Method and apparatus for detecting non-calibration of a sensor recording the surroundings of a vehicle

Abstract

The present invention relates to a method for detecting non-correction of at least one surrounding sensor (2) that records the surroundings of a vehicle (1), wherein a plurality of objects (3) in the surroundings are detected by the surrounding sensor (2), the angular velocity (dα / dt) of the detected objects (3) is determined during the driving of the vehicle (1), the accumulation (P) of the number (N) of the objects (3) in relation to the angular velocity (dα / dt) is determined by histogram evaluation, and when the accumulation (P) is confirmed when the angular velocity (dα / dt) is not zero, the non-correction of the surrounding sensor (2) due to an orientation error of the surrounding sensor (2) deviating from the normal position with respect to the vehicle axis (A) of the vehicle (1) is inferred. Furthermore, the present invention relates to a device for detecting non-correction and a vehicle (1) having such a device.

Description

Technology Field

[0001] The present invention relates to a method for detecting non-correction of a sensor recording the surroundings of a vehicle according to the preamble of claim 1. In addition, the present invention relates to an apparatus for detecting non-correction of a sensor recording the surroundings of a vehicle according to the preamble of claim 7 and a vehicle according to claim 8. Background Technology

[0002] For example, surrounding sensors for a vehicle may experience sudden changes in position and / or orientation, which can lead to a degradation of sensor performance. In particular, this can lead to errors in perception data. In the context of highly automated and / or autonomous vehicle systems, undetected sudden changes can lead to unwanted vehicle intervention and / or prevent necessary vehicle intervention.

[0003] Therefore, it is desirable to detect sudden changes in the position and / or orientation of surrounding sensors.

[0004] In prior art, sensor orientation is verified based on temporal notification of perception data. The algorithmic method used for this purpose is inherently slow compared to the update speed of the perception data. Because the filter constant is large, sudden changes in sensor orientation, particularly with respect to the vehicle axis, cannot be detected immediately.

[0005] German Patent DE 10 2021 000 360 A1 describes a device for detecting camera non-calibration. In the device, an image sensor is fixed to a frame by at least one mounting. Additionally, at least one electric capacitor is disposed in the frame, which is fully charged when the camera is deactivated. An evaluation unit is provided that, when the camera is activated, determines the charge of the capacitor, then checks whether the determined charge meets an estimate, and if an error is found between the determined charge and the estimate, infers / deduces camera non-calibration, or outputs corresponding information if an error is found. The problem to be solved

[0006] The present invention aims to provide a method and apparatus for detecting non-calibration of a sensor recording the surroundings of a vehicle, and a vehicle.

[0007] According to the present invention, this objective 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] The preferred configurations of the present invention are the subject of the dependent claims. means of solving the problem

[0009] The present invention relates to a method for detecting non-calibration of a sensor that records the surroundings of a vehicle.

[0010] According to the present invention, a plurality of objects in the surrounding area are detected by surrounding sensors. The angular velocity of the detected objects is determined during the driving of the vehicle, and the accumulation of the number of objects in relation to the angular velocity is determined by histogram evaluation, and when the accumulation is confirmed when the angular velocity is not zero, the non-correction of the surrounding sensors due to orientation errors of the surrounding sensors deviating from the normal position with respect to the vehicle axis is inferred.

[0011] The present invention allows for the direct detection of sudden changes in the position and / or orientation of surrounding sensors. The detection time is limited only by the cycle time of the sensor.

[0012] From this, the following advantages arise.

[0013] - Reduction in detection time for changes in the position and / or orientation of surrounding sensors,

[0014] - Prevention of erroneous intervention in vehicle control in relation to safety, and

[0015] - Increased detection reliability regarding the location and / or orientation of surrounding sensors.

[0016] In one embodiment, one or more of the yaw angular velocity, pitch angular velocity, and roll angular velocity are determined as angular velocities and undergo histogram evaluation.

[0017] In one embodiment, the determination of the angular velocity of a detected object is made by comparing the vehicle's movement data with data recorded by surrounding sensors or derived therefrom.

[0018] In one embodiment, a lidar and / or radar and / or camera and / or ultrasonic sensor is used as an ambient sensor.

[0019] In one embodiment, one or more of the following lists are used as vehicle motion data: steering angle, yaw angle, roll angle, pitch angle, yaw acceleration, roll acceleration, pitch acceleration, spring travel signal, wheel rotation pulse, position data, and redundant peripheral sensor device.

[0020] In one embodiment, a lidar and / or radar and / or camera and / or ultrasonic sensor is used as a redundant surrounding sensor device.

[0021] According to one aspect of the present invention, a device is provided that includes a data processing unit connected to at least one surrounding sensor and configured to perform one of the methods described above, and detects non-correction of at least one surrounding sensor that records the surroundings of a vehicle.

