Robot sensor degradation analysis method and device based on spinor theory

Abstract

The invention relates to the technical field of robot state analysis, and discloses a robot sensor degradation analysis method and device based on the spinor theory, and the method comprises the steps: constructing an observation feature measurement model corresponding to a target sensor according to an equivalent mechanism model corresponding to the target sensor; constructing a feature constraint model corresponding to the target sensor according to the observation feature measurement model, and constructing a degradation analysis model corresponding to the robot according to the feature constraint model; performing numerical analysis on the degradation analysis model to obtain candidate motion primitives of the robot; and determining a robust discrimination function corresponding to the candidate motion primitive, and performing degradation detection on the candidate motion primitive through the robust discrimination function to discriminate whether an application scene where the robot is located degrades towards the candidate motion primitive, so that the constraint effect of the sensor in state estimation is analyzed by using a spinor theory, and the state estimation accuracy is improved. The influence of scene noise and environment scale is reduced, and the degradation detection accuracy of the robot sensor is improved.

Description

Technical Field

[0001] This invention relates to the field of robot state analysis technology, and in particular to a method and apparatus for robot sensor degradation analysis based on spinor theory. Background Technology

[0002] In the field of robot control, state estimation, as a core technology, is widely used in scenarios such as Simultaneous Localization and Mapping (SLAM), sensor fusion, and calibration. Its accuracy has significantly improved in common environments. However, in degenerate scenarios with similar geometric structures, such as long corridors, circular tunnels, and spiral corridors, it still faces significant challenges. These environments, lacking features due to natural or human factors, exhibit high self-similarity, making it difficult to effectively constrain specific robot movements. This leads to a surge in uncertainty in state estimation, and even severe drift. Therefore, while observational data from external sensors can reflect the local state relationship between the robot's state and environmental features, geometric shape and texture information in unstructured outdoor environments can be misleading. For example, self-symmetric or axisymmetric structures can lead to incomplete spatial constraints, or repetitive or missing textures can cause the state estimator to get trapped in local optima, making the optimization problem sensitive to initial values ​​and rapidly accumulating errors in the degenerate space. Furthermore, in existing degradation detection methods, the degradation discrimination function value easily varies significantly with environmental scale, sensor noise, and other factors, making it difficult to determine the degradation threshold, resulting in low degradation detection accuracy.

[0003] In summary, there is an urgent need for a technical solution that can accurately address the effects of noise and environmental scale in degradation detection scenarios, thereby improving the accuracy of degradation detection. Summary of the Invention

[0004] This invention provides a robot sensor degradation analysis method and apparatus based on screw theory. By using screw theory to analyze the constraint effect of the sensor in state estimation, the influence of scene noise and environmental scale is reduced, which is beneficial to improving the accuracy of robot sensor degradation detection.

[0005] To address the aforementioned technical problems, the first aspect of this invention discloses a robot sensor degradation analysis method based on spinor theory, the method comprising:

[0006] Determine the equivalent mechanism model corresponding to the target sensor of the robot, and construct the observation feature measurement model corresponding to the target sensor based on the equivalent mechanism model corresponding to the target sensor;

[0007] Based on the observation feature measurement model corresponding to the target sensor, a feature constraint model corresponding to the target sensor is constructed, and based on the feature constraint model corresponding to the target sensor, a degradation analysis model corresponding to the robot is constructed.

[0008] Numerical analysis is performed on the degradation analysis model corresponding to the robot to obtain candidate motion primitives of the robot; the candidate motion primitives include candidate translational degenerate motion primitives and candidate spiral degenerate motion primitives;

[0009] A robust discriminant function is determined for the candidate motion primitive, and degradation detection is performed on the candidate motion primitive using the robust discriminant function to obtain the degradation detection result of the candidate motion primitive; the degradation detection result of the candidate motion primitive is used to determine whether the application scenario in which the robot is located is degrading towards the candidate motion primitive.

[0010] As an optional implementation, in the first aspect of the present invention, the target sensor includes a lidar sensor, a distance measurement sensor, and an orientation measurement sensor. The observation feature measurement model corresponding to the target sensor includes a point-area feature measurement model and a point-edge feature measurement model corresponding to the lidar sensor, a distance feature measurement model corresponding to the distance measurement sensor, and an orientation feature measurement model corresponding to the orientation measurement sensor.

[0011] The step of constructing a feature constraint model corresponding to the target sensor based on the observation feature measurement model corresponding to the target sensor includes:

[0012] Based on the point-to-surface feature measurement model corresponding to the lidar sensor, the first degenerate space spiral system corresponding to the point-to-surface feature measurement model is determined, and the first degenerate space spiral system is subjected to anti-spiral analysis to obtain the planar feature constraint spiral system corresponding to the lidar sensor, which is used as the first feature constraint model corresponding to the lidar sensor.

[0013] Based on the point-edge feature measurement model corresponding to the lidar sensor, the second degenerate space spiral system corresponding to the point-edge feature measurement model is determined, and the second degenerate space spiral system is subjected to anti-spiral analysis to obtain the linear feature constraint spiral system corresponding to the lidar sensor, which is used as the second feature constraint model corresponding to the lidar sensor.

[0014] Based on the distance feature measurement model corresponding to the distance measurement sensor, the third degenerate space spiral system corresponding to the distance feature measurement model is determined, and the third degenerate space spiral system is subjected to anti-spiral analysis to obtain the distance measurement constraint spiral system corresponding to the distance measurement sensor, which serves as the third feature constraint model corresponding to the distance measurement sensor.

[0015] Based on the orientation feature measurement model corresponding to the orientation measurement sensor, the fourth degenerate space spiral system corresponding to the orientation feature measurement model is determined, and the fourth degenerate space spiral system is subjected to anti-spiral analysis to obtain the orientation measurement constraint spiral system corresponding to the orientation measurement sensor, which serves as the fourth feature constraint model corresponding to the orientation measurement sensor.

[0016] As an optional implementation, in the first aspect of the present invention, the step of constructing a degradation analysis model for the robot based on the feature constraint model corresponding to the target sensor includes:

[0017] Based on the feature constraint model corresponding to the target sensor, construct the system constraint spiral system of the robot;

[0018] Based on the preset translational motion primitives of the robot and the system constraint spiral system, the translational degenerate motion space of the robot is constructed, and based on the preset spiral motion primitives of the robot and the system constraint spiral system, the spiral degenerate motion space of the robot is constructed.

[0019] Based on the translational degenerate motion space, the translational motion primitive constraint strength function of the robot is determined, and based on the helical degenerate motion space, the helical motion primitive constraint strength function of the robot is determined.

[0020] The constraint strength functions of the translational motion primitives and the constraint strength functions of the helical motion primitives are determined as the degradation analysis model corresponding to the robot.

