (S2) a plausibility check as to whether the microdoppler signal is assignable to a pedestrian (3) crossing ahead of the vehicle (2) by evaluating a power of the microdoppler signal and / or a microdoppler angle (α) of the microdoppler signal ,

Description

The present invention relates to a method for detecting pedestrians cruising in front of a vehicle, a driver assistance system for a vehicle for recognizing crossing pedestrians and a vehicle having a driver assistance system.

State of the art

Radar sensors prove to be beneficial in detecting pedestrians as they are less affected by external environmental conditions such as rain, fog or snow than, for example, cameras. In addition, modern vehicles already have standard radar sensors, which can thus be used without additional hardware effort for pedestrian detection.

If the radar waves emitted by the radar sensors are reflected by a moving object, a Doppler shift occurs, which can be evaluated by measuring the received reflected radar signal. As a result, speeds and general movements of objects can be detected.

Periodic motions, such as the typical pendulum movements of a pedestrian's arms or legs, produce significant patterns in the measured signals which interfere with the Doppler signal resulting from the total or centroid motion of the objects. This so-called μDoppler effect or microdoppler effect thus results from the simultaneous occurrence of different radial velocities, for example due to different speeds of the torso and the arms or legs of the pedestrian. The use of the microdoppler effect for gesture recognition is for example from the document WO 2012/152476 A1 known.

Of particular importance is the detection of intersecting pedestrians. For the purposes of this invention, an intersecting movement is to be understood as meaning that an interpolated trajectory of the pedestrian intersects with an interpolated trajectory of the vehicle. A crossing movement may be present, in particular, when the direction of movement of the pedestrian intersects with the direction of movement of the vehicle. The pedestrian thus moves, for example, in front of the vehicle from a first lane edge to an opposite second lane edge. In particular, the pedestrian can move perpendicular to the direction of movement of the vehicle in front of the vehicle.

Disclosure of the invention

The invention provides a method for detecting pedestrians crossing in front of a vehicle with the features of patent claim 1, a driver assistance system for a vehicle for recognizing crossing pedestrians with the features of patent claim 9 and a vehicle with the features of patent claim 10.

The invention accordingly provides, according to a first aspect, a method for recognizing pedestrians crossing in front of a vehicle. For this purpose, a microdoppler signal is determined and then a plausibility check is carried out as to whether the microdoppler signal can be assigned to a pedestrian crossing the vehicle. For plausibility, a power of the microdoppler signal and / or a microdoppler angle of the microdoppler signal are evaluated.

According to a second aspect of the invention, therefore, a driver assistance system for a vehicle for recognizing crossing pedestrians is provided. The driver assistance system comprises a radar device, which is designed to determine microdoppler signals on the basis of the radar device emitted and received radar signals. The driver assistance system further comprises an evaluation device which performs a plausibility check by evaluating a power of a microdoppler signal determined by the radar device and / or a microdoppler angle of the microdoppler signal as to whether the microdoppler signal can be assigned to a pedestrian crossing the vehicle.

According to a third aspect, the invention relates to a vehicle with a driver assistance system.

Preferred embodiments are the subject of the respective subclaims.

Advantages of the invention

On the one hand, the invention makes it possible to distinguish pedestrians from other objects. The invention makes use of the fact that the typical movements of a pedestrian, d. H. the pendulous movements of the arms or legs cause a typical behavior of the microdoppler signals. As a result, a pedestrian can be distinguished from other objects, which can also cause microdoppler signals. Such further objects may be objects with typical periodic movement patterns, such as leaves, flags or car tires, but the corresponding microdoppler signals differ from the pedestrian-induced microdoppler signals, in particular by the time course of the power. By evaluating the power or the time profile of the power of the microdoppler signal thereby a detection of pedestrians is possible.

On the other hand, the invention makes it possible to determine the exact direction of movement of the pedestrian relative to the vehicle. Thus, it is possible, in particular under observation of the microdoppler angle, to distinguish pedestrians crossing pedestrians, who move away from the vehicle or move parallel to the vehicle.