[0022] According to a further aspect of the present invention, a vehicle is provided that includes at least one surrounding sensor that records the surroundings of such a device and the vehicle.

[0023] In one embodiment, a lidar and / or radar and / or camera and / or ultrasonic sensor is provided as an ambient sensor.

[0024] In one embodiment, one or more sensors are provided to record one or more of the following data for recording vehicle movement data: steering angle, yaw angle, roll angle, pitch angle, yaw acceleration, roll acceleration, pitch acceleration, spring travel signal, wheel rotation pulse, position data, and redundant peripheral sensor device.

[0025] To prevent degradation, the present invention proposes a direct comparison of data from various sensors, in particular, available vehicle movement data and detected sensor data. The two data sources provide redundant information regarding the orientation of the vehicle and the sensor during normal operation, that is, there is no misorientation between the two orientations. Variations in sensor orientation are revealed in the offset between the two data sources with respect to the extracted orientation. Such variations can be detected immediately as soon as input data for processing is provided. Sensitivity is limited by a threshold selected for the allowed offset.

[0026] Possible sensor data are self-perceptual data (e.g., LiDAR point clouds, radar echoes, camera images, ultrasonic echoes) and data derived therefrom.

[0027] Possible vehicle data sources are as follows.

[0028] - Vehicle's own movement data:

[0029] - Steering angle,

[0030] - Yaw angle, roll angle, and pitch angle including related acceleration,

[0031] - Spring travel signal,

[0032] - Wheel rotation speed pulse,

[0033] - Location (GPS, GNSS, map data, etc.),

[0034] - Redundant environmental sensor devices (radar, camera, lidar, ultrasound).

[0035] Embodiments of the present invention will be described in more detail below with reference to the drawings. Brief explanation of the drawing

[0036] Figure 1 is a schematic diagram illustrating a vehicle in normal operation. FIG. 2 is a schematic diagram illustrating a vehicle during a sudden change in the orientation of surrounding sensors. Figure 3 is a schematic histogram illustrating the distribution of the number of objects detected across different angular velocities during normal operation of the vehicle. Figure 4 is a schematic histogram illustrating the distribution of the number of objects detected across different angular velocities when the orientation of the surrounding sensors changes abruptly. Members that match each other are indicated by the same drawing symbol in all drawings. Specific details for implementing the invention

[0037] Fig. 1 This is a schematic diagram illustrating a vehicle (1) in normal operation. At least one surrounding sensor (2) is positioned on the vehicle (1) and oriented in the direction of the vehicle axis (A) or relative to the vehicle axis. The position of surrounding objects detected by the surrounding sensor (2) is accurate.

[0038] Fig. 2This is a schematic diagram illustrating a vehicle (1) during a sudden change in the orientation of the surrounding sensor (2). For example, the orientation of the surrounding sensor (2) deviates from the vehicle axis (A) by an angle, specifically a yaw angle (α). Thus, the object (3) appears to be offset from the detection of the surrounding sensor (2) and is located at a position (3') on the surface. The change in the position of the object (3) in a reference frame moving with the vehicle (1) is the result of the object (3) itself moving or the change in the orientation of the surrounding sensor (2). The sudden change in orientation appears to the surrounding sensor (2) as if the entire arrangement of the object (3) has moved in the same direction. The displacement is r*α, where r is the radial distance of each object (3) relative to the surrounding sensor (2). All objects move simultaneously at the same angular velocity (dα / dt), specifically the azimuth velocity. This movement of the entire arrangement can be detected by an algorithmic filter of the angular velocity (dα / dt) or sensor position. The same analysis can be applied to pitch angle and roll angle, so sudden changes in orientation for all possible scenarios can be detected.

[0039] Fig. 3 It is a schematic histogram showing the distribution of the number of objects (N) detected over different angular velocities (dα / dt) during normal operation of the vehicle (1), that is, when the surrounding sensors (2) are correctly oriented in the direction of the vehicle axis (A) or relative to the vehicle axis.

[0040] The angular velocity (dα / dt) is defined in a reference frame moving with the vehicle (1). In a typical driving scenario, during normal operation, the distribution is wide and centered at 0. The amount of variation in the distribution is determined by the movement of the object (3) around the vehicle (1).