[0021] As an optional implementation, in the first aspect of the present invention, the translational motion primitive constraint strength function of the robot is:

[0022]

[0023] Where M1 is the number of translational environmental features of the robot obtained in advance. r S j Let be the principal vector of the j-th environmental constraint screw in the system constraint screw system. t S is the preset translation direction vector;

[0024] The constraint strength function of the helical motion primitive of the robot is:

[0025]

[0026] Where M2 is the number of spiral environment features of the robot obtained in advance. Let be the follower vector of the j-th environmental constraint screw in the system constraint screw system. θS is the preset rotation axis direction vector, r ξ is the preset target radius parameter, and h is the preset target pitch parameter.

[0027] As an optional implementation, in the first aspect of the present invention, the step of performing numerical analysis on the degradation analysis model corresponding to the robot to obtain candidate motion primitives of the robot includes:

[0028] For the translational motion primitive constraint strength function, the translational motion primitive constraint strength function is reconstructed to obtain the reconstructed translational motion primitive constraint strength function;

[0029] Singular value decomposition is performed on the translational degradation model information matrix in the reconstructed translational motion primitive constraint strength function to obtain multiple unit orthogonal eigenvectors corresponding to the translational motion primitive constraint strength function; each unit orthogonal eigenvector has a corresponding eigenvalue.

[0030] Based on the eigenvalues ​​corresponding to all the unit orthogonal eigenvectors, a target eigenvector whose eigenvalue is less than or equal to a preset eigenvalue threshold is determined from all the unit orthogonal eigenvectors;

[0031] Based on the target feature vector, candidate translational degenerate motion primitives of the robot are constructed; and,

[0032] For the constraint strength function of the spiral motion primitive, the constrained minimum value optimization function corresponding to the constraint strength function of the spiral motion primitive is determined, and the singular value decomposition operation is performed on the constrained minimum value optimization function to obtain the unconstrained matrix minimum eigenvalue solution function.

[0033] The adaptive damped Newton algorithm is used to solve the minimum eigenvalue solution function of the unconstrained matrix to obtain the target pitch value and target radius value corresponding to the constraint strength function of the helical motion primitive. Based on the target pitch value and the target radius value, the candidate helical degenerate motion primitive of the robot is constructed.

[0034] As an optional implementation, in the first aspect of the present invention, determining the robust discriminant function corresponding to the candidate motion primitive and performing degradation detection on the candidate motion primitive using the robust discriminant function to obtain the degradation detection result of the candidate motion primitive includes:

[0035] For the candidate translational degenerate motion primitive, based on the translational motion primitive of the robot and the system constraint screw system, the environmental constraint strength function of the translational motion primitive of the robot is determined, and the first maximum strength value of the environmental constraint strength function of the translational motion primitive is determined.

[0036] Based on the environmental constraint strength function of the translational motion primitive, the first maximum strength value, and the preset first feature vector angle error robustness threshold, the first robust discrimination function corresponding to the candidate translational degenerate motion primitive is determined;

[0037] Based on the first robust discriminant function and the pre-acquired parameters of the robot's translational environment features, a robust discriminant function for translational degradation corresponding to the candidate translational degradation motion primitive is constructed. Then, based on the robust discriminant function and a preset translational degradation ratio threshold, degradation detection is performed on the candidate translational degradation motion primitive to obtain the degradation detection result of the candidate translational degradation motion primitive; and...

[0038] For the candidate helical degenerate motion primitive, based on the helical motion primitive of the robot and the system constraint helical system, the environmental constraint strength function of the helical motion primitive of the robot is determined, and the second maximum strength value of the environmental constraint strength function of the helical motion primitive is determined.

[0039] Based on the environmental constraint strength function of the spiral motion primitive, the second maximum strength value, and the preset second feature vector angle error robust threshold, the second robust discrimination function corresponding to the candidate spiral degenerate motion primitive is determined;

[0040] Based on the second robust discriminant function, the position of the spiral environment feature points of the robot, the pre-acquired spiral environment feature quantity parameters of the robot, and the target spiral feature parameters, a spiral degradation robust discriminant function corresponding to the candidate spiral degradation motion primitive is constructed; the target spiral feature value includes the target pitch parameter and the target radius parameter of the spiral degradation model information matrix in the constraint strength function of the spiral motion primitive;

[0041] Based on the spiral degradation robust discrimination function and the preset spiral degradation ratio threshold, degradation detection is performed on the candidate spiral degradation motion primitives to obtain the degradation detection results of the candidate spiral degradation motion primitives.

[0042] As an optional implementation, in the first aspect of the present invention, the method further includes:

[0043] When the degradation detection result of the candidate translational degenerate motion primitive meets a preset first condition, the degradation detection result of the candidate translational degenerate motion primitive is determined to indicate that the application scenario in which the robot is located is degenerating towards the candidate translational degenerate motion primitive.

[0044] When the degradation detection result of the candidate spiral degenerate motion primitive meets the preset second condition, it is determined that the degradation detection result of the candidate spiral degenerate motion primitive is used to indicate that the application scenario in which the robot is located is degenerating towards the candidate spiral degenerate motion primitive;

[0045] The first condition is:

[0046]

[0047] Let tξ be the candidate translational degenerate motion primitive, and μ be the translational motion primitive of the robot. t The translation degradation ratio threshold is given, where, t c k ( t ξ) is the translational degradation robustness discriminant function and M1 is the quantity parameter of the translational environment features. t e j The translational motion primitive is subject to environmental constraint intensity function and r ξ j Let σ1 be the j-th environmental constraint spiral in the system constraint spiral system, σ1 be the robust threshold of the first eigenvector angle error, and ρ be the environmental constraint spiral. n ( t e j ,σ1) is the first robust discriminant function and U1 is the first maximum strength value;

[0048] The second condition is:

[0049]

[0050] The candidate spiral degenerate motion primitive is... θ ξ is the translational motion element of the robot, μ θ The spiral degradation ratio threshold is given, where, θ c k ( θ ξ) is the robust discriminant function for spiral degeneration and M2 is the quantity parameter of the spiral environment characteristics. θ e j The helical motion primitive is subject to environmental constraint intensity function and σ² is the robust threshold for the angle error of the second eigenvector, ρ n ( θ e j ,σ2) is the second robust discriminant function and U2 is the second maximum intensity value, r j Let r be the location of the spiral environmental feature point of the robot. ξ Let h be the target radius parameter and h be the target pitch parameter.

[0051] A second aspect of this invention discloses a robot sensor degradation analysis device based on spinor theory, the device comprising:

[0052] The determination module is used to determine the equivalent mechanism model corresponding to the target sensor of the robot;

[0053] The construction module is used to construct an observation feature measurement model corresponding to the target sensor based on the equivalent mechanism model corresponding to the target sensor; construct a feature constraint model corresponding to the target sensor based on the observation feature measurement model corresponding to the target sensor; and construct a degradation analysis model corresponding to the robot based on the feature constraint model corresponding to the target sensor.

