Perception degradation detection method and device of perception device, equipment and storage medium

Abstract

This application relates to a method, apparatus, device, and storage medium for detecting sensor degradation in a sensing device. The method includes: acquiring point cloud data of a current frame image collected by the sensing device; determining signal-to-noise ratio (SNR) probabilities in multiple pose directions of the sensing device based on the point cloud data and the pose information of the sensing device, wherein the SNR probabilities characterize the significance of the signal relative to noise in each pose direction; and performing degradation detection on each pose direction of the sensing device based on the SNR probabilities to obtain a detection result, wherein the detection result is used to determine the pose adjustment magnitude in each pose direction. This method enables accurate detection of the degradation degree in the pose directions of a sensor.

Description

Technical Field

[0001] This application relates to the field of sensor technology, and in particular to a method, apparatus, device, and storage medium for detecting sensor degradation in a sensing device. Background Technology

[0002] With the development of sensor technology, the effectiveness of applications such as Simultaneous Localization and Mapping (SLAM) using sensors, such as robot navigation and autonomous driving, depends on the accuracy of the measurement data.

[0003] However, in some data acquisition scenarios, such as utility tunnels and other similar environments, the lack of geometric information in the acquisition environment, or the mismatch between the sensor's perspective and the environment's geometry, leads to limitations in observation and a degradation in the sensor's pose orientation.

[0004] Therefore, it is necessary to accurately detect the degree of degradation in the sensor's pose orientation. Summary of the Invention

[0005] Therefore, it is necessary to provide a sensing degradation detection method, apparatus, device, and storage medium for a sensing device that can accurately detect the degree of degradation of the sensor's pose orientation, in order to address the aforementioned technical problems.

[0006] Firstly, this application provides a method for detecting sensor degradation in a sensing device. The method includes:

[0007] Acquire point cloud data of the current frame image collected by the sensing device;

[0008] Based on the point cloud data and the pose information of the sensing device, the signal-to-noise ratio probability in multiple pose directions of the sensing device is determined. The signal-to-noise ratio probability is used to characterize the significance of the signal relative to the noise in each pose direction.

[0009] Based on the signal-to-noise ratio probabilities, degradation detection is performed on each pose direction of the sensing device to obtain detection results, which are used to determine the pose adjustment range in each pose direction.

[0010] In one embodiment, determining the signal-to-noise ratio probability in multiple pose directions of the sensing device based on the point cloud data and the pose information of the sensing device includes:

[0011] Based on the point cloud data, the normal vector corresponding to each point cloud data, and the pose information, a contribution vector of each position point included in the point cloud data is constructed, and the Hessian matrix corresponding to the point cloud data is obtained by summing the contribution vectors.

[0012] Perturbation analysis is performed on each of the aforementioned locations to obtain perturbation analysis results. Based on the perturbation analysis results, the noise covariance matrix corresponding to each of the aforementioned locations is calculated, and the sum of the noise covariance matrices is determined as the total noise covariance matrix.

[0013] The signal-to-noise ratio probabilities are determined based on the Hessian matrix, each of the noise covariance matrices, and the total noise covariance matrix.

[0014] In one embodiment, determining each of the signal-to-noise ratio probabilities based on the Hessian matrix, each of the noise covariance matrices, and the total noise covariance matrix includes:

[0015] For each of the pose directions, the mean noise value in the pose direction is determined based on the total noise covariance matrix;

[0016] Based on the noise covariance matrices and the contribution vectors of each position point to the Hessian matrix, the noise variance in the pose direction is determined.

[0017] Projecting the Hessian matrix onto the pose direction yields the directional observation value in the pose direction;

[0018] Based on the noise mean, the noise variance, and the direction observation, the signal-to-noise ratio probability in the pose direction is determined to determine each of the signal-to-noise ratio probabilities.

[0019] In one embodiment, determining the signal-to-noise ratio probability in the pose direction based on the noise mean, the noise variance, and the direction observation includes:

[0020] For each pose direction, a normal distribution model of the noise components in the pose direction is established based on the noise mean and the noise variance.

[0021] Based on the preset signal-to-noise ratio coefficient and the observed direction value, construct the signal-to-noise ratio condition;

[0022] The probability of satisfying the signal-to-noise ratio condition is determined as the signal-to-noise ratio probability in the pose direction.

[0023] In one embodiment, determining the probability of satisfying the signal-to-noise ratio condition as the signal-to-noise ratio probability in the pose direction includes:

[0024] The noise components are standardized to obtain standardized noise components.

[0025] Based on the signal-to-noise ratio condition, determine the critical value of the normalized noise component;

[0026] The critical value is substituted into the cumulative distribution function of the preset standard normal distribution, and the function value of the cumulative distribution function is used as the signal-to-noise ratio probability.

[0027] In one embodiment, the step of performing degradation detection on each pose direction of the sensing device based on each signal-to-noise ratio probability to obtain a detection result includes:

[0028] For each of the signal-to-noise ratio probabilities, detect whether the signal-to-noise ratio probability is greater than a preset probability threshold;

[0029] If the signal-to-noise ratio probability is greater than the probability threshold, then the detection result is determined to be that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability;

[0030] If the signal-to-noise ratio probability is less than or equal to the probability threshold, then the detection result is determined to be that there is no degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0031] Secondly, this application also provides a sensing degradation detection device for a sensing device. The device includes:

[0032] The acquisition module is used to acquire point cloud data of the current frame image collected by the sensing device;

[0033] The determination module is used to determine the signal-to-noise ratio probability in multiple pose directions of the sensing device based on the point cloud data and the pose information of the sensing device. The signal-to-noise ratio probability is used to characterize the significance of the signal relative to the noise in each pose direction.

[0034] The detection module is used to perform degradation detection on each pose direction of the sensing device according to each signal-to-noise ratio probability, and obtain the detection result. The detection result is used to determine the pose adjustment range in each pose direction.

[0035] Thirdly, this application also provides an electronic device. The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method described in the first aspect.

[0036] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0037] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of the method described in the first aspect.

[0038] The aforementioned method, apparatus, device, and storage medium for detecting perception degradation in sensing devices involve the electronic device first acquiring point cloud data of the current frame image collected by the sensing device. Then, based on the point cloud data and the pose information of the sensing device, the signal-to-noise ratio (SNR) probabilities in multiple pose directions of the sensing device are determined. Since the SNR probabilities can characterize the significance of the signal relative to noise in each pose direction, they can provide a quantitative basis for degradation detection. Subsequently, based on each SNR probability, degradation detection is performed on each pose direction of the sensing device to obtain detection results used to determine the pose adjustment magnitude in each pose direction. Compared with the traditional binary hard thresholding rule, the method of determining the detection results based on the SNR probabilities can be adaptively applied to various sensing scenarios, significantly improving the accuracy and robustness of the detection results. This allows for accurate determination of the pose adjustment magnitude in each pose direction based on the detection results, effectively preventing the electronic device from blindly adjusting its pose in the presence of perception degradation, thereby improving the motion control performance and navigation stability of the electronic device. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a diagram illustrating the application environment of a sensing degradation detection method for a sensing device in one embodiment.