The invention thus increases safety by allowing faster and earlier detection of intersecting pedestrians. At the same time erroneous identifications can be avoided, such as objects that are not pedestrians, or pedestrians who do not cross in front of the vehicle.

According to a preferred development of the method, sensor data are evaluated in order to recognize an object. Furthermore, the object is assigned an object angle. The microdoppler angle is compared to the object angle to detect if the microdoppler signal is attributable to the detected object. The additional sensor data can be generated, for example, by further sensors of the vehicle, such as radar sensors, vehicle cameras, ultrasound sensors or lidar sensors. In particular, objects detected using this sensor data can already be identified as pedestrians if the microdoppler angle substantially corresponds to the object angle. Thus, the additionally determined sensor data can be used to check the plausibility of whether the microdoppler signal can actually be assigned to a pedestrian.

To determine whether the microdoppler angle substantially corresponds to the object angle, an absolute distance between the object angle and the microdoppler angle can be calculated. In particular, the absolute distance between the object angle and the microdoppler angle can be compared with a predetermined threshold value. Thus, according to one embodiment, a continuous function may be calculated, which depends on the amount of distance between the object angle and the microdoppler angle. If the difference between the object angle and the microdoppler angles is greater than the predetermined threshold, then the function may have a negative value. Conversely, if the absolute distance between the object angle and the microdoppler angle is less than the predetermined threshold, the function may have a positive value. The threshold value may for example have a value of 2 ° to 10 °, preferably a value between 3 ° and 4 °. The sign of the function can be used for plausibility. A positive value corresponds to a small deviation between the object angle and the microdoppler angle, so that it can be detected that the microdoppler signal can be assigned to the detected object. Conversely, a negative value corresponds to a larger deviation between the object angle and the microdoppler angle, so that with a certain probability the microdoppler signal can not be assigned to the detected object.

According to a preferred embodiment of the method, a temporal change in the amount of distance between the object angle and the microdoppler angle is evaluated. For example, if the microdoppler angle diverges increasingly from the object angle, it can be seen that the microdoppler signal can not be assigned to the object. If the absolute distance between the microdoppler angle and the object angle remains inversely small, ie in particular smaller than the predetermined threshold value, it can be assumed with a greater probability that the microdoppler signal can be assigned to the detected object. In particular, the above-described function of the absolute distance between the object angle and the microdoppler angle can be summed or integrated up in order to obtain a cumulative plausibility function. The value of the cumulative plausibility function is used to detect whether the microdoppler signal can be assigned to the detected object. If the cumulative plausibility function assumes a large positive value, the difference between the object angle and the microdoppler angle is also small in the course of time compared to the predetermined threshold, so that it can be recognized that the microdoppler signal can be assigned to the detected object. If the cumulative Conversely, if the plausibility function assumes a large negative value, it is recognized that the microdoppler signal can not be assigned to the detected object.

According to a preferred embodiment of the method, the temporal change of the microdoppler angle is evaluated in order to detect a crossing movement. In particular, it can be seen that the microdoppler signal is attributable to a crossing movement if the microdoppler angle decreases. When viewed from the vehicle, the pedestrian thus moves toward the direction of movement of the vehicle or to the expected trajectory of the vehicle. In this case, an intersecting movement is detected, whereby a risk of collision is increased.

According to a preferred development of the method, variations in the power of the microdoppler signal are evaluated in order to detect whether the microdoppler signal can be assigned to arm or leg movements which are typical for a pedestrian. In particular, the variations in power over time can be considered. The microdoppler signal is assigned to a pedestrian only if the magnitude of the variations in power exceeds a predetermined threshold. This makes it possible, for example, to distinguish movements of pedestrians from movements of leaves, since the latter cause smaller fluctuations in the performance of the microdoppler signal. This can ensure that the microdoppler signal is actually caused by a pedestrian.

