Method and apparatus for detecting decalibration of a sensor for detecting surroundings of a vehicle and vehicle
The method and apparatus rapidly detect sensor misalignment in vehicles by analyzing angular velocity and comparing vehicle motion data with sensor data, addressing the inefficiencies of existing methods and enhancing safety in autonomous vehicles.
Patent Information
- Application Number
- JP2024506891
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2021-08-06
- Filing Date
- 2022-06-28
- Publication Date
- 2024-12-09
- Estimated Expiration
- 2042-06-28
AI Technical Summary
Existing methods for detecting sudden changes in the alignment of vehicle sensors are slow and inefficient, leading to inaccurate recognition data and potential safety issues in autonomous vehicles.
A method and apparatus that utilize angular velocity analysis of detected objects to identify sudden changes in sensor alignment by comparing vehicle motion data with sensor data, allowing for rapid detection of decalibration through histogram evaluation.
Enables instantaneous detection of sensor misalignment, reducing detection time and improving safety by preventing erroneous vehicle interventions.
Smart Images

Figure 0007596598000001 
Figure 0007596598000002 
Figure 0007596598000003
Abstract
Description
[Technical field]
[0001] The invention relates to a method for detecting decalibration of a sensor for detecting the surroundings of a vehicle according to the preamble of claim 1. Furthermore the invention relates to a device for detecting decalibration of a sensor for detecting the surroundings of a vehicle according to claim 7 as well as to a vehicle according to claim 8. [Background technology]
[0002] For example, ambient sensors for automobiles may experience sudden changes in position and / or alignment that may lead to degradation of sensor performance. This may lead to, among other things, inaccurate perception data. From the perspective of highly automated and / or autonomous vehicle systems, sudden undetected changes may result in undesired vehicle interventions and / or prevent necessary vehicle interventions.
[0003] It is therefore desirable to detect sudden changes in the position and / or alignment of the perimeter sensor.
[0004] In the prior art, the alignment of the sensor is determined based on a time average of the recognition data. The mathematical techniques used for this are inherently slow compared to the update rate of the recognition data. Due to high filter constants, sudden changes in the alignment of the sensor, especially relative to the vehicle axis, cannot be determined immediately.
[0005] DE 102021000360 A1 describes a device for recognizing a decalibration of a camera. In that case, an image sensor is attached to a frame by at least one suspension. Furthermore, at least one electrical capacitor is arranged on the frame and is fully charged when the camera is not in operation. An evaluation unit is provided, which is configured to determine the charge of the capacitor when the camera is started up, to further check whether the determined charge corresponds to an expected value, and to determine a decalibration of the camera if an error between the determined charge and the expected value is identified and / or to output corresponding information if an error is identified. Summary of the Invention [Problem to be solved by the invention]
[0006] It is an object of the present invention to provide a method and device for detecting decalibration of a sensor for detecting the surroundings of a vehicle, and a vehicle. [Means for solving the problem]
[0007] According to the present invention, the above object is achieved by a method having the features of claim 1, an apparatus having the features of claim 7 and a vehicle having the features of claim 8.
[0008] Advantageous embodiments of the invention are the subject matter of the dependent claims.
[0009] The present invention relates to a method for detecting decalibration of a sensor for detecting the surroundings of a vehicle.
[0010] According to the present invention, a plurality of objects in the vicinity are detected by a perimeter sensor, the angular velocities of the detected objects are determined while the vehicle is moving, clusters of the number of objects are determined by a histogram evaluation based on the angular velocities, and if a cluster is identified whose angular velocities deviate from zero, a decalibration of the perimeter sensor caused by an error in alignment of the perimeter sensor from a normal position with respect to the vehicle axis of the vehicle is identified.
[0011] The present invention allows for instantaneous detection of sudden changes in the position and / or alignment of the peripheral sensor, with detection time limited only by the cycle time of the sensor.
[0012] This has the following advantages: -reducing the detection time of changes in the position and / or alignment of the peripheral sensors; - avoidance of safety-related erroneous interventions in vehicle control units, and - Improved reliability of detecting changes in the position and / or alignment of peripheral sensors.
[0013] In one embodiment, the angular velocity is determined to be one or more of the yaw angular velocity, pitch angular velocity, and roll angular velocity, and the histogram evaluation is performed.
[0014] In one embodiment, the angular velocity of a sensed object is determined by comparing vehicle motion data with data detected by or derived from surrounding sensors.
[0015] In one embodiment, lidar and / or radar and / or camera and / or ultrasonic sensors are used as ambient sensors.
[0016] In one embodiment, the vehicle motion data used is one or more of the following list: steering angle, yaw angle, roll angle, pitch angle, yaw acceleration, roll acceleration, pitch acceleration, suspension displacement signals, wheel speed pulses, position data, and redundant ambient sensor devices.