[0054] The analysis module is used to perform numerical analysis on the degradation analysis model corresponding to the robot to obtain candidate motion primitives of the robot; the candidate motion primitives include candidate translational degradation motion primitives and candidate spiral degradation motion primitives;

[0055] The detection module is used to determine the robust discriminant function corresponding to the candidate motion primitive, and to perform degradation detection on the candidate motion primitive through the robust discriminant function to obtain the degradation detection result of the candidate motion primitive; the degradation detection result of the candidate motion primitive is used to determine whether the application scenario in which the robot is located is degrading towards the candidate motion primitive.

[0056] As an optional implementation, in the second aspect of the present invention, the target sensor includes a lidar sensor, a distance measurement sensor, and an orientation measurement sensor, and the observation feature measurement model corresponding to the target sensor includes a point-area feature measurement model and a point-edge feature measurement model corresponding to the lidar sensor, a distance feature measurement model corresponding to the distance measurement sensor, and an orientation feature measurement model corresponding to the orientation measurement sensor.

[0057] Specifically, the method by which the construction module constructs the feature constraint model corresponding to the target sensor based on the observation feature measurement model corresponding to the target sensor includes:

[0058] Based on the point-to-surface feature measurement model corresponding to the lidar sensor, the first degenerate space spiral system corresponding to the point-to-surface feature measurement model is determined, and the first degenerate space spiral system is subjected to anti-spiral analysis to obtain the planar feature constraint spiral system corresponding to the lidar sensor, which is used as the first feature constraint model corresponding to the lidar sensor.

[0059] Based on the point-edge feature measurement model corresponding to the lidar sensor, the second degenerate space spiral system corresponding to the point-edge feature measurement model is determined, and the second degenerate space spiral system is subjected to anti-spiral analysis to obtain the linear feature constraint spiral system corresponding to the lidar sensor, which is used as the second feature constraint model corresponding to the lidar sensor.

[0060] Based on the distance feature measurement model corresponding to the distance measurement sensor, the third degenerate space spiral system corresponding to the distance feature measurement model is determined, and the third degenerate space spiral system is subjected to anti-spiral analysis to obtain the distance measurement constraint spiral system corresponding to the distance measurement sensor, which serves as the third feature constraint model corresponding to the distance measurement sensor.

[0061] Based on the orientation feature measurement model corresponding to the orientation measurement sensor, the fourth degenerate space spiral system corresponding to the orientation feature measurement model is determined, and the fourth degenerate space spiral system is subjected to anti-spiral analysis to obtain the orientation measurement constraint spiral system corresponding to the orientation measurement sensor, which serves as the fourth feature constraint model corresponding to the orientation measurement sensor.

[0062] As an optional implementation, in the second aspect of the present invention, the method by which the construction module constructs the degradation analysis model corresponding to the robot based on the feature constraint model corresponding to the target sensor specifically includes:

[0063] Based on the feature constraint model corresponding to the target sensor, construct the system constraint spiral system of the robot;

[0064] Based on the preset translational motion primitives of the robot and the system constraint spiral system, the translational degenerate motion space of the robot is constructed, and based on the preset spiral motion primitives of the robot and the system constraint spiral system, the spiral degenerate motion space of the robot is constructed.

[0065] Based on the translational degenerate motion space, the translational motion primitive constraint strength function of the robot is determined, and based on the helical degenerate motion space, the helical motion primitive constraint strength function of the robot is determined.

[0066] The constraint strength functions of the translational motion primitives and the constraint strength functions of the helical motion primitives are determined as the degradation analysis model corresponding to the robot.

[0067] As an optional implementation, in a second aspect of the invention, the translational motion primitive constraint strength function of the robot is:

[0068]

[0069] Where M1 is the number of translational environmental features of the robot obtained in advance. r S j Let be the principal vector of the j-th environmental constraint screw in the system constraint screw system. t S is the preset translation direction vector;

[0070] The constraint strength function of the helical motion primitive of the robot is:

[0071]

[0072] Where M2 is the number of spiral environment features of the robot obtained in advance. Let be the follower vector of the j-th environmental constraint screw in the system constraint screw system. θ S is the preset rotation axis direction vector, r ξ is the preset target radius parameter, and h is the preset target pitch parameter.

[0073] As an optional implementation, in the second aspect of the present invention, the analysis module performs numerical analysis on the degradation analysis model corresponding to the robot to obtain candidate motion primitives of the robot in the following specific ways:

[0074] For the translational motion primitive constraint strength function, the translational motion primitive constraint strength function is reconstructed to obtain the reconstructed translational motion primitive constraint strength function;

[0075] Singular value decomposition is performed on the translational degradation model information matrix in the reconstructed translational motion primitive constraint strength function to obtain multiple unit orthogonal eigenvectors corresponding to the translational motion primitive constraint strength function; each unit orthogonal eigenvector has a corresponding eigenvalue.

[0076] Based on the eigenvalues ​​corresponding to all the unit orthogonal eigenvectors, a target eigenvector whose eigenvalue is less than or equal to a preset eigenvalue threshold is determined from all the unit orthogonal eigenvectors;

[0077] Based on the target feature vector, candidate translational degenerate motion primitives of the robot are constructed; and,

[0078] For the constraint strength function of the spiral motion primitive, the constrained minimum value optimization function corresponding to the constraint strength function of the spiral motion primitive is determined, and the singular value decomposition operation is performed on the constrained minimum value optimization function to obtain the unconstrained matrix minimum eigenvalue solution function.

[0079] The adaptive damped Newton algorithm is used to solve the minimum eigenvalue solution function of the unconstrained matrix to obtain the target pitch value and target radius value corresponding to the constraint strength function of the helical motion primitive. Based on the target pitch value and the target radius value, the candidate helical degenerate motion primitive of the robot is constructed.

[0080] As an optional implementation, in a second aspect of the present invention, the detection module determines the robust discriminant function corresponding to the candidate motion primitive, and performs degradation detection on the candidate motion primitive using the robust discriminant function to obtain the degradation detection result of the candidate motion primitive. Specifically, this includes:

[0081] For the candidate translational degenerate motion primitive, based on the translational motion primitive of the robot and the system constraint screw system, the environmental constraint strength function of the translational motion primitive of the robot is determined, and the first maximum strength value of the environmental constraint strength function of the translational motion primitive is determined.

[0082] Based on the environmental constraint strength function of the translational motion primitive, the first maximum strength value, and the preset first feature vector angle error robustness threshold, the first robust discrimination function corresponding to the candidate translational degenerate motion primitive is determined;

[0083] Based on the first robust discriminant function and the pre-acquired parameters of the robot's translational environment features, a robust discriminant function for translational degradation corresponding to the candidate translational degradation motion primitive is constructed. Then, based on the robust discriminant function and a preset translational degradation ratio threshold, degradation detection is performed on the candidate translational degradation motion primitive to obtain the degradation detection result of the candidate translational degradation motion primitive; and...

[0084] For the candidate helical degenerate motion primitive, based on the helical motion primitive of the robot and the system constraint helical system, the environmental constraint strength function of the helical motion primitive of the robot is determined, and the second maximum strength value of the environmental constraint strength function of the helical motion primitive is determined.