list of figures

• 1 ein schematisches Blockdiagramm eines Fahrerassistenzsystems gemäß einer Ausführungsform der Erfindung;
• 2 ein schematisches Blockschaltbild eines Fahrzeugs gemäß einer Ausführungsform der Erfindung;
• 3 ein schematisches Verkehrsszenario, wobei sich zwei Fußgänger entlang der Erfassungsrichtung des Radars befinden;
• 4 beispielhafte Verläufe eines Objekt-Winkels und eines Mikrodoppler-Winkels im Falle eines kreuzenden Fußgängers;
• 5 beispielhafte Verläufe eines Objekt-Winkels und eines Mikrodoppler-Winkels für ein Verkehrsszenario;
• 6 ein beispielhafter Verlauf einer Hilfsfunktion zur Auswertung des Abstandes zwischen Objekt-Winkel und Mikrodoppler-Winkel,
• 7 bis 10 beispielhafte Verläufe von Objekt-Winkel, Mikrodoppler-Winkel und ersten Plausibilitätsfunktionen, falls das Mikrodoppler-Signal dem Objekt zuordenbar ist;
• 11 bis 14 beispielhafte Verläufe von Objekt-Winkel, Mikrodoppler-Winkel und ersten Plausibilitätsfunktionen, falls das Mikrodoppler-Signal nicht dem Objekt zuordenbar ist;
• 15 bis 17 beispielhafte Verkehrsszenarien mit kreuzenden und nicht kreuzenden Fußgängern;
• 18 beispielhafte Verläufe eines Mikrodoppler-Winkels für kreuzende und nicht kreuzende Fußgänger;
• 19 bis 22 beispielhafte Verläufe von Mikrodoppler-Winkel und zweiten Plausibilitätsfunktionen im Falle kreuzender Fußgänger;
• 23 bis 26 beispielhafte Verläufe von Mikrodoppler-Winkel und zweiten Plausibilitätsfunktionen im Falle nicht kreuzender Fußgänger;
• 27 ein schematischer Verlauf einer Mikrodoppler-Leistung für Fußgänger und Objekte, welche keine Fußgänger sind; und
• 28 ein Flussdiagramm eines Verfahrens zum Erkennen von kreuzenden Fußgängern gemäß einer Ausführungsform der Erfindung.

Show it:

• 1 a schematic block diagram of a driver assistance system according to an embodiment of the invention;
• 2 a schematic block diagram of a vehicle according to an embodiment of the invention;
• 3 a schematic traffic scenario, with two pedestrians are along the detection direction of the radar;
• 4 exemplary courses of an object angle and a microdoppler angle in the case of a crossing pedestrian;
• 5 exemplary histories of an object angle and a microdoppler angle for a traffic scenario;
• 6 an exemplary course of an auxiliary function for the evaluation of the distance between object angle and microdoppler angle,
• 7 to 10 exemplary gradients of object angle, microdoppler angle, and first plausibility functions if the microdoppler signal is assignable to the object;
• 11 to 14 exemplary gradients of object angle, microdoppler angle, and first plausibility functions if the microdoppler signal can not be assigned to the object;
• 15 to 17 exemplary traffic scenarios with intersecting and non-intersecting pedestrians;
• 18 exemplary courses of a microdoppler angle for pedestrians crossing and not crossing;
• 19 to 22 exemplary courses of microdoppler angle and second plausibility functions in case of crossing pedestrians;
• 23 to 26 exemplary courses of microdoppler angle and second plausibility functions in the case of non-crossing pedestrians;
• 27 a schematic course of a microdoppler performance for pedestrians and objects that are not pedestrians; and
• 28 a flowchart of a method for detecting crossing pedestrians according to an embodiment of the invention.

In all figures, the same or functionally identical elements and devices are provided with the same reference numerals. The numbering of method steps is for the sake of clarity and generally should not imply a particular chronological order. In particular, several method steps can be carried out simultaneously.

Description of the embodiments

1 shows a schematic block diagram of a driver assistance system 1 for a vehicle. The driver assistance system 1 may be located entirely within the vehicle, however, individual functions may be outsourced. In particular, certain calculation and evaluation operations can be performed on an external server.