[0017] In one embodiment, lidar and / or radar and / or camera and / or ultrasonic sensors are used as redundant ambient sensor devices.
[0018] In one aspect of the invention, an apparatus for detecting decalibration of at least one surrounding sensor for detecting the surroundings of a vehicle is proposed, the apparatus comprising a data processing unit connected to the at least one surrounding sensor and configured to perform any one of the methods described above.
[0019] In another aspect of the invention, a vehicle is proposed that is equipped with such a device and at least one surroundings sensor for detecting the surroundings of the vehicle.
[0020] In one embodiment, the surrounding sensors include lidar and / or radar and / or cameras and / or ultrasonic sensors.
[0021] In one embodiment, to detect vehicle motion data, one or more sensors are provided to detect one or more data from the following list: steering angle, yaw angle, roll angle, pitch angle, yaw acceleration, roll acceleration, pitch acceleration, suspension displacement signals, wheel speed pulses, position data, and redundant ambient sensor devices.
[0022] To avoid degradation, the invention proposes a direct comparison of data from the various sensors, in particular the available vehicle motion data and the recognized sensor data. During normal operation, both data sources provide identical information on the vehicle and sensor alignment redundantly, i.e. without misalignment between the two alignments. Any change in the alignment of the sensors appears as a deviation between the two data sources with respect to the extracted alignment. Such a change can be detected as soon as the input data is available for processing. Its sensitivity is limited by a threshold value selected for the allowed deviation.
[0023] Possible sensor data are own perception data (e.g. point clouds from lidar, radar echoes, camera images, ultrasonic echoes) and data derived from this.
[0024] Possible vehicle data sources are: -Vehicle self-motion data: -Steering angle, - Yaw angle, roll angle, pitch angle, and corresponding acceleration, -Suspension displacement signal, -Wheel speed pulse, - Location (GPS, GNSS, map data, etc.), -Redundant ambient sensor equipment (radar, camera, lidar, ultrasonic).
[0025] In the following, an embodiment of the present invention will be explained in detail with reference to the drawings. [Brief description of the drawings]
[0026] [Figure 1] FIG. 1 is a schematic diagram of a vehicle during normal operation. [Diagram 2] 1 is a schematic diagram of a vehicle when the alignment of a surrounding sensor suddenly changes. [Diagram 3] 4 is a schematic histogram showing the distribution of the number of recognized objects for different angular velocities during normal operation of the vehicle; [Figure 4] 13 is a schematic histogram showing the distribution of the number of recognized objects for different angular velocities when the alignment of the peripheral sensor is suddenly changed;
[0027] In all the drawings, the same reference numerals are used to designate corresponding parts. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0028] 1 is a schematic diagram of a vehicle 1 in normal operation. At least one perimeter sensor 2 is arranged on the vehicle 1, oriented in or aligned with respect to a vehicle axis A. The positions of objects in the perimeter are sensed by the perimeter sensor 2 with precision.
[0029] FIG. 2 is a schematic diagram of a vehicle 1 when the alignment of the perimeter sensor 2 changes suddenly. By way of example, the alignment of the perimeter sensor 2 is misaligned from the vehicle axis A, in particular by the yaw angle α. The object 3 therefore appears misaligned in the perception of the perimeter sensor 2 at an apparent position 3′. In the reference frame moving with the vehicle 1, the variation in the position of the object 3 is either the result of the object 3's own movement or the result of a change in the alignment of the perimeter sensor 2. A sudden change in alignment makes the perimeter sensor 2 see the entire array of objects 3 moving in the same direction. This displacement is r*α, where r is the radial distance of each object 3 to the perimeter sensor 2. All objects move simultaneously with the same angular velocity dα / dt, in particular the azimuth velocity. Such a movement of the entire array can be detected by algorithmic filtering of the angular velocity dα / dt or the sensor position. This same analysis can also be applied to the pitch and roll angles, so that a sudden change in alignment can be detected for all possible scenarios.
[0030] FIG. 3 is a schematic histogram showing the distribution of the number N of recognized objects for different angular velocities dα / dt during normal operation of the vehicle 1, i.e. when the surroundings sensor 2 is correctly aligned in the direction of or with respect to the vehicle axis A.
[0031] The angular velocity dα / dt is defined in a reference frame moving with the vehicle 1. During normal operation, typical driving scenarios result in a wide distribution centered around zero. The variance of the distribution is determined by the motion of objects 3 around the vehicle 1.
