Granted Patent B2

US 12,326,499 · App. 17/802,057 · Granted Jun 10, 2025

Inventor: Dominik Kellner (Dachau, DE)

Assignee: Bayerische Motoren Werke Aktiengesellschaft

G01S13/867G01S13/931G06V10/764G06V20/58

Abstract

Claims (142)

1. A method for detecting objects from a motor vehicle that move in surroundings of the motor vehicle, the method comprising:

recording camera data of the surroundings;

recording radar echoes from the surroundings;

recognizing at least one object depicted in the camera data on a basis of the camera data;

generating an azimuth angle range occupied by the at least one object from a viewpoint of the motor vehicle and a delimiting frame enclosing the at least one object having an assumed distance to the motor vehicle for the at least one object;

determining an associated Doppler velocity for the radar echoes with respect to a current position of the motor vehicle;

determining an associated distance from the motor vehicle and an associated azimuth angle at least for moving radar echoes for which a non-negligible Doppler velocity is determined;

carrying out an assignment between at least one of the moving radar echoes and the delimiting frame by comparing the respective azimuth angle to the azimuth angle range and by comparing the respective radar-based determined distance to the camera-based assumed distance;

for the delimiting frame to which the at least one of the moving radar echoes was successfully assigned, correcting the respective assumed distance according to the radar-based determined distance of the at least one of the moving radar echoes; and

outputting the delimiting frame with the respective corrected distance as an object data set indicating successful object detection,

wherein carrying out the assignment between the at least one of the moving radar echoes and the delimiting frame comprises carrying out the assignment between at least three moving radar echoes and the delimiting frame;

wherein the method further comprises:

determining a Doppler velocity profile on a basis of the at least three moving radar echoes such that as many as possible of the radar echoes are consistent with the Doppler velocity profile; and

correcting the distance of the respective delimiting frame based on the determined distances of only the radar echoes consistent with the Doppler velocity profile.

2. The method according to claim 1 , further comprising:

carrying out the method for radar echoes of one measurement cycle in each case;

determining that no camera data recorded simultaneously is available for a recording time of the measurement cycle; and

interpolating or extrapolating available camera data onto the recording time of the measurement cycle in response to determining that no camera data recorded simultaneously is available for the recording time of the measurement cycle.

3. The method according to claim 1 , wherein

the radar echoes and the delimiting frame are embedded in a predefined coordinate system,

the method further comprising:

determining a scaling error in the predefined coordinate system at least for the moving radar echoes, which scaling error indicates a distance of the respective radar echo to the delimiting frame in a radial direction with respect to the current position of the motor vehicle; and

determining that the scaling error for the respective radar echo is less than a predefined threshold value for a successful assignment.

4. The method according to claim 1 , further comprising:

assessing the assignment based on a predefined probability function for the azimuth angle and/or for a scaling error which indicates for a radar echo in each case its distance to the respective delimiting frame in a predefined coordinate system,

wherein the predefined probability function has a constant central range from which the probability drops toward both sides.

5. The method according to claim 1 , further comprising:

estimating an orientation of the at least one object on a basis of the radar echoes;

comparing the estimated orientation to an orientation of the respective delimiting frame; and

assessing the assignment on a basis of a deviation between the orientation of the at least one object and the orientation of the respective delimiting frame, wherein the assignment is retained in response to the deviation being less than a predefined threshold value.

6. The method according to claim 1 , further comprising:

determining an orientation of the respective delimiting frame;

estimating a radar-based orientation and movement velocity of the at least one object in a direction of the radar-based determined orientation on a basis of the radar echoes consistent with the Doppler velocity profile; and

checking whether the corresponding respective data are consistent with a yawing object using a predefined velocity-dependent yaw rate and/or a predefined velocity as a plausible reference, in the event of a deviation of the orientations from one another lying above a predefined threshold value.

7. The method according to claim 1 , further comprising:

determining a scaling error for the radar echoes, which indicates a distance of the respective radar echo to the respective delimiting frame in a radial direction with respect to the current position of the motor vehicle;

upon successful assignment of a plurality of consistently moving radar echoes to a delimiting frame, analysing stationary radar echoes detected in a predefined spatial vicinity to these radar echoes, which have a negligible Doppler velocity; and

assigning analysed stationary radar echoes to the respective delimiting frame that are also consistent with the Doppler velocity profile and the scaling error of which deviates by at most a predefined absolute value from the scaling errors of the moving radar echoes already assigned to the respective delimiting frame.