The driver assistance system 1 includes a radar device 11 , which is adapted to emit and receive radar beams. The radar device 11 For this purpose, at least 1 and preferably a plurality of transceiver devices 111 exhibit. The transceiver devices 111 are designed to emit and receive radar signals according to a known radar method. The transceiver devices 111 For example, they can be configured as continuous wave radar (CW radar) or as frequency-modulated continuous wave radar (FMCW radar). The radar device 11 is further hardware or software side designed to evaluate the received reflected radar signals. The radar device 11 is particularly adapted to detect microdoppler signals.

The driver assistance system 1 also has an evaluation device 12 which is determined by the radar device 11 detected microdoppler signals detects whether the microdoppler signal is attributable to a crossing in front of the vehicle pedestrian. The evaluation device 12 can use the microdoppler signals a microdoppler angle α _μD calculate which of the microdoppler signals is only assigned. The microdoppler angle α _μD indicates the angle which extends between a first direction and a second direction, wherein the first direction corresponds to the direction of travel of the vehicle and the second direction corresponds to the direction from which the microdoppler signal was received. The second direction may be based on the characteristics of the transmitted and received signals and / or on the knowledge of the transceiver device 111 , which has received the reflected signal, are determined. The first direction results from the fixed predetermined arrangement of the transceiver devices 111 on the vehicle.

The evaluation device 12 is further adapted to determine a power of the microdoppler signal. Using the power of the microdoppler signal and the microdoppler angle α _μD recognizes the evaluation device 12 whether the microdoppler signal corresponds to a pedestrian and whether the pedestrian crosses in front of the vehicle.

The driver assistance system 1 Optionally, a control device can continue 13 have, which is designed to, depending on the evaluation of the evaluation 12 to control a driving function of the vehicle. In particular, the control device 13 accelerate, decelerate or steer the vehicle. Optionally, a warning message can also be output, for example via an optical or acoustic signal, which warns the driver of the vehicle of a crossing pedestrian and a possible risk of collision.

In 2 is a schematic block diagram of a vehicle 2 illustrated according to an embodiment of the invention. The vehicle 2 includes for this purpose a driver assistance system described above 1 , Optionally, the vehicle 2 additional sensors or sensor systems 21 such as vehicle cameras, radar systems, infrared cameras or lidar systems. Sensor data of the sensor systems 21 In addition to the plausibility check, it is possible to use whether the microdoppler signal can be assigned to a pedestrian crossing in front of the vehicle.

In the following figures, exemplary embodiments of the evaluation device 12 explained in more detail.

In 3 illustrates an exemplary traffic situation, being a first pedestrian 3a and a second pedestrian 3b on one of the radar device 11 groundbreaking axis A lie. The angle α between the direction of travel of the vehicle 2 and the direction under which the first pedestrian 3a from the radar device is thus identical to the angle α between the direction of travel of the vehicle 2 and the direction under which the second pedestrian 3b from the radar device 11 out appears. Accordingly, it is difficult or impossible, the detected microdoppler signals one of the pedestrians 3a . 3b assigned. In addition, often the microdoppler angle α _μD not exactly determinable and can have a mistake of several degrees. In order to ensure a high robustness of the measurement, a plausibility check of the potentially relevant objects is therefore necessary.

The driver assistance system 1 can increase the robustness of sensor data from other sensor systems 21 of the vehicle 2 receive. In particular, an object can be detected on the basis of the sensor data. The subject can be classified and the object can have an object angle α _obj be assigned, which is relative to the direction of travel v2 of the vehicle 2 is measured.

In 4 is an exemplary measurement of the object angle α _obj illustrated. Except for small measurement inaccuracies corresponds to the object angle α _obj essentially the microdoppler angle α _μD , The object can be identified from the sensor data as a pedestrian. The vehicle 2 and the object move toward each other, and as the radius decreases, the amount of object angle α _obj or microdoppler angle α _μD gets smaller. The pedestrian 3 thus moves in the direction of travel of the vehicle 2 too, so that it is a crossing or crossing pedestrian.