[0041] Figure 2 This is a flowchart illustrating a sensing degradation detection method for a sensing device in one embodiment;

[0042] Figure 3 This is a flowchart illustrating step 202 in one embodiment;

[0043] Figure 4 This is a flowchart illustrating step 303 in one embodiment;

[0044] Figure 5 This is a flowchart illustrating step 404 in one embodiment;

[0045] Figure 6 This is a flowchart illustrating step 503 in one embodiment;

[0046] Figure 7 This is a flowchart illustrating step 203 in one embodiment;

[0047] Figure 8 This is a structural block diagram of a sensing degradation detection device in one embodiment;

[0048] Figure 9 This is a diagram of the internal structure of an electronic device in one embodiment. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0050] In SLAM applications, point cloud registration algorithms are required, and degradation detection is usually an intermediate step in the point cloud registration process. When updating point cloud data, the update magnitude is adjusted by identifying the degradation direction of the sensing device. For example, for pose directions with significant degradation, update suppression is used to match the geometric characteristics of the surrounding environment.

[0051] However, in scenarios such as utility tunnels and underground passages, the sensing environment of radar and other sensing devices often lacks sufficient feature points, textures, and geometric information. This makes it difficult for these devices to acquire enough constraints for accurate localization and mapping. Furthermore, due to factors such as changes in lighting, occlusion, and dynamic changes, the data observed by the sensing devices may not correctly match the elements in the map, leading to significant degradation in the sensing capabilities and impacting application performance. Current technologies typically employ hard thresholding rules or learning-based models to detect this degradation.

[0052] For hard threshold rule-based methods, manually set thresholds often only apply to a single specific scenario in different utility tunnel environments. When applied to other scenarios, significant manpower is required to re-adjust and set new thresholds. Therefore, this method struggles to cope with the diversity and dynamic changes of actual utility tunnel environments. Furthermore, the lack of quantifiable error constraints in threshold setting makes it impossible to pre-assess the potential positioning error range introduced by the detection results. This leads to unstable reliability of degradation detection. When the sensing data from the equipment within the utility tunnel is affected by environmental interference such as dust or water reflection, the blindness of the threshold further amplifies the probability of misjudgment, making it difficult to support the need for precise adjustments to subsequent positioning strategies.

[0053] Most learning-based methods employ complex model structures such as deep neural networks. Their decision-making process relies on the nonlinear combination and weight allocation of hidden neurons. The intermediate features of these hidden neurons often lack intuitive physical meaning or clear geometric constraints, making it impossible to trace the basis for the generation of degradation detection results through quantifiable physical indicators such as feature value magnitude and condition number threshold. When the model makes an incorrect prediction about the degradation of a certain utility tunnel scene, it is difficult to pinpoint the root cause of the problem from the weight update trajectory or feature mapping process within the model. For example, it is impossible to determine whether the lack of samples of this type of scene in the training data leads to distribution mismatch, whether the model fails to filter out interference information such as dust and water reflection during the feature extraction stage, or whether the activation function setting of the output layer is unreasonable, causing the decision boundary to shift. Therefore, this black-box characteristic makes the troubleshooting and optimization of learning-based methods lack a clear direction, greatly limiting their application in utility tunnel positioning systems with stringent reliability requirements.

[0054] In view of this, this application proposes a method for detecting perceptual degradation in a sensing device. The electronic device first acquires point cloud data of the current frame image collected by the sensing device. Then, based on the point cloud data and the pose information of the sensing device, it determines the signal-to-noise ratio (SNR) probabilities in multiple pose directions of the sensing device. Since the SNR probabilities can characterize the significance of the signal relative to noise in each pose direction, it can provide a quantitative basis for degradation detection. Subsequently, based on each SNR probability, degradation detection is performed on each pose direction of the sensing device to obtain detection results used to determine the pose adjustment magnitude in each pose direction. Compared with the traditional binary hard thresholding rule, the method of determining the detection results based on the SNR probabilities can be adaptively applied to various sensing scenarios, significantly improving the accuracy and robustness of the detection results. Thus, the pose adjustment magnitude in each pose direction can be accurately determined based on the detection results, effectively avoiding the electronic device blindly adjusting its pose in the presence of perceptual degradation, thereby improving the motion control performance and navigation stability of the electronic device.

[0055] The sensing degradation detection method for sensing devices provided in this application embodiment can be applied to, for example... Figure 1In the application environment shown, electronic device 102 communicates with sensing device 104 via a network. A data storage system can store the data that electronic device 102 needs to process. The data storage system can be integrated into electronic device 102 or placed in the cloud or on another network server. Electronic device 102 first acquires point cloud data of the current frame image collected by sensing device 104; then, based on the point cloud data and the pose information of the sensing device, it determines the signal-to-noise ratio (SNR) probability in multiple pose directions of the sensing device. The SNR probability characterizes the significance of the signal relative to noise in each pose direction; subsequently, based on each SNR probability, it performs degradation detection on each pose direction of the sensing device to obtain detection results, which are used to determine the pose adjustment range in each pose direction. Electronic device 102 can be, but is not limited to, various drones, robots, and other electronic devices. Sensing device 104 includes, but is not limited to, one or more of LiDAR, millimeter-wave radar, ultrasonic sensors, sonar, and depth cameras.

[0056] In one exemplary embodiment, such as Figure 2 As shown, a method for detecting sensor degradation in a sensing device is provided, which is then applied to... Figure 1 Taking an electronic device as an example, the explanation includes the following steps:

[0057] 201. Obtain the point cloud data of the current frame image collected by the sensing device.

[0058] It should be noted that the sensing device can collect image information from the sensing environment in real time. Each frame of image information includes a set of point cloud data, that is, the point cloud data includes the position information of multiple points, for example, the position information can be represented as (x, y, z). It can be understood that the point cloud data is measurement data affected by noise.

[0059] The current frame image refers to the frame image captured by the sensing device at the current moment.

[0060] In this embodiment, the electronic device can obtain in real time the point cloud data included in the current frame image collected by the sensing device at the current moment from the sensing device.

[0061] 202. Based on the point cloud data and the pose information of the sensing device, determine the signal-to-noise ratio probability in multiple pose directions of the sensing device.

[0062] The signal-to-noise ratio probability is used to characterize the significance of the signal relative to noise in each pose direction.