[0085] Based on the environmental constraint strength function of the spiral motion primitive, the second maximum strength value, and the preset second feature vector angle error robust threshold, the second robust discrimination function corresponding to the candidate spiral degenerate motion primitive is determined;

[0086] Based on the second robust discriminant function, the position of the spiral environment feature points of the robot, the pre-acquired spiral environment feature quantity parameters of the robot, and the target spiral feature parameters, a spiral degradation robust discriminant function corresponding to the candidate spiral degradation motion primitive is constructed; the target spiral feature value includes the target pitch parameter and the target radius parameter of the spiral degradation model information matrix in the constraint strength function of the spiral motion primitive;

[0087] Based on the spiral degradation robust discrimination function and the preset spiral degradation ratio threshold, degradation detection is performed on the candidate spiral degradation motion primitives to obtain the degradation detection results of the candidate spiral degradation motion primitives.

[0088] As an optional implementation, in a second aspect of the invention, the determining module is further configured to:

[0089] When the degradation detection result of the candidate translational degenerate motion primitive meets a preset first condition, the degradation detection result of the candidate translational degenerate motion primitive is determined to indicate that the application scenario in which the robot is located is degenerating towards the candidate translational degenerate motion primitive.

[0090] When the degradation detection result of the candidate spiral degenerate motion primitive meets the preset second condition, it is determined that the degradation detection result of the candidate spiral degenerate motion primitive is used to indicate that the application scenario in which the robot is located is degenerating towards the candidate spiral degenerate motion primitive;

[0091] The first condition is:

[0092]

[0093] Let tξ be the candidate translational degenerate motion primitive, and μ be the translational motion primitive of the robot. t The translation degradation ratio threshold is given, where, t c k ( t ξ) is the translational degradation robustness discriminant function and M1 is the quantity parameter of the translational environment features. t e j The translational motion primitive is subject to environmental constraint intensity function and r ξ j Let σ1 be the j-th environmental constraint spiral in the system constraint spiral system, σ1 be the robust threshold of the first eigenvector angle error, and ρ be the environmental constraint spiral. n ( t e j ,σ1) is the first robust discriminant function and U1 is the first maximum strength value;

[0094] The second condition is:

[0095]

[0096] The candidate spiral degenerate motion primitive is... θ ξ is the translational motion element of the robot, μ θ The spiral degradation ratio threshold is given, where, θ c k ( θ ξ) is the robust discriminant function for spiral degeneration and M2 is the quantity parameter of the spiral environment characteristics. θ e j The helical motion primitive is subject to environmental constraint intensity function and σ² is the robust threshold for the angle error of the second eigenvector, ρ n ( θ e j ,σ2) is the second robust discriminant function and U2 is the second maximum intensity value, r j Let r be the location of the spiral environmental feature point of the robot. ξ Let h be the target radius parameter and h be the target pitch parameter.

[0097] A third aspect of this invention discloses another robot sensor degradation analysis device based on spinor theory, the device comprising:

[0098] Memory containing executable program code;

[0099] A processor coupled to the memory;

[0100] The processor calls the executable program code stored in the memory to execute the robot sensor degradation analysis method based on spinor theory disclosed in the first aspect of the present invention.

[0101] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute the robot sensor degradation analysis method based on spinor theory disclosed in the first aspect of the present invention.

[0102] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0103] In this embodiment of the invention, an observation feature measurement model corresponding to the target sensor is constructed based on the equivalent mechanism model corresponding to the target sensor; a feature constraint model corresponding to the target sensor is constructed based on the observation feature measurement model; and a degradation analysis model corresponding to the robot is constructed based on the feature constraint model; numerical analysis is performed on the degradation analysis model to obtain candidate motion primitives of the robot; a robust discriminant function corresponding to the candidate motion primitives is determined, and degradation detection is performed on the candidate motion primitives through the robust discriminant function to determine whether the application scenario in which the robot is located is degrading towards the candidate motion primitives. In this way, the constraint effect of the sensor in state estimation is analyzed using spinor theory, which reduces the influence of scene noise and environmental scale, and is conducive to improving the accuracy of robot sensor degradation detection. Attached Figure Description

[0104] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0105] Figure 1 This is a schematic flowchart of a robot sensor degradation analysis method based on spinor theory disclosed in an embodiment of the present invention;

[0106] Figure 2 This is a flowchart illustrating another robot sensor degradation analysis method based on spinor theory disclosed in an embodiment of the present invention.

[0107] Figure 3 This is a schematic diagram of the structure of a robot sensor degradation analysis device based on spinor theory disclosed in an embodiment of the present invention;

[0108] Figure 4 This is a schematic diagram of another robot sensor degradation analysis device based on spinor theory disclosed in an embodiment of the present invention.

[0109] Figure 5 This is a schematic diagram illustrating the equivalent relationship between sensor observation features and mechanical transmission structure disclosed in an embodiment of the present invention;

[0110] Figure 6 This is a schematic diagram of a scenario for robot sensor degradation analysis disclosed in an embodiment of the present invention;

[0111] Figure 7 This is a schematic diagram of a feature extraction framework based on feature map resampling disclosed in an embodiment of the present invention. Detailed Implementation

[0112] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0113] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0114] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0115] This invention discloses a robot sensor degradation analysis method and apparatus based on screw theory. By using screw theory to analyze the constraint effect of the sensor in state estimation, the influence of scene noise and environmental scale is reduced, which is beneficial to improving the accuracy of robot sensor degradation detection.

[0116] Example 1

[0117] Please see Figure 1 , Figure 1 This is a schematic flowchart of a robot sensor degradation analysis method based on screw theory disclosed in an embodiment of the present invention. Optionally, this method can be implemented by a sensor degradation analysis system, which can be integrated into the robot, or it can be a local server or cloud server used to process the robot sensor degradation analysis process based on screw theory, etc., and the embodiments of the present invention are not limited thereto. Figure 1 As shown, this robot sensor degradation analysis method based on spinor theory can include the following operations:

[0118] 101. Determine the equivalent mechanism model corresponding to the target sensor of the robot, and construct the observation feature measurement model corresponding to the target sensor based on the equivalent mechanism model corresponding to the target sensor.

[0119] In this embodiment of the invention, the target sensor includes a lidar sensor, a distance measurement sensor, and an orientation measurement sensor.

[0120] It should be noted that this invention proposes to introduce spinor analysis into the feature observation modeling process for SLAM state estimation, thereby enabling the modeling of feature constraints for common sensor systems such as lidar sensors, orientation measurement sensors, and distance measurement sensors. This invention has the advantages of requiring no derivative and being independent of the cost function definition.