8. The method according to claim 1 , further comprising:

determining that fewer than three moving radar echoes have been assigned to the delimiting frame, or that such an assignment has resulted in an error;

determining an orientation of the delimiting frame;

estimating a plausible Doppler velocity of the at least one object in relation to the current position of the motor vehicle based on the orientation of the delimiting frame;

for each moving radar echo of the moving radar echoes having been assigned to the delimiting frame, determining a probability that one moving radar echo of the fewer than three moving radar echoes having been assigned to the delimiting frame originates from the at least one object enclosed by the respective delimiting frame by comparing the estimated Doppler velocity to the radar-based determined Doppler velocity for the one moving radar echo; and

using, as the basis for correcting the assumed distance of the delimiting frame, only or at least the radar echo for which the highest probability was determined.

9. The method according to claim 1 , further comprising:

automatically classifying objects recognized on a basis of the camera data with respect to their type based on predefined classes, wherein plausible radar-based determined Doppler velocity directions are assigned to each class, and the assignment of a radar echo to a delimiting frame is precluded in response to a Doppler velocity direction determined for the respective radar echo being different from the plausible Doppler velocity directions for the type of the object enclosed by the respective delimiting frame.

10. The method according to claim 1 , further comprising:

automatically classifying objects recognized on a basis of the camera data with respect to their type based on predefined classes, wherein a plausible Doppler velocity range is assigned to each class, and the assignment of a radar echo to a delimiting frame is precluded in response to the Doppler velocity determined for the respective radar echo being outside the plausible Doppler velocity range for the type of the object enclosed by the respective delimiting frame.

11. The method according to claim 1 , further comprising:

determining a respective radar cross section at least for the moving radar echoes;

automatically classifying objects recognized on the basis of the camera data with respect to their type based on predefined classes, wherein a plausible radar cross section range is assigned to each class; and

precluding the assignment of a radar echo to a delimiting frame in response to the radar cross section of the respective radar echo being outside the plausible radar cross section range for the type of the object enclosed by the respective delimiting frame.

12. A motor vehicle, comprising:

a surroundings sensor system having at least one camera for recording camera data of the surroundings of the motor vehicle and at least one radar unit for emitting radar pulses into the surroundings and recording resulting radar echoes from the surroundings; and

a data processing unit connected to the surroundings sensor system, wherein the data processing unit is configured to:

recognize at least one object depicted in the camera data on a basis of the camera data;

generate an azimuth angle range occupied by the at least one object from a viewpoint of the motor vehicle and a delimiting frame enclosing the at least one object having an assumed distance to the motor vehicle for the at least one object;

determine an associated Doppler velocity for the radar echoes with respect to a current position of the motor vehicle;

determine an associated distance from the motor vehicle and an associated azimuth angle at least for moving radar echoes for which a non-negligible Doppler velocity is determined;

carry out an assignment between at least one of the moving radar echoes and the delimiting frame by comparing the respective azimuth angle to the azimuth angle range and by comparing the respective radar-based determined distance to the camera-based assumed distance;

for the delimiting frame to which the at least one of the moving radar echoes was successfully assigned, correct the respective assumed distance according to the radar-based determined distance of the at least one of the moving radar echoes; and

output the delimiting frame with the respective corrected distance as an object data set indicating successful object detection;

wherein the data processing unit is further configured to:

carry out the assignment between the at least one of the moving radar echoes and the delimiting frame by carrying out the assignment between at least three moving radar echoes and the delimiting frame;

determine a Doppler velocity profile on a basis of the at least three moving radar echoes such that as many as possible of the radar echoes are consistent with the Doppler velocity profile; and

correct the distance of the respective delimiting frame based on the determined distances of only the radar echoes consistent with the Doppler velocity profile.