[0063] The pose information of the sensing device includes its position and orientation in three-dimensional space. For example, the position information includes the X-axis, Y-axis, and Z-axis coordinates of the sensing device in the global coordinate system, and the orientation information includes the roll angle (the angle of rotation around the X-axis), pitch angle (the angle of rotation around the Y-axis), and yaw angle (the angle of rotation around the Z-axis). Multiple pose directions include the X-axis direction, Y-axis direction, Z-axis direction, rotation direction around the X-axis, rotation direction around the Y-axis, and rotation direction around the Z-axis.

[0064] Understandably, in practical applications, even without actual scene degradation, minute noise disturbances can trigger erroneous degradation constraint instructions, leading to decreased pose update accuracy or even interruption of the registration process. Therefore, during degradation detection, the signal-to-noise ratio (SNR) probabilities in multiple pose directions of the sensing device can be determined. Then, based on the SNR probabilities, the presence of degradation can be determined, thereby achieving adaptive determination of the degradation direction. For example, if the SNR probability of the sensing device in a certain pose direction indicates that the noise intensity is significantly stronger than the signal intensity, then it can be determined that sensing degradation exists in that pose direction.

[0065] In this embodiment, the electronic device can perform uncertainty analysis on point cloud data to obtain noise statistical characteristics in each pose direction. Then, it calculates whether each noise statistical characteristic meets the preset signal-to-noise ratio (SNR) condition. After that, it calculates the proportion that meets the condition and determines the proportion as the SNR probability in each pose direction.

[0066] 203. Based on the probability of each signal-to-noise ratio, degradation detection is performed on each pose direction of the sensing device to obtain the detection results.

[0067] The detection results are used to determine the pose adjustment range in each pose direction. The pose adjustment range refers to the amount by which an electronic device is allowed to change in a certain pose direction.

[0068] Understandably, if perceptual degradation exists in a certain pose direction, it can cause the electronic device to move erroneously in the unconstrained direction, leading to collisions, or project erroneous data to the wrong location, resulting in map distortion, ghosting, etc. Therefore, if the detection result indicates perceptual degradation in a certain pose direction, it is necessary to suppress the update of the pose adjustment magnitude of the electronic device in that pose direction. That is, set the corresponding increment in that pose direction to zero, or reduce the update weight in that pose direction, thereby preventing the electronic device from relying on the erroneous data control system of the sensing device. If the detection result indicates that there is no perceptual degradation in a certain pose direction, the electronic device can rely on the data control system in that pose direction to adjust the movement direction and amplitude of the electronic device.

[0069] Optionally, the detection result may include whether degradation exists or not. Optionally, the higher the signal-to-noise ratio probability, the greater the pose adjustment amplitude; the lower the signal-to-noise ratio probability, the smaller the pose adjustment amplitude, up to zero.

[0070] In this embodiment, for the signal-to-noise ratio (SNR) probability in a certain pose direction, the electronic device can compare the SNR probability with the degradation probability and the non-degradation probability respectively. If the SNR probability is closer to the degradation probability, that is, the SNR probability is approximately equal to the degradation probability, then the detection result can be determined that there is degradation in that pose direction; if the SNR probability is closer to the non-degradation probability, that is, the SNR probability is approximately equal to the non-degradation probability, then the detection result can be determined that there is no degradation in that pose direction, thereby obtaining the detection result of whether degradation exists in each pose direction.

[0071] In this way, when abnormal point cloud frames are generated in the utility tunnel due to local pipeline obstruction or temporary strong light interference, the filtering mechanism based on the signal-to-noise ratio probability criterion can effectively remove invalid data, thereby improving the continuous and stable output rate of the positioning results. Moreover, the root cause of the degradation direction can be quickly located through the signal-to-noise ratio probability distribution curve.

[0072] In the aforementioned method for detecting sensor degradation, the electronic device first acquires point cloud data of the current frame image collected by the sensor. Then, based on the point cloud data and the pose information of the sensor, it determines the signal-to-noise ratio (SNR) probabilities in multiple pose directions of the sensor. Since the SNR probabilities can characterize the significance of the signal relative to noise in each pose direction, it can provide a quantitative basis for degradation detection. Subsequently, based on each SNR probability, degradation detection is performed on each pose direction of the sensor to obtain detection results used to determine the pose adjustment magnitude in each pose direction. Compared with the traditional binary hard thresholding rule, the method of determining the detection results based on the SNR probabilities can be adaptively applied to various sensing scenarios, significantly improving the accuracy and robustness of the detection results. This allows for accurate determination of the pose adjustment magnitude in each pose direction based on the detection results, effectively preventing the electronic device from blindly adjusting its pose in the presence of sensor degradation, thereby improving the motion control performance and navigation stability of the electronic device.

[0073] In one exemplary embodiment, such as Figure 3 As shown, this embodiment relates to the process by which an electronic device determines the signal-to-noise ratio probability in multiple pose directions of the sensing device based on point cloud data and the pose information of the sensing device. Step 202 includes:

[0074] 301. Based on each point cloud data, the corresponding normal vector of each point cloud data, and each pose information, construct the contribution vector of each position point included in the point cloud data, and obtain the Hessian matrix corresponding to the point cloud data by summing the contribution vectors.

[0075] It should be noted that in the Iterative Closest Point (IPC) algorithm, the sensing device continuously scans environmental data, and the point clouds from different times need to be unified into the same coordinate system for environmental modeling. The core lies in obtaining the homogeneous matrix. This transforms the source point cloud to the same coordinate system as the target point cloud, thereby integrating all scanned geometric structures into a unified coordinate system. Homogeneous matrix In It is a 3D matrix used to represent the rotation operation of the sensing device's coordinate system relative to a unified coordinate system. It is the translation vector in the coordinates of the sensing device, therefore The format is as follows:

[0076]

[0077] The point-to-plane measurement is achieved through normal vector projection, thus ignoring reasonable slippage of sampling points on the plane and focusing on eliminating structural deviations perpendicular to the surface. This provides the correct gradient direction conforming to physical geometry for the iterative process, improving registration accuracy and convergence speed. The point-to-plane distance can be expressed as:

[0078]

[0079] in, This represents the distance from a point to a plane. It is noise-free point cloud data in the coordinate system of the sensing device. Let be the plane normal vector per unit length of the plane containing the target point cloud. This is the signed orthogonal distance from the origin to the plane.

[0080] in, This can be obtained by optimizing the minimum error distance from the point to the plane:

[0081]

[0082] in, It is the L2 norm. For non-negative weights, given a differential perturbation , For rotation vector perturbation, To obtain the error due to the translation vector perturbation, performing a first-order Taylor decomposition on the above equation yields the result. Linear relationship:

[0083]

[0084] in, , Linearization is performed using the initial guesses, and the optimization problem is formulated as a least-squares problem (minimizing the extreme value):

[0085]

[0086] The optimal differential perturbation can be obtained by solving the following equation. :

[0087]

[0088] in, The Hessian matrix representing the optimization problem:

[0089]

[0090] In this embodiment, since the point cloud data is measurement data subject to additive Gaussian noise, it is necessary to construct the Hessian matrix of the point cloud data obtained from the sensing device based on the Hessian matrix mentioned above, according to each point cloud data, the normal vector corresponding to each point cloud data, and the pose information of each point cloud data. This Hessian matrix is ​​a Hessian matrix containing noise.