[0121] Because the unit screw reciprocity product is independent of coordinate system definition, this invention uses unit screws to characterize environmental feature constraints and motion primitives. In screw theory, when two screws are reciprocal, they have a dual relationship: one screw can be considered as motion, and the other is correspondingly considered as a constraint on that motion; conversely, one represents a constraint force, and the other represents the motion allowed under the constraint conditions. If the reciprocity product of a motion screw and a constraint screw is not zero, it indicates that under the current motion state, the "constraint force" generated by the environmental feature constraint will do work on the rigid body system to suppress motion that does not conform to the system's specifications. In summary, the specific sensor observation constraint modeling (i.e., observation feature measurement model) process can be understood as follows:

[0122] First, such as Figure 5 As shown, common sensor observation features are equivalent to their mechanical transmission structures. For example, the point-to-plane and edge features of lidar are respectively equivalent to... Figure 5 The mechanical transmission structure in (a) and (b) is equivalent, and the distance sensor measurement constraint is... Figure 5 (c) The mechanical transmission structure is equivalent, while the direction sensor measurement constraint is the same as... Figure 5 (d) Equivalent mechanical transmission structure.

[0123] Furthermore, the observation feature measurement models corresponding to the target sensors include the point-area feature measurement model and the point-edge feature measurement model corresponding to the lidar sensor, the distance feature measurement model corresponding to the distance measurement sensor, and the orientation feature measurement model corresponding to the direction measurement sensor.

[0124] 102. Based on the observation feature measurement model corresponding to the target sensor, construct the feature constraint model corresponding to the target sensor, and based on the feature constraint model corresponding to the target sensor, construct the degradation analysis model corresponding to the robot.

[0125] In this embodiment of the invention, based on the above equivalent relationship and following the screw modeling method for mechanical transmission systems, equivalent constraint screw systems and motion screw systems of various measurement constraints can be constructed, thereby constructing the corresponding degradation analysis model for the robot, including translational degradation analysis model and screw degradation analysis model.

[0126] 103. Perform numerical analysis on the degradation analysis model corresponding to the robot to obtain the candidate motion primitives of the robot.

[0127] In this embodiment of the invention, the candidate motion primitives include candidate translational degenerate motion primitives and candidate spiral degenerate motion primitives.

[0128] 104. Determine the robust discriminant function corresponding to the candidate motion primitive, and perform degradation detection on the candidate motion primitive through the robust discriminant function to obtain the degradation detection result of the candidate motion primitive.

[0129] In this embodiment of the invention, the degradation detection result of the candidate motion primitive is used to determine whether the application scenario in which the robot is located degrades towards the candidate motion primitive, that is, to determine whether the application scenario in which the robot is located degrades towards the candidate translational motion primitive or the candidate spiral degraded motion primitive.

[0130] As can be seen, implementing the embodiments of the present invention can construct a sensor observation feature measurement and constraint model through screw analysis, and use unit screws to characterize environmental constraints and motion primitives. Through equivalent mechanism modeling, feature constraint derivation, numerical analysis, and robust discrimination, the robot sensor degradation detection process is realized. In this way, it can solve the problems of high complexity and poor adaptability of state estimation in traditional methods without demand derivation or cost function dependence. It also reduces the influence of scene noise and environmental scale and improves modeling efficiency and versatility. At the same time, through the degradation analysis model and robust discrimination function, the translational and spiral degradation trends can be accurately located, which is conducive to improving the accuracy of robot sensor degradation detection. This provides a key basis for robot scene perception and state estimation, thereby helping the system to provide early warning and optimize control, and ensuring the reliability and stability of operation in complex environments.

[0131] Example 2

[0132] Please see Figure 2 , Figure 2 This is a schematic flowchart of another robot sensor degradation analysis method based on screw theory disclosed in an embodiment of the present invention. Optionally, this method can be implemented by a sensor degradation analysis system, which can be integrated into the robot, or it can be a local server or cloud server used to process the robot sensor degradation analysis process based on screw theory, etc., and the embodiments of the present invention are not limited thereto. Figure 2As shown, this robot sensor degradation analysis method based on spinor theory can include the following operations:

[0133] 201. Determine the equivalent mechanism model corresponding to the target sensor of the robot, and construct the observation feature measurement model corresponding to the target sensor based on the equivalent mechanism model corresponding to the target sensor.

[0134] 202. Based on the point-to-surface feature measurement model corresponding to the lidar sensor, determine the first degenerate space spiral system corresponding to the point-to-surface feature measurement model, and perform anti-spiral analysis on the first degenerate space spiral system to obtain the planar feature constraint spiral system corresponding to the lidar sensor, which serves as the first feature constraint model corresponding to the lidar sensor.

[0135] In an embodiment of the present invention, for example, r is uniformly defined. i Let be the position vector of the i-th feature in the sensor coordinate system. For the eigenvector (plane normal vector n) i Or the direction vector d of the line i The corresponding orthogonal complement space contains a set of identity orthogonal bases, which in orientation or distance sensor analysis are characteristic position vectors r. i The corresponding orthogonal complement space unit orthogonal basis.

[0136] Among them, the degenerate space spiral system corresponding to the point-area feature measurement model of the lidar sensor can be: n i Let be the normal vector of the plane feature in the i-th information pair.

[0137] Furthermore, the anti-screw of its degenerate spatial spinor system is obtained, which can represent the constraint spiral system of the planar feature on the lidar sensor coordinate system (i.e., the first feature constraint model):

[0138]

[0139] 203. Based on the point-edge feature measurement model corresponding to the lidar sensor, determine the second degenerate space spiral system corresponding to the point-edge feature measurement model, and perform anti-spiral analysis on the second degenerate space spiral system to obtain the linear feature constraint spiral system corresponding to the lidar sensor, which serves as the second feature constraint model corresponding to the lidar sensor.

[0140] In this embodiment of the invention, similarly, the degenerate space spiral system corresponding to the point-edge feature measurement model of the lidar sensor can be:

[0141] Let be the midline feature direction vector of the i-th information pair.

[0142] Furthermore, the anti-screw of its degenerate space spinor system is obtained, which can represent the constraint spiral system of the linear feature on the coordinate system of the lidar sensor (i.e., the second feature constraint model):

[0143]

[0144] 204. Based on the distance feature measurement model corresponding to the distance measurement sensor, determine the third degenerate space spiral system corresponding to the distance feature measurement model, and perform anti-spiral analysis on the third degenerate space spiral system to obtain the distance measurement constraint spiral system corresponding to the distance measurement sensor, which serves as the third feature constraint model corresponding to the distance measurement sensor.

[0145] In this embodiment of the invention, similarly, the degenerate spatial spiral system corresponding to the distance feature measurement model of the distance measurement sensor can be:

[0146]

[0147] Furthermore, the anti-helix of its degenerate spatial spinor system can be obtained, which represents the distance measurement constraint corresponding to the distance measurement sensor:

[0148]

[0149] 205. Based on the orientation feature measurement model corresponding to the orientation measurement sensor, determine the fourth degenerate space spiral system corresponding to the orientation feature measurement model, and perform anti-spiral analysis on the fourth degenerate space spiral system to obtain the orientation measurement constraint spiral system corresponding to the orientation measurement sensor, which serves as the fourth feature constraint model corresponding to the orientation measurement sensor.