13. A method for detecting objects from a motor vehicle that move in surroundings of the motor vehicle, the method comprising:

recording camera data of the surroundings;

recording radar echoes from the surroundings;

recognizing at least one object depicted in the camera data on a basis of the camera data;

generating an azimuth angle range occupied by the at least one object from a viewpoint of the motor vehicle and a delimiting frame enclosing the at least one object having an assumed distance to the motor vehicle for the at least one object;

determining an associated Doppler velocity for the radar echoes with respect to a current position of the motor vehicle;

determining an associated distance from the motor vehicle and an associated azimuth angle at least for moving radar echoes for which a non-negligible Doppler velocity is determined;

carrying out an assignment between at least one of the moving radar echoes and the delimiting frame by comparing the respective azimuth angle to the azimuth angle range and by comparing the respective radar-based determined distance to the camera-based assumed distance;

for the delimiting frame to which the at least one of the moving radar echoes was successfully assigned, correcting the respective assumed distance according to the radar-based determined distance of the at least one of the moving radar echoes;

outputting the delimiting frame with the respective corrected distance as an object data set indicating successful object detection,

wherein the radar echoes and the delimiting frame are embedded in a predefined coordinate system,

the method further comprising:

determining a scaling error in the predefined coordinate system at least for the moving radar echoes, which scaling error indicates a distance of the respective radar echo to the delimiting frame in a radial direction with respect to the current position of the motor vehicle; and

determining that the scaling error for the respective radar echo is less than a predefined threshold value for a successful assignment.

14. A method for detecting objects from a motor vehicle that move in surroundings of the motor vehicle, the method comprising:

recording camera data of the surroundings;

recording radar echoes from the surroundings;

recognizing at least one object depicted in the camera data on a basis of the camera data;

generating an azimuth angle range occupied by the at least one object from a viewpoint of the motor vehicle and a delimiting frame enclosing the at least one object having an assumed distance to the motor vehicle for the at least one object;

determining an associated Doppler velocity for the radar echoes with respect to a current position of the motor vehicle;

determining an associated distance from the motor vehicle and an associated azimuth angle at least for moving radar echoes for which a non-negligible Doppler velocity is determined;

carrying out an assignment between at least one of the moving radar echoes and the delimiting frame by comparing the respective azimuth angle to the azimuth angle range and by comparing the respective radar-based determined distance to the camera-based assumed distance;

for the delimiting frame to which the at least one of the moving radar echoes was successfully assigned, correcting the respective assumed distance according to the radar-based determined distance of the at least one of the moving radar echoes;

outputting the delimiting frame with the respective corrected distance as an object data set indicating successful object detection; and

assessing the assignment based on a predefined probability function for the azimuth angle and/or for a scaling error which indicates for a radar echo in each case its distance to the respective delimiting frame in a predefined coordinate system,

wherein the predefined probability function has a constant central range from which the probability drops toward both sides.

15. A method for detecting objects from a motor vehicle that move in surroundings of the motor vehicle, the method comprising:

recording camera data of the surroundings;

recording radar echoes from the surroundings;

recognizing at least one object depicted in the camera data on a basis of the camera data;

generating an azimuth angle range occupied by the at least one object from a viewpoint of the motor vehicle and a delimiting frame enclosing the at least one object having an assumed distance to the motor vehicle for the at least one object;

determining an associated Doppler velocity for the radar echoes with respect to a current position of the motor vehicle;

determining an associated distance from the motor vehicle and an associated azimuth angle at least for moving radar echoes for which a non-negligible Doppler velocity is determined;

carrying out an assignment between at least one of the moving radar echoes and the delimiting frame by comparing the respective azimuth angle to the azimuth angle range and by comparing the respective radar-based determined distance to the camera-based assumed distance;

for the delimiting frame to which the at least one of the moving radar echoes was successfully assigned, correcting the respective assumed distance according to the radar-based determined distance of the at least one of the moving radar echoes;

outputting the delimiting frame with the respective corrected distance as an object data set indicating successful object detection;

determining that fewer than three moving radar echoes have been assigned to the delimiting frame, or that such an assignment has resulted in an error;

determining an orientation of the delimiting frame;

estimating a plausible Doppler velocity of the at least one object in relation to the current position of the motor vehicle based on the orientation of the delimiting frame;

for each moving radar echo of the moving radar echoes having been assigned to the delimiting frame, determining a probability that one moving radar echo of the fewer than three moving radar echoes having been assigned to the delimiting frame originates from the at least one object enclosed by the respective delimiting frame by comparing the estimated Doppler velocity to the radar-based determined Doppler velocity for the one moving radar echo; and

using, as the basis for correcting the assumed distance of the delimiting frame, only or at least the radar echo for which the highest probability was determined.