[0091] The point cloud data includes the contribution vectors of each location point. It can be represented as:

[0092]

[0093] in, This represents the contribution weight of the i-th position. The position noise of the i-th point in the point cloud data can be represented as:

[0094]

[0095] The noise of the normal vector corresponding to the i-th location point in the point cloud data can be expressed as:

[0096]

[0097] in, Let i be the actual position of the i-th position point. Let be the multivariate Gaussian distribution of the covariance of the i-th location point. Let be the value of the small perturbation that the normal vector at the i-th position point experiences. Let be the normal vector of the i-th position point.

[0098] Thus, the Hessian matrix corresponding to the point cloud data can be represented as:

[0099]

[0100] 302. Perform disturbance analysis on each location point to obtain the disturbance analysis results, calculate the noise covariance matrix corresponding to each location point based on the disturbance analysis results, and determine the total noise covariance matrix by summing the noise covariance matrices.

[0101] The perturbation analysis result is obtained by considering point position noise and normal vector rotation perturbation, establishing the relationship between the contribution vector and noise at each point through a first-order Taylor expansion, and calculating the covariance matrix of the contribution vector accordingly. This covariance matrix serves as the perturbation analysis result for that point, which is used to describe how noise propagates to subsequent matrices such as the Hessian matrix.

[0102] The noise covariance matrix at each location point describes the contribution vector of each location point to that location point. The uncertainty of the total noise covariance matrix refers to the sum of the covariance matrices of all points, which reflects the total noise intensity distribution of the entire point cloud data in each pose direction and is used to calculate the mean of directional noise.

[0103] It should be noted that, combining the aforementioned position noise and normal vector noise, the contribution vector... By performing analysis, we can obtain the contribution vector. Compared with the true contribution vector under noise-free conditions Point noise Normal vector noise The relationship between them:

[0104]

[0105] in, .

[0106] Then, by combining the aforementioned position noise and normal vector noise, signal-to-noise separation and decoupling are performed on the Hessian matrix corresponding to the point cloud data, the total noise covariance matrix can be obtained. :

[0107]

[0108] The noise covariance matrix corresponding to the i-th location point is: . The Hessian matrix is ​​noise-free, and the pose direction is... The perturbation variance on can be expressed as:

[0109]

[0110] 303. Based on the Hessian matrix, each noise covariance matrix, and the total noise covariance matrix, determine each signal-to-noise ratio probability.

[0111] In this implementation, the electronic device can generate a large number of random noise vectors for each position point based on the noise covariance matrix of that position point to simulate the possible fluctuations of the contribution vector at that point. Then, the noise samples of each point are accumulated to obtain the samples of the overall noise matrix. Next, for each set of samples, a noisy Hessian matrix is ​​constructed. Then, for each pose direction, the directional observation value of each set of samples is calculated. Finally, the frequency at which the signal is submerged by noise is counted, which is the signal-to-noise ratio probability of that pose direction, thereby obtaining the signal-to-noise ratio probability of each pose direction.

[0112] In this embodiment, the electronic device first constructs the contribution vector of each position point in the point cloud data based on each point cloud data, the corresponding normal vector, and each pose information. Then, it obtains the Hessian matrix corresponding to the point cloud data by summing the contribution vectors. Simultaneously, it performs perturbation analysis on each position point to obtain the perturbation analysis results. Based on these results, it calculates the noise covariance matrix corresponding to each position point and determines the total noise covariance matrix by summing the noise covariance matrices. Finally, it determines each signal-to-noise ratio (SNR) probability based on the Hessian matrix, each noise covariance matrix, and the total noise covariance matrix. Since the determination of the SNR probability comprehensively considers the geometric constraint information reflected by the Hessian matrix, the local uncertainties described by each noise covariance matrix, and the overall noise distribution represented by the total noise covariance matrix, it more comprehensively and accurately quantifies the significance of the signal relative to noise in each pose direction, thereby improving the accuracy of the determined SNR probabilities.

[0113] In one exemplary embodiment, such as Figure 4 As shown, this embodiment relates to the process by which an electronic device determines each signal-to-noise ratio probability based on the Hessian matrix, each noise covariance matrix, and the total noise covariance matrix. Step 303 includes:

[0114] 401. For each pose direction, determine the mean noise value in the pose direction based on the total noise covariance matrix.

[0115] It should be noted that the process of determining the signal-to-noise ratio probability is similar for each pose direction. Therefore, in this embodiment, the process is exemplarily described using a pose direction as an example.

[0116] The noise mean refers to the average magnitude of the noise components in a certain pose direction, describing the systematic deviation or central tendency of the noise components in that pose direction. If the noise mean is positive, it means that the noise tends to overestimate the observed value; if it is negative, it overestimates the observed value.

[0117] In this embodiment, the total noise covariance matrix can be projected onto a certain pose direction, that is, a quadratic operation is performed between the unit direction vector and the total noise covariance matrix, and the result is the noise mean in that pose direction.

[0118] Wherein, the noise mean in the pose direction u It can be represented as:

[0119]

[0120] 402. Based on the noise covariance matrices and the contribution vectors of each position point to the Hessian matrix, determine the noise variance in the pose direction.

[0121] Noise variance refers to the degree of fluctuation of the noise component in a certain pose direction, describing the uncertainty or dispersion of the noise in that pose direction. Optionally, a larger noise variance indicates that the noise is more unstable and the observations in that pose direction are less reliable, while a smaller noise variance indicates that the noise is more concentrated and the observations in that pose direction are more stable.

[0122] The contribution vector of each point to the Hessian matrix represents the contribution of that point to the Hessian matrix.

[0123] It should be noted that the process of determining the signal-to-noise ratio probability is similar for each pose direction. Therefore, in this embodiment, the process is exemplarily described using a pose direction as an example.

[0124] In this embodiment, the noise covariance matrix of each position point can be projected onto a certain pose direction to obtain the numerical value of the concentration of noise at each position point in that pose direction. Then, the contribution vector of each position point is projected onto the same pose direction to obtain a value reflecting the magnitude of the contribution of each position point in the same pose direction. Then, the above two values ​​are used as the contribution of each position point to the noise variance in that pose direction. Finally, the contributions of the noise variance of all position points are added together to obtain the noise variance in that pose direction.

[0125] The noise variance in the pose direction can be expressed as:

[0126]

[0127] 403. Project the Hessian matrix onto the pose direction to obtain the directional observation value in the pose direction.

[0128] The orientation observation reflects the geometric constraint strength in a certain pose direction, but contains noise.