In 5 In contrast, a scenario is illustrated where an object 5 , which is not a pedestrian, such as a Gullideckel, located on the street or on the roadside. At an initial time A and an end time C, the object appears 5 and the pedestrian 3 from different angles. At a middle time B is the object 5 between the radar device 11 and the pedestrian 3 so that a unique assignment of microdoppler signals to the object 5 or to the pedestrian 3 is no longer possible.

The driver assistance system 1 is now designed to detect whether the microdoppler signal is actually from a crossing pedestrian 3 or from another object 5 which is not a pedestrian crossing.

The evaluation device 12 may be configured to calculate a first plausibility function. At a certain time t calculates the evaluation 12 this is the amount between the microdoppler angle determined on the basis of the microdoppler signal α _μD and the object angle determined from the sensor data α _obj , The determined amount goes into an auxiliary function f, which, for example, can have the following form: $$f\left(x.y\right)=a*e^{\left(b*\left|x-y\right|\right)}+c$$

For example, factors a, b, and c may take the following values: $$\begin{array}{ccc}\text{a}=21.21& \text{b}=-.1057& \text{c}=-15.1\end{array}$$

In 6 the corresponding course of the auxiliary function f is shown. As can be seen, the function assumes positive values if the difference of microdoppler angles α _μD and object angle α _obj is less than 4 °. Otherwise, the auxiliary function f assumes negative values. The value of 4 ° thus represents a predetermined threshold.

The evaluation device 12 Now calculate the first plausibility function by multiplying the auxiliary function f by a prefactor conf _μD which determines the reliability of the measurement of the microdoppler angle α _μD and / or the object angle α _obj indicates. To calculate the reliability conf _μD In particular, the signal strength, ie the power of the microdoppler signal or the sensor signal can be used. For example, the reliability conf _μD be proportional to the signal strength. The first plausibility function thus has the following form: $${\text{plaus}}_{\alpha \text{match}}\left(t\right)={\text{conf}}_{\mu D}\left(t\right)*f\left({\alpha }_{\mu D}\left(t\right).{\alpha }_{\text{obj}}\left(t\right)\right)$$

Next, the evaluation 12 a cumulative first plausibility function p _αMatch by summing up the values of the plausibility function at all preceding times, that is, according to the following formula: $${\text{p}}_{\alpha \text{match}}\left(t\right)={\displaystyle \underset{i=1}{\overset{t}{\Sigma }}{\text{plaus}}_{\alpha \text{match}}\left(i\right)}$$

The invention is not limited thereto. Thus, only the last N values of the plausibility function can be added up, where N is a natural number. Also other values a, b, c of the auxiliary function f are possible. The auxiliary function f may further take other linear or non-linear forms. Preferably, the auxiliary function f is continuous. The sign is also convention.

In the 7 to 14 the (cumulative) plausibility functions are illustrated for different scenarios.

7 and 8th correspond to a first scenario with a crossing object. In 7 are the object angle α _obj and the microdoppler angle α _μD recorded as a function of distance, with each value of the microdoppler angle α _μD the corresponding reliability conf _μD assigned. The angles are substantially equal or at least mostly smaller than the predetermined threshold of 4 °. Correspondingly, the first plausibility function usually assumes positive values and the cumulative first plausibility function also assumes positive values at smaller intervals, ie later times.

The 9 and 10 correspond to a second exemplary scenario, again with the microdoppler angle α _μD essentially the object angle α _obj equivalent.

In the 11 and 12 a third scenario is illustrated where the microdoppler angle α _μD stronger from the object angle α _obj differs. Mostly the difference is greater than the preset threshold of 4 °. Correspondingly, the first plausibility function generally assumes negative values and the cumulative first plausibility function assumes negative values, in particular at small intervals, ie later times.

In 13 and 14 is a fourth scenario illustrated, again, the difference between the object angle α _obj and the microdoppler angle α _μD compared to the threshold is illustrated. The comparison of the first and second scenarios with the third and fourth scenarios shows that the sign indicates whether or not the microdoppler signal can be assigned to the object. The amount of the accumulated first plausibility function indicates the reliability or probability of this assignment.