16. A method for detecting objects from a motor vehicle that move in surroundings of the motor vehicle, the method comprising:

recording camera data of the surroundings;

recording radar echoes from the surroundings;

recognizing at least one object depicted in the camera data on a basis of the camera data;

generating an azimuth angle range occupied by the at least one object from a viewpoint of the motor vehicle and a delimiting frame enclosing the at least one object having an assumed distance to the motor vehicle for the at least one object;

determining an associated Doppler velocity for the radar echoes with respect to a current position of the motor vehicle;

determining an associated distance from the motor vehicle and an associated azimuth angle at least for moving radar echoes for which a non-negligible Doppler velocity is determined;

carrying out an assignment between at least one of the moving radar echoes and the delimiting frame by comparing the respective azimuth angle to the azimuth angle range and by comparing the respective radar-based determined distance to the camera-based assumed distance;

for the delimiting frame to which the at least one of the moving radar echoes was successfully assigned, correcting the respective assumed distance according to the radar-based determined distance of the at least one of the moving radar echoes;

outputting the delimiting frame with the respective corrected distance as an object data set indicating successful object detection; and

automatically classifying objects recognized on a basis of the camera data with respect to their type based on predefined classes, wherein plausible radar-based determined Doppler velocity directions are assigned to each class, and the assignment of a radar echo to a delimiting frame is precluded in response to a Doppler velocity direction determined for the respective radar echo being different from the plausible Doppler velocity directions for the type of the object enclosed by the respective delimiting frame.

17. A method for detecting objects from a motor vehicle that move in surroundings of the motor vehicle, the method comprising:

recording camera data of the surroundings;

recording radar echoes from the surroundings;

recognizing at least one object depicted in the camera data on a basis of the camera data;

generating an azimuth angle range occupied by the at least one object from a viewpoint of the motor vehicle and a delimiting frame enclosing the at least one object having an assumed distance to the motor vehicle for the at least one object;

determining an associated Doppler velocity for the radar echoes with respect to a current position of the motor vehicle;

determining an associated distance from the motor vehicle and an associated azimuth angle at least for moving radar echoes for which a non-negligible Doppler velocity is determined;

carrying out an assignment between at least one of the moving radar echoes and the delimiting frame by comparing the respective azimuth angle to the azimuth angle range and by comparing the respective radar-based determined distance to the camera-based assumed distance;

for the delimiting frame to which the at least one of the moving radar echoes was successfully assigned, correcting the respective assumed distance according to the radar-based determined distance of the at least one of the moving radar echoes;

outputting the delimiting frame with the respective corrected distance as an object data set indicating successful object detection; and

automatically classifying objects recognized on a basis of the camera data with respect to their type based on predefined classes, wherein a plausible Doppler velocity range is assigned to each class, and the assignment of a radar echo to a delimiting frame is precluded in response to the Doppler velocity determined for the respective radar echo being outside the plausible Doppler velocity range for the type of the object enclosed by the respective delimiting frame.

18. A method for detecting objects from a motor vehicle that move in surroundings of the motor vehicle, the method comprising:

recording camera data of the surroundings;

recording radar echoes from the surroundings;

recognizing at least one object depicted in the camera data on a basis of the camera data;

generating an azimuth angle range occupied by the at least one object from a viewpoint of the motor vehicle and a delimiting frame enclosing the at least one object having an assumed distance to the motor vehicle for the at least one object;

determining an associated Doppler velocity for the radar echoes with respect to a current position of the motor vehicle;

determining an associated distance from the motor vehicle and an associated azimuth angle at least for moving radar echoes for which a non-negligible Doppler velocity is determined;

carrying out an assignment between at least one of the moving radar echoes and the delimiting frame by comparing the respective azimuth angle to the azimuth angle range and by comparing the respective radar-based determined distance to the camera-based assumed distance;

for the delimiting frame to which the at least one of the moving radar echoes was successfully assigned, correcting the respective assumed distance according to the radar-based determined distance of the at least one of the moving radar echoes;

outputting the delimiting frame with the respective corrected distance as an object data set indicating successful object detection;

determining a respective radar cross section at least for the moving radar echoes;

automatically classifying objects recognized on the basis of the camera data with respect to their type based on predefined classes, wherein a plausible radar cross section range is assigned to each class; and

precluding the assignment of a radar echo to a delimiting frame in response to the radar cross section of the respective radar echo being outside the plausible radar cross section range for the type of the object enclosed by the respective delimiting frame.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (24)