[0129] It is understandable that the Hessian matrix contains constraint information in all pose directions. In order to clarify the constraint information in each pose direction, the constraint information in each pose direction can be extracted by projection.

[0130] In this embodiment, for a certain pose direction, the result of the quadratic operation between the unit direction vector and the Hessian matrix can be determined as the direction observation value in that pose direction.

[0131] The direction observation value in the pose direction u can be expressed as:

[0132]

[0133] in, Let be the intensity of the true signal in the pose direction u. Let be the noise intensity in the pose direction u.

[0134] 404. Based on the noise mean, noise variance, and direction observations, determine the signal-to-noise ratio probability in the pose direction to determine each signal-to-noise ratio probability.

[0135] It should be noted that the expectation and variance of the Hessian matrix correspond to the mean and variance of a non-central chi-square distribution, respectively. Since noise-free ground truth points / normal vectors are unobservable in real-world scenarios, noisy observation points / normal vectors are used to estimate the noise covariance matrix and the total noise covariance matrix.

[0136] Although the measurement error at a single location point follows an asymmetric distribution, calculating the total error of the electronic device's sensing system requires the errors from multiple locations. According to the central limit theorem in statistics, if the number of samples involved in the accumulation is sufficiently large, its statistical characteristics will automatically exhibit a symmetric, smooth Gaussian distribution. Therefore, a simple Gaussian model can be used to describe the noise characteristics of the entire sensing system, assuming that the noise in the pose direction u follows a distribution... .

[0137] In this embodiment, for each pose direction, the electronic device can construct a noise distribution function based on the noise mean, noise variance, and direction observations of that pose direction. Then, based on the properties of the distribution function, the signal and noise distribution information of that pose direction is determined. Subsequently, based on the signal and noise distribution information, the signal-to-noise ratio probability is determined, thereby obtaining the signal-to-noise ratio probability for each pose direction.

[0138] In this embodiment, the electronic device determines the noise mean in each pose direction based on the total noise covariance matrix, and the noise variance in each pose direction based on the noise covariance matrices and the contribution vector of each position point to the Hessian matrix. The Hessian matrix is ​​then projected onto the pose direction to obtain the directional observation value in that pose direction. Subsequently, the signal-to-noise ratio (SNR) probability in the pose direction is determined based on the noise mean, noise variance, and directional observation value. By fully integrating global noise statistics and local point cloud uncertainty information, the device can more comprehensively and accurately quantify the significance of the signal relative to noise in each pose direction, thereby effectively improving the accuracy of the determined SNR probabilities.

[0139] In one exemplary embodiment, such as Figure 5 As shown, this embodiment relates to the process by which an electronic device determines the signal-to-noise ratio probability in the pose direction based on the noise mean, noise variance, and orientation observations. Step 404 includes:

[0140] 501. For each pose direction, a normal distribution model of the noise components in the pose direction is established based on the noise mean and noise variance.

[0141] It should be noted that by establishing a normal distribution model, the statistical regularity of noise can be fully described, and the signal-to-noise ratio probability can be determined based on the properties of the normal distribution model. Specifically, the noise mean determines the central location of the normal distribution model, and the noise variance determines the degree of dispersion of the normal distribution model.

[0142] In this embodiment, taking a certain pose direction as an example, the electronic device can set the noise component in the pose direction as a random variable, and then use the noise mean as the center position of the random variable and the noise variance as the dispersion of the random variable to obtain the normal distribution model of the noise component in the pose direction.

[0143] 502. Based on the preset signal-to-noise ratio coefficient and direction observation value, construct the signal-to-noise ratio condition.

[0144] The signal-to-noise ratio coefficient can be set to a constant s to control the conservatism of the detection. Optionally, the larger the constant s, the stronger the signal is required to be considered non-degenerate; the smaller the constant s, the more sensitive the detection.

[0145] It should be noted that the orientation observation is composed of the real signal and noise, i.e., orientation observation = real signal + noise. When the pose orientation u is a reliable estimate, the real signal in that pose orientation... It should be significantly greater than the noise. So, the real signal Constant s × noise.

[0146] In this embodiment, the probability that the signal strength is s times greater than the noise strength in the pose direction u is... for:

[0147]

[0148] in, These are directional observations.

[0149] 503, the probability of satisfying the signal-to-noise ratio condition is determined as the signal-to-noise ratio probability in the pose direction.

[0150] Understandably, since noise is random and its exact magnitude in a single measurement cannot be determined, only its statistical regularity is known, the probability that the signal strength is greater than the noise strength is defined as the signal-to-noise ratio probability.

[0151] Here, the probability of satisfying the signal-to-noise ratio condition refers to the probability that makes it possible to achieve the desired signal-to-noise ratio. The noise value that is valid. Therefore, in this embodiment, as long as the probability value that satisfies the signal-to-noise ratio condition is obtained, it can be determined as the signal-to-noise ratio probability in the pose direction.

[0152] In one possible implementation, such as Figure 6 As shown, step 503 above may include:

[0153] 601. The noise components are standardized to obtain the standardized noise components.

[0154] Standardization transformation refers to converting general noise into standard form noise, that is, transforming the normal distribution of mean and variance into a standard normal distribution with a mean of 0 and a variance of 1.

[0155] In this embodiment, a new random variable is obtained by subtracting the noise mean from the noise component and then dividing by its standard deviation. This standardized noise component can be expressed as:

[0156]

[0157] 602. Determine the critical value of the normalized noise component based on the signal-to-noise ratio condition.

[0158] The critical value of the normalized noise component refers to the value that the normalized noise component must be less than for the signal-to-noise ratio condition to be met.

[0159] In this embodiment, the noise component can be represented as Then, substituting the noise components into the signal-to-noise ratio condition, we obtain... Finally, solve... The inequalities that are satisfied yield the critical value. .

[0160] 603, substitute the critical value into the cumulative distribution function of the preset standard normal distribution, and use the function value of the cumulative distribution function as the signal-to-noise ratio probability.

[0161] The cumulative distribution function of the standard normal distribution along the attitude direction u can be expressed as:

[0162]

[0163] In this embodiment, the signal-to-noise ratio probability can be obtained by solving the cumulative distribution function described above.

[0164] In this embodiment, the electronic device establishes a normal distribution model of the noise components in each pose direction based on the noise mean and noise variance. Then, based on the preset signal-to-noise ratio coefficient and the direction observation value, it constructs the signal-to-noise ratio condition. The probability of satisfying the signal-to-noise ratio condition can then be determined as the signal-to-noise ratio probability in the pose direction. This allows the significance of the signal relative to the noise to be quantified into a clear probability value, making the degradation detection results continuous, interpretable, and reliable.