The evaluation device 12 is further adapted to calculate a second plausibility function indicating whether it is a pedestrian crossing pedestrian or a non-pedestrian pedestrian.

In 15 an exemplary scenario is illustrated, wherein a first pedestrian 3a and a second pedestrian 3b each crossing, ie the current velocity vector v2 or the current direction of movement of the vehicle 2 with the current speed v3a or the current direction of movement of the first pedestrian 3a and with the current speed v3b or direction of movement of the second pedestrian 3b cuts.

In the in 16 As shown, the speeds intersect v3 . v5 or directions of movement of a pedestrian 3 or an object 5 not with the speed v2 or direction of movement of the vehicle 2 , These are therefore not intersecting objects, but rather objects moving in parallel. A collision is not expected in these cases.

In the in 17 The scenario shown moves a pedestrian 3 also along a direction of movement v3 on the direction of movement v2 of the vehicle 2 to, taking the view of the vehicle 2 on the pedestrian 3 initially in sections through objects 5a and 5b is covered.

The evaluation device determines to calculate the second plausibility function 12 whether the microdoppler angle α _μD decreases in time, ie goes to 0 or changes its sign. This is for example in the scenarios of 15 and 17 the case, but not in the scenario of 16 in which the microdoppler angle α _μD at an approach of vehicle 2 and pedestrians 3 or object 5 increases. The microdoppler angle α _μD follows at a crossing pedestrian a first curve f1, which in 18 is illustrated, while in a non-crossing object of the microdoppler angle α _μD one in 18 illustrated second curve f2 follows.

The evaluation device 12 Thus, the second plausibility function calculates by calculating the change in the microdoppler angle α _μD at the given time t relative to a value before a certain number w of time steps, for example w = 5. The second plausibility function thus calculates a slope or regression line which can be given by the following formula: $${\text{plaus}}_{{\alpha }_{\mu D}}\left(t\right)=\text{slope}\left(t-w\right)$$

Anstelle einer Regressionsgerade kann auch ein Mittelwert zur Berechnung des Winkeltrends ermittelt werden.The evaluation device 12 is configured to calculate a cumulative second plausibility function by summing the first plausibility function over the past time steps, ie according to the following formula: $${\text{p}}_{{\alpha }_{\mu D}}\left(t\right)={\displaystyle \underset{i=1}{\overset{t}{\Sigma }}{\text{plaus}}_{{\alpha }_{\mu D}}\left(i\right)}$$

Instead of a regression line, a mean value for calculating the angle trend can also be determined.

The 19 to 26 show exemplary courses of the microdoppler angle α _μD and the second plausibility functions for different scenarios.

The 19 and 21 show the microdoppler angle α _μD as a function of the distance for a first or second scenario and the 20 and 22 show the second plausibility function and cumulative second plausibility function as a function of the distance for the first and second scenario, respectively. The microdoppler angle α _μD in each case approaches 0, ie the second plausibility function assumes positive values and the cumulative second plausibility function also assumes positive and increasing values for decreasing distances, ie increasing time. It is thus a crossing object.

Of the 23 and 25 show the microdoppler angle α _μD as a function of the distance for a third or fourth scenario and the 24 and 26 show the second plausibility function and cumulative second plausibility function as a function of distance for the third and fourth scenario, respectively. The microdoppler angle α _μD increases with increasing time, ie with a reduced distance, and in particular does not approach 0. The second plausibility function therefore assumes negative values and the cumulative second plausibility function also assumes negative and increasing values for decreasing distances, ie increasing time. It is therefore not a crossing object.