[0041] Fig. 4This is a schematic histogram illustrating the distribution of the number of objects (N) detected across different angular velocities (dα / dt) when the orientation has changed abruptly, that is, when the surrounding sensor (2) is no longer correctly oriented in the direction of the vehicle axis (A) or relative to the vehicle axis. From the perspective of the surrounding sensor (2), the displacement of the orientation of the surrounding sensor (2) appears to be that the entire periphery of the vehicle (1) has moved by an angle (α). Consequently, all perceptual data show a strong component of angular velocity (dα / dt) that overlaps with the actual movement around the vehicle (1). This creates a peak (P) or accumulation (P) that is not at zero in the distribution. The larger the magnitude of each displacement, the faster the displacement occurred. Detecting this accumulation (P) in the distribution of angular velocity (dα / dt) is an effective method for identifying abrupt changes in the orientation of the surrounding sensor (2).

[0042] The angular velocity (dα / dt) of the detected object (3) can be determined by comparing the movement data of the vehicle (1) with the data recorded by or derived from the surrounding sensors (2). For example, lidar and / or radar and / or camera and / or ultrasonic sensors may be used as the surrounding sensors (2). One or more of the following list may be used as the movement data of the vehicle (1): steering angle, yaw angle, roll angle, pitch angle, yaw acceleration, roll acceleration, pitch acceleration, spring travel signal, wheel rotation pulse, position data, and redundant surrounding sensor devices, for example, lidar and / or radar and / or camera and / or ultrasonic sensors.

[0043] The method of the present invention can be performed, for example, in a data processing unit (4) that can be placed in a vehicle (1).

Claims

Claim 1 A method for detecting decalibration of at least one surrounding sensor (2) for detecting the surroundings of a vehicle (1), the method comprises: a step of detecting a plurality of objects (3) within the surroundings by the at least one surrounding sensor (2); a step of determining the angular velocity (dα / dt) of the detected objects (3) when the vehicle (1) is in motion - the angular velocity (dα / dt) of the detected objects (3) is determined by comparing the movement data of the vehicle (1) with the data recorded by or derived therefrom by the surrounding sensor (2); and a step of determining one or more accumulations (P) from a distribution of the number (N) of the objects (3) by histogram evaluation based on the determined angular velocity (dα / dt). A method for detecting non-correction, comprising the step of identifying non-correction of the surrounding sensor (2) in response to the case where one of the determined accumulations (P) is located at a non-zero angular velocity (dα / dt) — wherein the non-correction is caused by the orientation of the surrounding sensor (2) deviating from the normal position relative to the vehicle axis (A) of the vehicle (1). Claim 2 A method for detecting non-correction according to claim 1, characterized in that one or more of yaw angular velocity, pitch angular velocity and roll angular velocity are determined as angular velocity (dα / dt) and undergo histogram evaluation. Claim 3 delete Claim 4 A method for detecting non-correction according to claim 1, characterized in that the surrounding sensor (2) is used, for example, a lidar and / or radar and / or a camera and / or an ultrasonic sensor. Claim 5 A method for detecting non-correction according to claim 1, characterized in that one or more of the following list: steering angle, yaw angle, roll angle, pitch angle, yaw acceleration, roll acceleration, pitch acceleration, spring travel signal, wheel rotation pulse, position data and redundant peripheral sensor device are used as movement data of the vehicle (1). Claim 6 A method for detecting non-correction according to claim 5, characterized in that a lidar and / or radar and / or camera and / or ultrasonic sensor is used as a redundant peripheral sensor device. Claim 7 A device for detecting non-correction of at least one surrounding sensor (2) for detecting the surroundings of a vehicle (1), comprising a data processing unit (4) connected to the at least one surrounding sensor (2), wherein the data processing unit (4) detects a plurality of objects (3) in the surroundings; determines the angular velocity (dα / dt) of the detected objects (3) while the vehicle (1) is driving, wherein the angular velocity (dα / dt) of the detected objects (3) is determined by comparing the movement data of the vehicle (1) with the data recorded by or derived therefrom by the surrounding sensor (2); and determines at least one accumulation (P) from the distribution of the number (N) of the objects (3) by histogram evaluation based on the angular velocity (dα / dt); A non-correcting detection device configured to identify non-correcting of the surrounding sensor (2) in response to when one of the above accumulations (P) is located at a non-zero angular velocity (dα / dt), wherein the non-correcting is caused by the orientation of the surrounding sensor (2) deviating from the normal position relative to the vehicle axis (A) of the vehicle (1). Claim 8 A vehicle (1) comprising a device according to claim 7 and at least one surrounding sensor (2) for detecting the surroundings of the vehicle (1). Claim 9 A vehicle (1) characterized in that, in claim 8, a surrounding sensor (2) is provided with a lidar and / or radar and / or camera and / or ultrasonic sensor. Claim 10 In claim 9, the vehicle (1) comprises one or more sensors configured to acquire one or more of the following data as movement data of the vehicle (1): steering angle, yaw angle, roll angle, pitch angle, yaw acceleration, roll acceleration, pitch acceleration, suspension travel signal, wheel speed pulse, position data and redundant surrounding sensors.

Images