[0032] FIG. 4 is a schematic histogram showing the distribution of the number N of recognized objects for different angular velocities dα / dt when the alignment suddenly changes, i.e. when the perimeter sensor 2 is no longer correctly aligned in the direction of or in relation to the vehicle axis A. From the perspective of the perimeter sensor 2, a misalignment of the perimeter sensor 2 appears as if the entire perimeter of the vehicle 1 has moved by an angle α. As a result, all recognition data show a strong component of angular velocities dα / dt that is superimposed on the actual movement of the perimeter of the vehicle 1. This produces non-zero peaks P or clusters P in the distribution. The magnitude of the angular displacement is larger the faster the misalignment occurs. Detecting such clusters P in the distribution of angular velocities dα / dt is an effective way to identify sudden changes in the alignment of the perimeter sensor 2.
[0033] The determination of the angular velocity dα / dt of the detected object 3 can be made by comparing the motion data of the vehicle 1 with data detected by or derived from the surroundings sensor 2. As surroundings sensor 2, for example, a lidar and / or radar and / or a camera and / or an ultrasonic sensor can be used. As motion data of the vehicle 1, one or more of the following list are used: steering angle, yaw angle, roll angle, pitch angle, yaw acceleration, roll acceleration, pitch acceleration, suspension displacement signals, wheel speed pulses, position data and redundant surroundings sensor devices, for example a lidar and / or radar and / or a camera and / or an ultrasonic sensor.
[0034] The method can be executed in a data processing unit 4 which can be located in the vehicle 1 for example. [Prior art documents] [Patent documents]
[0035] [Patent Document 1] DE 102021000360
Claims
1. A method for detecting decalibration of at least one surroundings sensor (2) for detecting the surroundings of a vehicle (1), comprising: a plurality of objects (3) in the surroundings are detected by the surroundings sensor (2); The angular velocities (dα / dt) of the detected objects (3) are calculated while the vehicle (1) is moving, and when it is confirmed that the angular velocity (dα / dt) of the peak (P) of the distribution deviates from zero based on a histogram showing a distribution of the number (N) of the detected objects (3) for each angular velocity (dα / dt), decalibration of the surrounding sensor (2) caused by an alignment error of the surrounding sensor (2) from a normal position with respect to the vehicle axis (A) of the vehicle (1) is identified.
13. A method for detecting decalibration of a sensor, comprising:
2. Calculating one or more of a yaw angular velocity, a pitch angular velocity, and a roll angular velocity as the angular velocity (dα / dt) of each of the detected objects (3).
2. A method for detecting decalibration of a sensor according to claim 1.
3. The angular velocity (dα / dt) of each of the detected objects (3) is calculated based on the motion data of the vehicle (1) and on data detected by or derived from the surroundings sensors (2).
2. A method for detecting decalibration of a sensor according to claim 1.
4. As the surrounding sensor (2), a lidar and / or a radar and / or a camera and / or an ultrasonic sensor are used.
2. A method for detecting decalibration of a sensor according to claim 1.
5. As the motion data of the vehicle (1), one or more of the following list is used: steering angle, yaw angle, roll angle, pitch angle, yaw acceleration, roll acceleration, pitch acceleration, suspension displacement signals, wheel speed pulses, position data, and data sensed by redundant surrounding sensor devices.
4. A method for detecting decalibration of a sensor according to claim 3.
6. When data detected by the redundant surrounding sensor device is used as the motion data of the vehicle (1), a lidar and / or a radar and / or a camera and / or an ultrasonic sensor is used as the redundant surrounding sensor device.
6. A method for detecting decalibration of a sensor according to claim 5.
7. 1. An apparatus for detecting decalibration of at least one surroundings sensor (2) for detecting the surroundings of a vehicle (1), comprising: A device for detecting sensor decalibration, comprising a data processing unit (4) connected to said at least one ambient sensor (2) and adapted to execute the method according to any one of claims 1 to 6.
8. A vehicle (1) comprising a device according to claim 7 and at least one surroundings sensor (2) for detecting the surroundings of the vehicle (1).
9. As the surrounding sensor (2), a lidar and / or a radar and / or a camera and / or an ultrasonic sensor are provided.
9. A vehicle (1) according to claim 8, characterized in that
10. In order to detect motion data of the vehicle (1), one or more sensors are provided for detecting one or more data from the following list: steering angle, yaw angle, roll angle, pitch angle, yaw acceleration, roll acceleration, pitch acceleration, suspension displacement signals, wheel speed pulses, position data, and data sensed by a redundant surroundings sensor device; The data processing unit (4) calculates the angular velocity (dα / dt) of each of the detected objects (3) based on the motion data of the vehicle (1) and on data detected by or derived from the surroundings sensors (2).
9. A vehicle (1) according to claim 8, characterized in that
Citation Information
Patent Citations
Device and method for detecting a camera decalibration
DE102021000360A1
Method and device for recognizing obstacle
JP2005329779A
Calibration device and calibration method
JP2020026955A
Method and device for correcting sensing errors among lidar sensors
KR1020210081151A
Apparatus and method for detecting alignment of sensor in an automotive detection system
US20180341007A1