The evaluation device 12 Finally, it is designed to calculate a third cumulative plausibility function, which indicates whether the microdoppler signal is to be assigned to a pedestrian or to another object. The evaluation device calculates this 12 the value of the third accumulated plausibility function at a current time t _m by subtracting the maximum value of all determined power values of the microdoppler signal and the minimum value of all determined power values, ie according to the following formula: $${\text{p}}_{\left|\mu D\right|}\left(t_m\right)=\underset{i=0...m}{\text{Max}}\left({\left|\mu D\right|}_{t_i}\right)-\underset{i=0...m}{\text{min}}\left({\left|\mu D\right|}_{t_i}\right)$$

Exemplary curves of the determined and normalized powers Pn of the microdoppler signal are in 27 displayed. A first course C1 always assumes values close to 1. Accordingly, the difference of maximum value and minimum value is small. In particular, for distances greater than 5 meters, the value of the third cumulative plausibility function is less than 0.3. A small variation in performance usually results from objects that are not pedestrians, such as leaves blowing in the wind or turning wheels of cars or cyclists.

In a second course C2 and a third course C3 On the other hand, larger power fluctuations occur so that the value of the third accumulated plausibility function increases, especially at smaller intervals. The larger power fluctuations result from the periodic oscillations of the arms and legs and are typical of a pedestrian.

The behavior in the near range, ie for distances smaller than 5 meters is usually less meaningful and can therefore be neglected to calculate the third plausibility function. According to further embodiments, the evaluation device 12 To calculate the third cumulative plausibility function, only take into account values that are determined for distances smaller than a predetermined distance threshold, approximately 15 meters.

The evaluation device 12 can now be designed to perform a plausibility check based on the first to third cumulative plausibility functions, whether the microdoppler signal is attributable to a crossing in front of the vehicle pedestrian.

So can the evaluation 12 recognize that the microdoppler signal is to be assigned to the detected object, if the value of the first cumulative plausibility function is greater than a predetermined positive first threshold, for example, greater than 30. The evaluation 12 can recognize that the micro signal is not to be assigned to the detected object, if the value of the first cumulative plausibility function is less than a predetermined negative second threshold, for example less than -100.

Next, the evaluation 12 recognize that the microdoppler signal corresponds to a crossing movement, if the value of the second accumulated plausibility function is greater than a predetermined positive third threshold, for example, greater than 3. The evaluation 12 can recognize that the microdoppler signal does not correspond to any crossing motion if the value of the second cumulative plausibility function is less than a predetermined negative fourth threshold, for example less than -5.

Next, the evaluation 12 recognize that the microdoppler signal is to be assigned to a pedestrian if the value of the third accumulated plausibility function is greater than a predetermined positive fifth threshold, for example, greater than 0.4. The evaluation device 12 can recognize that the microdoppler signal is not attributable to a pedestrian if the value of the third accumulated plausibility function is less than a predetermined positive sixth threshold, for example less than 0.3.

wobei: $$\begin{array}{ccc}{f}_1\left(x\right)=\frac{1}{1+e^{-x\text{/}15}},& f_2\left(x\right)=\frac{1}{1+e^{-x\text{/}6}},& f_3\left(x\right)=\frac{1}{1+e^{-\left(x-0.3\right)\text{/}10}}\end{array}$$

According to further embodiments, the evaluation device 12 the first to third cumulative plausibility functions combine to provide a plausibility indicator p _FG to calculate. The plausibility indicator p _FG preferably takes values between 0 and 1. For example, the plausibility indicator p _FG calculated according to the following formula: $${\text{p}}_{\text{FG}}\left(t\right)=\frac{1}{3}\left(f_1\left({\text{p}}_{\alpha \text{match}}\left(t\right)\right)+f_2\left({\text{p}}_{{\alpha }_{\mu D}}\left(t\right)\right)+f_3\left({\text{p}}_{\left|\mu D\right|}\left(t\right)\right)\right)$$

in which: $$\begin{array}{ccc}{f}_1\left(x\right)=\frac{1}{1+e^{-x\text{/}15}}.& f_2\left(x\right)=\frac{1}{1+e^{-x\text{/}6}}.& f_3\left(x\right)=\frac{1}{1+e^{-\left(x-0.3\right)\text{/}10}}\end{array}$$

If the recognizer 12 a plausibility indicator p _FG calculated, which is greater than a predetermined threshold, one in front of the vehicle 2 crossing pedestrians 3 recognized. The value of the plausibility indicator may more generally indicate a value for the likelihood of a pedestrian crossing in front of the vehicle.