[0165] In one exemplary embodiment, such as Figure 7 As shown, this embodiment relates to the process by which an electronic device performs degradation detection on each pose direction of a sensing device based on each signal-to-noise ratio probability, and obtains the detection result. Step 203 includes:

[0166] Step 701: For each signal-to-noise ratio probability, detect whether the signal-to-noise ratio probability is greater than a preset probability threshold.

[0167] The probability threshold is a critical value used to determine whether degradation exists in each pose direction. For example, the probability threshold can be set to 0.5.

[0168] In this embodiment, the electronic device can compare the signal-to-noise ratio probability with a probability threshold to obtain a comparison result, and then determine whether the signal-to-noise ratio probability is greater than the probability threshold based on the comparison result.

[0169] Step 702: If the signal-to-noise ratio probability is greater than the probability threshold, then the detection result is determined to be that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0170] In this embodiment, if the electronic device determines that the noise intensity in the pose direction is greater than the signal intensity when the signal-to-noise ratio probability is greater than the probability threshold, then it can determine that there is degradation in the pose direction, that is, the detection result is that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0171] Step 703: If the signal-to-noise ratio probability is less than or equal to the probability threshold, then the detection result is determined to be that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0172] In this embodiment, if the electronic device determines that the noise intensity in the pose direction is less than or equal to the signal intensity when the signal-to-noise ratio probability is less than or equal to the probability threshold, then it can be determined that there is no degradation in the pose direction, that is, the detection result is that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0173] In this embodiment, the electronic device detects whether the signal-to-noise ratio (SNR) probability is greater than a preset probability threshold for each SNR probability. When the SNR probability is greater than the probability threshold, it determines that there is degradation in the pose direction corresponding to the SNR probability. When the SNR probability is less than or equal to the probability threshold, it determines that there is no degradation in the pose direction corresponding to the SNR probability. Thus, the detection result can be quickly determined based on the probability threshold and the SNR probability, thereby improving the efficiency of the detection result determination.

[0174] To facilitate understanding by those skilled in the art, the sensing degradation detection method for the sensing device provided in this application will be described in detail below. This method may include:

[0175] S1, acquire point cloud data of the current frame image collected by the sensing device.

[0176] S2. Based on each point cloud data, the corresponding normal vector of each point cloud data, and each pose information, construct the contribution vector of each position point included in the point cloud data, and obtain the Hessian matrix corresponding to the point cloud data by summing the contribution vectors.

[0177] S3, perform disturbance analysis on each location point to obtain the disturbance analysis results.

[0178] S4. Calculate the noise covariance matrix corresponding to each location point based on the results of each disturbance analysis, and determine the total noise covariance matrix by summing the noise covariance matrices.

[0179] S5. For each pose direction, determine the mean noise value in the pose direction based on the total noise covariance matrix.

[0180] S6. Determine the noise variance in the pose direction based on each noise covariance matrix and the contribution vector of each position point to the Hessian matrix.

[0181] S7. Project the Hessian matrix onto the pose direction to obtain the directional observation value in the pose direction.

[0182] S8. For each pose direction, establish a normal distribution model of the noise components in the pose direction based on the noise mean and noise variance.

[0183] S9. Based on the preset signal-to-noise ratio coefficient and direction observation value, construct the signal-to-noise ratio condition.

[0184] S10 performs a normalization transformation on the noise components to obtain normalized noise components.

[0185] S11, Determine the critical value of the normalized noise component based on the signal-to-noise ratio condition.

[0186] S12, substitute the critical value into the cumulative distribution function of the preset standard normal distribution, and use the function value of the cumulative distribution function as the signal-to-noise ratio probability.

[0187] S13, for each signal-to-noise ratio probability, detect whether the signal-to-noise ratio probability is greater than a preset probability threshold.

[0188] S14. If the signal-to-noise ratio probability is greater than the probability threshold, then the detection result is determined to be that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0189] S15, if the signal-to-noise ratio probability is less than or equal to the probability threshold, then the detection result is determined to be that there is no degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0190] It should be noted that the descriptions in S1-S15 above can be found in the relevant descriptions in the above embodiments, and their effects are similar, so they will not be repeated here.

[0191] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0192] Based on the same inventive concept, this application also provides a sensing degradation detection device for implementing the sensing degradation detection method of the sensing device described above. The solution provided by this device is similar to the implementation described in the above method. Therefore, the specific limitations in one or more sensing degradation detection device embodiments provided below can be found in the limitations of the sensing degradation detection method of the sensing device described above, and will not be repeated here.

[0193] In one embodiment, such as Figure 8 As shown, a sensing degradation detection device for a sensing device is provided, comprising: an acquisition module 801, a determination module 802, and a detection module 803, wherein:

[0194] The acquisition module 801 is used to acquire point cloud data of the current frame image collected by the sensing device;

[0195] The determination module 802 is used to determine the signal-to-noise ratio probability in multiple pose directions of the sensing device based on point cloud data and pose information of the sensing device. The signal-to-noise ratio probability is used to characterize the significance of the signal relative to the noise in each pose direction.

[0196] The detection module 803 is used to perform degradation detection on each pose direction of the sensing device according to each signal-to-noise ratio probability, and obtain the detection result. The detection result is used to determine the pose adjustment range in each pose direction.

[0197] The sensing degradation detection device for the sensing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0198] In one embodiment, the determining module 802 includes:

[0199] The construction unit is used to construct the contribution vector of each position point included in the point cloud data based on each point cloud data, the normal vector corresponding to each point cloud data, and the pose information of each point cloud data, and to obtain the Hessian matrix corresponding to the point cloud data based on the sum of each contribution vector.

[0200] The first determining unit is used to perform disturbance analysis on each location point, obtain disturbance analysis results, calculate the noise covariance matrix corresponding to each location point based on each disturbance analysis result, and determine the total noise covariance matrix by summing the noise covariance matrices.

[0201] The second determining unit is used to determine each signal-to-noise ratio probability based on the Hessian matrix, each noise covariance matrix, and the total noise covariance matrix.

[0202] The sensing degradation detection device for the sensing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0203] In one embodiment, the second determining unit is specifically used for:

[0204] For each pose direction, the mean noise value in the pose direction is determined based on the total noise covariance matrix.

[0205] Based on the noise covariance matrices and the contribution vectors of each position point to the Hessian matrix, the noise variance in the pose direction is determined.

[0206] Projecting the Hessian matrix onto the pose direction yields the directional observations in the pose direction;

[0207] Based on the noise mean, noise variance, and direction observations, the signal-to-noise ratio probability in the pose direction is determined to determine the probability of each signal-to-noise ratio.

[0208] The sensing degradation detection device for the sensing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0209] In one embodiment, the second determining unit is specifically used for:

[0210] For each pose direction, a normal distribution model of the noise components in the pose direction is established based on the noise mean and noise variance.

[0211] Based on the preset signal-to-noise ratio coefficient and direction observation values, construct the signal-to-noise ratio conditions;

[0212] The probability of satisfying the signal-to-noise ratio condition is defined as the signal-to-noise ratio probability in the pose direction.