[0150] In this embodiment of the invention, similarly, the degenerate spatial spiral system corresponding to the orientation feature measurement model of the orientation measurement sensor can be:

[0151]

[0152] Furthermore, the anti-helix of its degenerate spatial spinor system can be obtained, which represents the orientation measurement constraint corresponding to the orientation measurement sensor:

[0153]

[0154] 206. Based on the feature constraint model corresponding to the target sensor, construct the degradation analysis model corresponding to the robot.

[0155] 207. Perform numerical analysis on the degradation analysis model corresponding to the robot to obtain the candidate motion primitives of the robot.

[0156] 208. Determine the robust discriminant function corresponding to the candidate motion primitive, and perform degradation detection on the candidate motion primitive through the robust discriminant function to obtain the degradation detection result of the candidate motion primitive.

[0157] In this embodiment of the invention, for other descriptions of steps 201 and 206-208, please refer to the detailed description of steps 101-104 in Embodiment 1, which will not be repeated in this embodiment. Furthermore, steps 202-205 can be performed simultaneously or sequentially. After obtaining the feature constraint models corresponding to all sensors, step 206 can be executed.

[0158] As can be seen, implementing the embodiments of the present invention first constructs equivalent mechanism models and observation feature measurement models for robot target sensors (LiDAR, distance measurement, and orientation measurement sensors), and determines corresponding degradation space spiral systems for different feature models of each sensor. Then, through anti-spiral analysis, the feature constraint models of each sensor are obtained, and a degradation analysis model of the robot is constructed based on this. The degradation detection process of robot sensors is realized through numerical analysis and robust discrimination. In this way, accurate constraint modeling of LiDAR point-to-surface and point-to-edge features, as well as features of distance and orientation sensors, is achieved, improving the reliability and accuracy of feature constraints for each sensor. At the same time, through anti-spiral analysis, the correspondence between degradation space and constraint spiral system is effectively established, making the degradation detection process more consistent with the characteristics of different sensors. This significantly improves the accuracy of the robot's judgment of sensor degradation trends in complex environments, providing a more reliable basis for robot state estimation and motion control.

[0159] In an optional embodiment, step 206 above, which involves constructing a degradation analysis model for the robot based on the feature constraint model corresponding to the target sensor, includes:

[0160] Based on the feature constraint model corresponding to the target sensor, construct the robot's system constraint spiral system;

[0161] Based on the preset translational motion primitives of the robot and the system constraint spiral system, the translational degenerate motion space of the robot is constructed, and based on the preset spiral motion primitives of the robot and the system constraint spiral system, the spiral degenerate motion space of the robot is constructed.

[0162] Based on the translational degenerate motion space, determine the constraint strength function of the robot's translational motion primitives, and based on the helical degenerate motion space, determine the constraint strength function of the robot's helical motion primitives.

[0163] The constraint strength functions of the translational motion primitive and the constraint strength functions of the helical motion primitive are determined as the corresponding degradation analysis models for the robot.

[0164] In this optional embodiment, a translational degradation analysis model (i.e., translational motion primitive constraint strength function) and a spiral degradation analysis model (i.e., spiral motion primitive constraint strength function) are respectively established using the environmental characteristics corresponding to the robot.

[0165] The constraint strength function for the translational motion primitives of the robot is:

[0166]

[0167] Where M1 is the number of translational environmental features of the robot obtained in advance. r S j Let be the principal vector of the j-th environmental constraint screw in the system constraint screw system. t S is the preset translation direction vector;

[0168] The constraint strength function of the robot's helical motion primitive is:

[0169]

[0170] Where M2 is the number of pre-acquired spiral environment features of the robot. Let be the follower vector of the j-th environmental constraint screw in the system constraint screw system. θ S is the preset rotation axis direction vector, r ξ is the preset target radius parameter, and h is the preset target pitch parameter.

[0171] For example, the process of constructing the degradation analysis model for the robot in this embodiment can be understood as follows: First, the observation constraint spirals of each sensor are aggregated to form a system constraint spiral system: Similar to the parallel mechanism model with redundant constraint characteristics, each characteristic constraint is equivalent to a branch constraint in a mechanical system. And similar to the workspace analysis of parallel mechanisms, in an idealized environment, if the surrounding environmental conditions degrade, a motion spiral will inevitably exist. d ξ, which is related to the constraint spiral system of the current observation system. r The product of reciprocity Ξ is zero.

[0172] Furthermore, in a specific general scenario, when the sensor coordinate system undergoes a given helical motion, the reciprocity product of the motion helix and the constraint helix system is calculated to define the strength of the environmental constraint on the motion element:

[0173]

[0174] Ideally, when the sum of the squares of the constraint strengths of all constraints on a specific motion element or set of motion elements is less than a certain threshold, that motion element... d ξ represents the degenerate motion primitives, and the set they form is the degenerate space:

[0175]

[0176] In practical applications, degenerate spinors can be subdivided into two categories: pure even spinors (involving only translation along a vector) and helical spinors (combining rotation and translation along an axis). In degeneracy analysis, these correspond to translational degeneracy and helical degeneracy, respectively. Given the difficulty in unifying these two helical types in form, and the fundamental dimensional differences between helical and translational motions, they need to be analyzed separately. Therefore, the translational degenerate motion space of this robot can be described in the following form: Where tξ is the translational motion primitive of the robot, r ξ j Let j be the j-th environmental constraint screw in the system constraint screw system. After simplification, the constraint strength function of the robot's translational motion primitive can be obtained as described above. t E.

[0177] Furthermore, the helical degenerate motion space of the robot can be described in the following form:

[0178] in, θ Let ξ be the translational motion primitive of the robot. After simplification, the constraint strength function for the helical motion primitive of the robot described above can be obtained. θ E.

[0179] As can be seen, this optional embodiment can first construct a system constraint spiral system for the robot based on the feature constraint model corresponding to the target sensor, and then construct the corresponding degenerate motion space by combining preset translational and spiral motion primitives. This allows for the determination of the constraint strength functions corresponding to the translational and spiral motion primitives, which are then used as the robot's degradation analysis model. By separating the translational and spiral degenerate motion primitives, the threshold setting problem caused by their dimensional differences is solved, reducing threshold confusion. Furthermore, the unit spiral constraint ensures consistent rotational measurements of the motion primitives, reducing the interference of scale differences on feature values ​​and unifying degradation evaluation standards across different scenarios. Simultaneously, by establishing spiral motion primitives, various complex spatial motion forms such as rotation, turning, and spiraling can be covered, significantly expanding the model's applicability. Moreover, the application of the spinor reciprocity product allows the constraint strength measurement to be independent of the specific form and coordinate system of the pose estimation cost function, enabling accurate assessment of degradation at any position. It also eliminates the need for derivation and supports seamless fusion of multi-sensor information, providing a flexible and convenient analysis tool for multi-sensor fusion systems to handle complex problems, significantly improving the reliability and accuracy of robot state estimation in degradation scenarios.