Instead of the additive combination of the first to third cumulative plausibility functions, a logical AND connection is also possible.

Furthermore, instead of the cumulative totals for the cumulative plausibility functions, a product can also be calculated.

In 28 FIG. 10 is a flowchart of a method of detecting in front of a vehicle. FIG 2 crossing pedestrians 3 illustrated. In a first process step S1 a radar signal is transmitted, received and evaluated and a microdoppler signal determined.

In one process step S2 a plausibility check is carried out as to whether the microdoppler signal can be assigned to a pedestrian crossing in front of the vehicle. For this purpose, a power of the microdoppler signal and / or a microdoppler angle α _μD evaluated the microdoppler signal.

For this purpose, process steps S21 . S22 and S23 be executed. These process steps can be performed serially or in parallel. According to further embodiments, two of the method steps can also be carried out in parallel and the third method step can be carried out in series. In accordance with certain embodiments, only one or two of the method steps may be performed.

In the process step S21 is determined by an amount of distance between the microdoppler angle α _μD and an object angle detected from sensor data α _obj an object detects whether the microdoppler signal can be assigned to the object. In particular, a first cumulative plausibility function described above can be calculated for this purpose.

In the process step S22 it is calculated whether the microdoppler signal corresponds to a crossing movement, wherein a second cumulative plausibility function described above can be calculated.

Finally, in the process step S23 determines if the microdoppler signal corresponds to a pedestrian or other object. For this, the third cumulative plausibility function can be calculated.

In one process step S24 can the results of the process steps S21 to S23 in particular, by calculating the plausibility indicator described above p _FG , Based on the plausibility indicator p _FG can be detected, whether it is a pedestrian crossing in front of the vehicle.

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

• WO 2012/152476 A1 [0004]

Claims (10)

A method of detecting pedestrians (3) crossing in front of a vehicle (2), comprising the steps of: determining (S1) a microdoppler signal; and carrying out (S2) a plausibility check as to whether the microdoppler signal can be assigned to a pedestrian (3) crossing in front of the vehicle (2) by evaluating a power of the microdoppler signal and / or a microdoppler angle (α _μD ) of the microdoppler signal. Method according to Claim 1 wherein the microdoppler angle (α _μD ) is compared with an object angle (α _obj ) of an object detected from sensor data to detect whether the microdoppler signal is assignable to the detected object. Method according to Claim 2 in which, in order to detect whether the microdoppler signal can be assigned to the object, an absolute distance between the object angle (α _obj ) and the microdoppler angle (α _μD ) is compared with a predetermined threshold value. Method according to one of Claims 2 or 3 in which a temporal development of the comparison of the microdoppler angle (α _μD ) with the object angle (α _obj ) is evaluated in order to detect whether the microdoppler signal can be assigned to the detected object. Method according to one of the preceding claims, wherein a temporal change of the microdoppler angle (α _μD ) is evaluated in order to detect a crossing movement. Method according to Claim 5 , wherein it is recognized that the microdoppler signal is attributable to a crossing movement if the microdoppler angle (α _μD ), measured with respect to the direction of movement of the vehicle (2), decreases. A method according to any one of the preceding claims, wherein fluctuations in the power of the microdoppler signal are evaluated to detect whether the microdoppler signal is attributable to arm or leg movements typical of a pedestrian (3). Method according to Claim 7 recognizing that the microdoppler signal is attributable to arm or leg movements typical of a pedestrian (3) if the variations in the power of the microdoppler signal exceed a predetermined threshold. Driver assistance system (1) for a vehicle (2) for detecting crossing pedestrians (3), having a radar device (11) which is designed to detect microdoppler signals; an evaluation device (12) which is designed to perform a plausibility check by evaluating a power of a microdoppler signal determined by the radar device (11) and / or a microdoppler angle (α _μD ) of the microdoppler signal, whether the microdoppler signal a pedestrian (3) crossing in front of the vehicle (2) can be assigned. Vehicle (2) with a driver assistance system (1) according to Claim 9 ,

Images