[0213] The sensing degradation detection device for the sensing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0214] In one embodiment, the second determining unit is specifically used for:

[0215] The noise components are standardized to obtain the standardized noise components.

[0216] Based on the signal-to-noise ratio condition, determine the critical value of the normalized noise component;

[0217] The critical value is substituted into the cumulative distribution function of the preset standard normal distribution, and the function value of the cumulative distribution function is used as the signal-to-noise ratio probability.

[0218] The sensing degradation detection device for the sensing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0219] In one embodiment, the detection module 803 includes:

[0220] The detection unit is used to detect whether the signal-to-noise ratio probability is greater than a preset probability threshold for each signal-to-noise ratio probability.

[0221] The third determining unit is used to determine that if the signal-to-noise ratio probability is greater than the probability threshold, the detection result is that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0222] The fourth determining unit is used to determine that if the signal-to-noise ratio probability is less than or equal to the probability threshold, the detection result is that there is no degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0223] The sensing degradation detection device for the sensing device provided in this embodiment can execute the above method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0224] Each module in the aforementioned sensing degradation detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the electronic device in hardware form or independently of it, or stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of each module.

[0225] In one exemplary embodiment, an electronic device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, this electronic device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores point cloud data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for detecting sensor degradation in a sensing device.

[0226] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the electronic device to which the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0227] In one embodiment, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0228] Acquire point cloud data of the current frame image collected by the sensing device;

[0229] Based on point cloud data and pose information of the sensing device, the signal-to-noise ratio (SNR) probability is determined in multiple pose directions of the sensing device. The SNR probability is used to characterize the significance of the signal relative to noise in each pose direction.

[0230] Based on the signal-to-noise ratio probabilities, degradation detection is performed on each pose direction of the sensing device to obtain the detection results. The detection results are used to determine the pose adjustment range in each pose direction.

[0231] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0232] Based on each point cloud data, the corresponding normal vector and pose information, the contribution vector of each position point included in the point cloud data is constructed, and the Hessian matrix corresponding to the point cloud data is obtained by summing the contribution vectors.

[0233] Perturbation analysis is performed on each location point to obtain the perturbation analysis results. The noise covariance matrix corresponding to each location point is calculated based on the perturbation analysis results, and the sum of each noise covariance matrix is ​​determined as the total noise covariance matrix.

[0234] Based on the Hessian matrix, each noise covariance matrix, and the total noise covariance matrix, determine each signal-to-noise ratio probability.

[0235] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0236] For each pose direction, the mean noise value in the pose direction is determined based on the total noise covariance matrix.

[0237] Based on the noise covariance matrices and the contribution vectors of each position point to the Hessian matrix, the noise variance in the pose direction is determined.

[0238] Projecting the Hessian matrix onto the pose direction yields the directional observations in the pose direction;

[0239] Based on the noise mean, noise variance, and direction observations, the signal-to-noise ratio probability in the pose direction is determined to determine the probability of each signal-to-noise ratio.

[0240] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0241] For each pose direction, a normal distribution model of the noise components in the pose direction is established based on the noise mean and noise variance.

[0242] Based on the preset signal-to-noise ratio coefficient and direction observation values, construct the signal-to-noise ratio conditions;

[0243] The probability of satisfying the signal-to-noise ratio condition is defined as the signal-to-noise ratio probability in the pose direction.

[0244] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0245] The noise components are standardized to obtain the standardized noise components.

[0246] Based on the signal-to-noise ratio condition, determine the critical value of the normalized noise component;

[0247] The critical value is substituted into the cumulative distribution function of the preset standard normal distribution, and the function value of the cumulative distribution function is used as the signal-to-noise ratio probability.

[0248] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0249] For each signal-to-noise ratio probability, check whether the signal-to-noise ratio probability is greater than a preset probability threshold;

[0250] If the signal-to-noise ratio probability is greater than the probability threshold, then the detection result is determined to be that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0251] If the signal-to-noise ratio probability is less than or equal to the probability threshold, then the detection result is determined to be that there is no degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0252] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0253] Acquire point cloud data of the current frame image collected by the sensing device;

[0254] Based on point cloud data and pose information of the sensing device, the signal-to-noise ratio (SNR) probability is determined in multiple pose directions of the sensing device. The SNR probability is used to characterize the significance of the signal relative to noise in each pose direction.

[0255] Based on the signal-to-noise ratio probabilities, degradation detection is performed on each pose direction of the sensing device to obtain the detection results. The detection results are used to determine the pose adjustment range in each pose direction.

[0256] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0257] Based on each point cloud data, the corresponding normal vector and pose information, the contribution vector of each position point included in the point cloud data is constructed, and the Hessian matrix corresponding to the point cloud data is obtained by summing the contribution vectors.

[0258] Perturbation analysis is performed on each location point to obtain the perturbation analysis results. The noise covariance matrix corresponding to each location point is calculated based on the perturbation analysis results, and the sum of each noise covariance matrix is ​​determined as the total noise covariance matrix.

[0259] Based on the Hessian matrix, each noise covariance matrix, and the total noise covariance matrix, determine each signal-to-noise ratio probability.

[0260] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0261] For each pose direction, the mean noise value in the pose direction is determined based on the total noise covariance matrix.

[0262] Based on the noise covariance matrices and the contribution vectors of each position point to the Hessian matrix, the noise variance in the pose direction is determined.

[0263] Projecting the Hessian matrix onto the pose direction yields the directional observations in the pose direction;

[0264] Based on the noise mean, noise variance, and direction observations, the signal-to-noise ratio probability in the pose direction is determined to determine the probability of each signal-to-noise ratio.

[0265] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0266] For each pose direction, a normal distribution model of the noise components in the pose direction is established based on the noise mean and noise variance.

[0267] Based on the preset signal-to-noise ratio coefficient and direction observation values, construct the signal-to-noise ratio conditions;

[0268] The probability of satisfying the signal-to-noise ratio condition is defined as the signal-to-noise ratio probability in the pose direction.

[0269] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0270] The noise components are standardized to obtain the standardized noise components.

[0271] Based on the signal-to-noise ratio condition, determine the critical value of the normalized noise component;

[0272] The critical value is substituted into the cumulative distribution function of the preset standard normal distribution, and the function value of the cumulative distribution function is used as the signal-to-noise ratio probability.

[0273] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0274] For each signal-to-noise ratio probability, check whether the signal-to-noise ratio probability is greater than a preset probability threshold;

[0275] If the signal-to-noise ratio probability is greater than the probability threshold, then the detection result is determined to be that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0276] If the signal-to-noise ratio probability is less than or equal to the probability threshold, then the detection result is determined to be that there is no degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0277] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0278] Acquire point cloud data of the current frame image collected by the sensing device;

[0279] Based on point cloud data and pose information of the sensing device, the signal-to-noise ratio (SNR) probability is determined in multiple pose directions of the sensing device. The SNR probability is used to characterize the significance of the signal relative to noise in each pose direction.