[0180] In another optional embodiment, step 207 above involves performing numerical analysis on the degradation analysis model corresponding to the robot to obtain candidate motion primitives for the robot, including:

[0181] For the constraint strength function of the translational motion primitive, the constraint strength function of the translational motion primitive is reconstructed to obtain the reconstructed constraint strength function of the translational motion primitive;

[0182] Singular value decomposition is performed on the translational degradation model information matrix in the reconstructed translational motion primitive constraint strength function to obtain multiple unit orthogonal eigenvectors corresponding to the translational motion primitive constraint strength function; each unit orthogonal eigenvector has a corresponding eigenvalue.

[0183] Based on the eigenvalues ​​corresponding to all orthogonal eigenvectors, target eigenvectors whose eigenvalues ​​are less than or equal to a preset eigenvalue threshold are determined from all orthogonal eigenvectors.

[0184] Based on the target feature vector, candidate translational degenerate motion primitives for the robot are constructed; and,

[0185] For the constraint strength function of the helical motion primitive, the constrained minimum value optimization function corresponding to the constraint strength function of the helical motion primitive is determined, and the singular value decomposition operation is performed on the constrained minimum value optimization function to obtain the solution function for the minimum eigenvalue of the unconstrained matrix.

[0186] The adaptive damped Newton algorithm is used to solve the minimum eigenvalue solution function of the unconstrained matrix, thereby obtaining the target pitch value and target radius value corresponding to the constraint strength function of the helical motion primitive. Based on the target pitch value and target radius value, candidate helical degenerate motion primitives of the robot are constructed.

[0187] In this optional embodiment, for the translational degradation case, since the possible degenerate motion primitives constitute a subspace in the motion space, within this subspace, the error generated by any motion primitive is less than a threshold. Due to the nature of translational motion, this space is a linear space. For ease of characterization, this embodiment uses SVD (Singular Value Decomposition) to calculate a set of orthonormal bases in the degenerate space to characterize the degenerate space.

[0188] First, the original discriminant function is reconstructed to explicitly separate the variables to be solved (i.e., the reconstructed translational motion primitive constraint strength function): t E = S T t Q T t QS, where tQ = [ r S1 r S2… r S M ] TtQ T t Q is a symmetric and positive definite translational degradation model information matrix. Performing SVD decomposition on this information matrix yields a set of orthogonal eigenvectors arranged in descending order of their corresponding eigenvalues: [VΣ] = SVD( t I), where, t I = t Q T t Q, V, and Σ are the eigenvector matrix and eigenvalue matrix, respectively. Where V = [v1 v2 v3], and Σ = diag(λ1λ2λ3). Furthermore, eigenvectors v3 and v2 correspond to the two translational motion primitives with the minimum constraint strength in this environment. Therefore, in one- or two-dimensional translational degradation environments with sufficient features, it is sufficient to construct candidate degenerate motion primitives using only these two eigenvectors and verify whether the motion primitives satisfy the degradation condition. Therefore, the candidate translational degenerate motion primitives for the robot are:

[0189] Spiral degradation scenarios differ from translational degradation scenarios. In reality, only one-dimensional spiral degradation scenarios (e.g., pipes, circular buildings) and three-dimensional spiral degradation scenarios (e.g., spherical environments) typically exist. For one-dimensional spiral degradation problems, it is only necessary to verify that the decision function... θ The optimization problem can be solved by minimizing the spiral motion primitive E to determine whether it degenerates. Therefore, the following optimization problem (i.e., a constrained minimum optimization function) is constructed: Then, using SVD decomposition, this 7-dimensional constrained optimization problem is equivalently transformed into a 4-dimensional unconstrained optimization problem (i.e., solving for the minimum eigenvalue of an unconstrained matrix): Here, the matrix function λ3(*) represents solving for the minimum eigenvalue of the matrix, and the corresponding eigenvector v3 is the rotation axis direction of the helical motion primitive. This embodiment uses the adaptive damped Newton method to solve the above optimization problem to obtain the information matrix I(r) ξ The pitch h that minimizes the minimum eigenvalue. ★ and vector diameter Therefore, the expression for the unit spiral degenerate motion primitive with the minimum constraint strength (i.e., the candidate spiral degenerate motion primitive of the robot) is obtained as follows: Among them, S ★ Information matrix The eigenvector corresponding to the smallest eigenvalue.

[0190] As can be seen, this optional embodiment can conduct numerical analysis on robot degradation analysis models for two types of motion primitives: translational and helical. After reconstructing the constraint strength function of the translational motion primitive, the target eigenvectors with qualified eigenvalues ​​are selected through singular value decomposition of the translational degradation model information matrix to construct candidate translational degradation motion primitives. For the constraint strength function of the helical motion primitive, it is first transformed into a constrained minimum value optimization function. After obtaining the minimum eigenvalue solution function of the unconstrained matrix through singular value decomposition, the target pitch value and target radius value are solved using the adaptive damped Newton algorithm to construct candidate helical degradation motion primitives. In this way, for translational degradation analysis, the eigenvector with the minimum constraint strength can be accurately extracted to construct candidate primitives to adapt to the characteristics of linear space, improving the accuracy and efficiency of translational degradation scenario analysis. For helical degradation analysis, through optimization problem transformation and the adaptive damped Newton algorithm, the high-dimensional constraint problem can be solved in a reduced dimension, efficiently obtaining key parameters and adapting to one-dimensional and three-dimensional helical degradation scenarios, thus broadening the applicability of the model.

[0191] In another optional embodiment, the step 208 above, which involves determining the robust discriminant function corresponding to the candidate motion primitive and performing degradation detection on the candidate motion primitive using the robust discriminant function to obtain the degradation detection result of the candidate motion primitive, includes:

[0192] For candidate translational degenerate motion primitives, based on the robot's translational motion primitives and the system constraint screw system, the environmental constraint strength function of the robot's translational motion primitives is determined, and the first maximum strength value of the environmental constraint strength function of the translational motion primitives is determined.

[0193] Based on the environmental constraint strength function of the translational motion primitive, the first maximum strength value, and the preset first feature vector angle error robust threshold, the first robust discrimination function corresponding to the candidate translational degenerate motion primitive is determined;

[0194] Based on the first robust discriminant function and the pre-acquired parameters of the robot's translational environment features, a robust discriminant function for translational degradation corresponding to candidate translational degradation motion primitives is constructed. Then, based on the robust discriminant function and a preset translational degradation ratio threshold, degradation detection is performed on the candidate translational degradation motion primitives to obtain the degradation detection results; and...

[0195] For candidate spiral degenerate motion primitives, based on the robot's spiral motion primitives and the system's constrained spiral system, the environmental constraint strength function of the robot's spiral motion primitives is determined, and the second maximum strength value of the environmental constraint strength function of the spiral motion primitives is determined.