[0280] Based on the signal-to-noise ratio probabilities, degradation detection is performed on each pose direction of the sensing device to obtain the detection results. The detection results are used to determine the pose adjustment range in each pose direction.

[0281] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0282] Based on each point cloud data, the corresponding normal vector and pose information, the contribution vector of each position point included in the point cloud data is constructed, and the Hessian matrix corresponding to the point cloud data is obtained by summing the contribution vectors.

[0283] Perturbation analysis is performed on each location point to obtain the perturbation analysis results. The noise covariance matrix corresponding to each location point is calculated based on the perturbation analysis results, and the sum of each noise covariance matrix is ​​determined as the total noise covariance matrix.

[0284] Based on the Hessian matrix, each noise covariance matrix, and the total noise covariance matrix, determine each signal-to-noise ratio probability.

[0285] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0286] For each pose direction, the mean noise value in the pose direction is determined based on the total noise covariance matrix.

[0287] Based on the noise covariance matrices and the contribution vectors of each position point to the Hessian matrix, the noise variance in the pose direction is determined.

[0288] Projecting the Hessian matrix onto the pose direction yields the directional observations in the pose direction;

[0289] Based on the noise mean, noise variance, and direction observations, the signal-to-noise ratio probability in the pose direction is determined to determine the probability of each signal-to-noise ratio.

[0290] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0291] For each pose direction, a normal distribution model of the noise components in the pose direction is established based on the noise mean and noise variance.

[0292] Based on the preset signal-to-noise ratio coefficient and direction observation values, construct the signal-to-noise ratio conditions;

[0293] The probability of satisfying the signal-to-noise ratio condition is defined as the signal-to-noise ratio probability in the pose direction.

[0294] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0295] The noise components are standardized to obtain the standardized noise components.

[0296] Based on the signal-to-noise ratio condition, determine the critical value of the normalized noise component;

[0297] The critical value is substituted into the cumulative distribution function of the preset standard normal distribution, and the function value of the cumulative distribution function is used as the signal-to-noise ratio probability.

[0298] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0299] For each signal-to-noise ratio probability, check whether the signal-to-noise ratio probability is greater than a preset probability threshold;

[0300] If the signal-to-noise ratio probability is greater than the probability threshold, then the detection result is determined to be that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0301] If the signal-to-noise ratio probability is less than or equal to the probability threshold, then the detection result is determined to be that there is no degradation in the pose direction corresponding to the signal-to-noise ratio probability.

[0302] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0303] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0304] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for detecting sensor degradation in a sensing device, characterized in that, The method includes: Acquire point cloud data of the current frame image collected by the sensing device; Based on the point cloud data and the pose information of the sensing device, the signal-to-noise ratio probability in multiple pose directions of the sensing device is determined. The signal-to-noise ratio probability is used to characterize the significance of the signal relative to the noise in each pose direction. Based on the signal-to-noise ratio probabilities, degradation detection is performed on each pose direction of the sensing device to obtain detection results, which are used to determine the pose adjustment range in each pose direction.

2. The method according to claim 1, characterized in that, The step of determining the signal-to-noise ratio probability in multiple pose directions of the sensing device based on the point cloud data and the pose information of the sensing device includes: Based on the point cloud data, the normal vector corresponding to each point cloud data, and the pose information, a contribution vector of each position point included in the point cloud data is constructed, and the Hessian matrix corresponding to the point cloud data is obtained by summing the contribution vectors. Perturbation analysis is performed on each of the aforementioned locations to obtain perturbation analysis results. Based on the perturbation analysis results, the noise covariance matrix corresponding to each of the aforementioned locations is calculated, and the sum of the noise covariance matrices is determined as the total noise covariance matrix. The signal-to-noise ratio probabilities are determined based on the Hessian matrix, each of the noise covariance matrices, and the total noise covariance matrix.

3. The method according to claim 2, characterized in that, The step of determining each signal-to-noise ratio probability based on the Hessian matrix, each of the noise covariance matrices, and the total noise covariance matrix includes: For each of the pose directions, the mean noise value in the pose direction is determined based on the total noise covariance matrix; Based on the noise covariance matrices and the contribution vectors of each position point to the Hessian matrix, the noise variance in the pose direction is determined. Projecting the Hessian matrix onto the pose direction yields the directional observation value in the pose direction; Based on the noise mean, the noise variance, and the direction observation, the signal-to-noise ratio probability in the pose direction is determined to determine each of the signal-to-noise ratio probabilities.

4. The method according to claim 3, characterized in that, The step of determining the signal-to-noise ratio probability in the pose direction based on the noise mean, the noise variance, and the direction observation includes: For each pose direction, a normal distribution model of the noise components in the pose direction is established based on the noise mean and the noise variance. Based on the preset signal-to-noise ratio coefficient and the observed direction value, construct the signal-to-noise ratio condition; The probability of satisfying the signal-to-noise ratio condition is determined as the signal-to-noise ratio probability in the pose direction.

5. The method according to claim 4, characterized in that, Determining the probability of satisfying the signal-to-noise ratio condition as the signal-to-noise ratio probability in the pose direction includes: The noise components are standardized to obtain standardized noise components. Based on the signal-to-noise ratio condition, determine the critical value of the normalized noise component; The critical value is substituted into the cumulative distribution function of the preset standard normal distribution, and the function value of the cumulative distribution function is used as the signal-to-noise ratio probability.

6. The method according to any one of claims 1-5, characterized in that, The step of performing degradation detection on each pose direction of the sensing device based on each signal-to-noise ratio probability to obtain a detection result includes: For each of the signal-to-noise ratio probabilities, detect whether the signal-to-noise ratio probability is greater than a preset probability threshold; If the signal-to-noise ratio probability is greater than the probability threshold, then the detection result is determined to be that there is degradation in the pose direction corresponding to the signal-to-noise ratio probability; If the signal-to-noise ratio probability is less than or equal to the probability threshold, then the detection result is determined to be that there is no degradation in the pose direction corresponding to the signal-to-noise ratio probability.

7. A sensing degradation detection device for a sensing device, characterized in that, The device includes: The acquisition module is used to acquire point cloud data of the current frame image collected by the sensing device; The determination module is used to determine the signal-to-noise ratio probability in multiple pose directions of the sensing device based on the point cloud data and the pose information of the sensing device. The signal-to-noise ratio probability is used to characterize the significance of the signal relative to the noise in each pose direction. The detection module is used to perform degradation detection on each pose direction of the sensing device according to each signal-to-noise ratio probability, and obtain the detection result. The detection result is used to determine the pose adjustment range in each pose direction.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images