[0196] Based on the environmental constraint strength function of the helical motion primitive, the second maximum strength value, and the preset robust threshold for the second feature vector angle error, the second robust discrimination function corresponding to the candidate helical degenerate motion primitive is determined;

[0197] Based on the second robust discriminant function, the position of the robot's spiral environment feature points, the pre-acquired number of spiral environment feature parameters of the robot, and the target spiral feature parameters, a spiral degradation robust discriminant function corresponding to the candidate spiral degradation motion primitive is constructed; the target spiral feature values ​​include the target pitch parameter and the target radius parameter of the spiral degradation model information matrix in the spiral motion primitive constraint strength function;

[0198] Based on the robust discriminant function for spiral degradation and the preset spiral degradation ratio threshold, degradation detection is performed on candidate spiral degradation motion primitives to obtain the degradation detection results of candidate spiral degradation motion primitives.

[0199] In this optional embodiment, it should be noted that when the environment is relatively ideal and the feature extraction noise is low, the constraint strength corresponding to the degraded motion primitive should approach zero. Using a discriminant function and a fixed threshold, it is sufficient to accurately determine whether the environment is degraded and calculate the degraded motion primitive. However, in common robot application scenarios, complex factors such as dynamic distortion, measurement noise, and environmental scale diversity are inevitably encountered, making the boundary between degradation and non-degradation perceived by sensors such as LiDAR very blurry. Due to interference from factors such as scene measurement noise, small non-zero components may be mixed into the constraint conditions. These components will significantly affect the accuracy of degradation discrimination under the cumulative effect. At the same time, changes in scene scale will dynamically adjust the relative weights of these noise components, further increasing the difficulty of discrimination. Therefore, this embodiment constructs an environmental constraint metric representation that is both robust to noise and insensitive to changes in scene scale based on the error function.

[0200] To achieve more effective noise suppression, such as Figure 6 As shown, this embodiment designs a robust function as a noise suppression factor to weaken the influence of external points on the constraint strength. The specific expression of the designed robust function is as follows:

[0201]

[0202] Where U is the constraint strength e j The upper bound of σ is the robust threshold of the eigenvector angle error, which depends on the magnitude of environmental measurement noise.

[0203] After identifying the degenerate motion primitives, this embodiment also provides a scale factor to maintain the consistency of the constraint scale and establishes a scaled degradation threshold. Specifically, the judgment value... t E, θThe upper bound of E is used as a scaling factor to normalize the value of the judgment formula, thereby improving the stability, adaptability, and discriminative ability of the degradation threshold at different scales. In summary, after fusing the noise suppression factor and the scale change suppression factor, a specific scheme for translational and spiral degradation discrimination can be obtained (refer to the discrimination process of the first and second conditions below).

[0204] In this optional embodiment, the method further includes:

[0205] When the degradation detection result of the candidate translational degenerate motion primitive meets the preset first condition, the degradation detection result of the candidate translational degenerate motion primitive is determined to indicate that the application scenario in which the robot is located is degenerating towards the candidate translational degenerate motion primitive:

[0206] When the degradation detection result of the candidate spiral degenerate motion primitive meets the preset second condition, the degradation detection result of the candidate spiral degenerate motion primitive is determined to indicate that the application scenario in which the robot is located is degenerating towards the candidate spiral degenerate motion primitive.

[0207] The first condition is:

[0208]

[0209] Let tξ be a candidate translational degenerate motion primitive, and μ be a translational motion primitive of the robot. t The translation degradation ratio threshold is given by, where, t c k ( t ξ) is the translational degeneracy robustness discriminant function and M1 is the quantitative parameter of the translational environment features. t e j The intensity function of the translational motion primitive under environmental constraints and r ξ j Let σ1 be the j-th environmental constraint screw in the system constraint screw system, σ1 be the robust threshold of the first eigenvector angle error, and ρ be the environmental constraint screw. n ( t e j ,σ1) is the first robust discriminant function and U1 is the first maximum strength value;

[0210] The second condition is:

[0211]

[0212] As a candidate spiral degenerate motion unit, θ ξ is the translational motion primitive of the robot, μ θThe threshold value for spiral degradation is denoted as , where θ c k ( θ ξ) is the robust discriminant function for spiral degeneration and M2 is a quantitative parameter representing the characteristics of the spiral environment. θ e j The helical motion primitive is subject to environmental constraint intensity function and σ² is the robust threshold for the angle error of the second eigenvector, ρ n ( θ e j ,σ2) is the second robust discriminant function and U2 is the second maximum intensity value, r j r represents the location of the helical environmental feature point of the robot. ξ is the target radius parameter, and h is the target pitch parameter.

[0213] In this optional embodiment, further, this embodiment can also set up a degradation analysis acceleration framework based on feature map resampling. This framework realizes parallel computation of degradation analysis and feature extraction, thereby significantly reducing feature analysis time, reducing the impact of feature analysis on the real-time performance of the SLAM system, and improving the practicality of degradation analysis algorithms. The framework is as follows: Figure 7 As shown, the system is divided into a feature analysis module, a local feature map, and feature resampling. When a new set of sensor scan points is input, it is processed synchronously in two threads (feature extraction thread and feature analysis thread). In the feature analysis thread, the feature analysis module uses sensor perception information to extract features based on screw representation and saves them to a local feature map data structure based on ikd-Tree, using spatial coordinates as indices. In the feature extraction thread, the spatial coordinates of the input sensor perception information are used as seeds to sample and search for neighboring extracted features in the local feature map, which serve as the input feature set for degradation analysis.

[0214] In the feature analysis module, this embodiment can employ relatively strict thresholds to improve the accuracy and effectiveness of features. To effectively supplement features within the FOV and enhance the coverage of the feature set, this embodiment can introduce a playback mechanism for historical feature constraint modeling results (unlike traditional real-time feature extraction methods, this framework enables the reuse and long-term management of historical features; furthermore, it makes it possible to predict future pose positioning uncertainty, broadening the application scenarios of degradation analysis). Specifically, this embodiment designs and implements a local feature map based on ikd-Tree: whenever new feature points are extracted from the point cloud, they are immediately added to this local feature map. Figure 7As shown, the local feature map can display the center location of planar features in the tunnel and the latest set of feature constraint information through point clouds in different states (such as different colors). These point cloud data structures embed feature constraint models based on unit spinors. In this way, this embodiment not only makes full use of the constraint model information of historical feature points, but also continuously incorporates newly detected feature points into the feature set, thereby significantly improving the coverage of the feature set. To effectively control the consumption of map storage resources, this embodiment only retains feature points within a certain distance range of the robot and dynamically deletes old feature points outside this range. This strategy ensures the validity of information while also achieving efficient management of storage resources.

[0215] Using the point cloud of the current frame in the world coordinate system as seed points in the initial state, the nearest neighbor features within a certain range around these seed points are searched in the local feature map. This step ensures that only environmental features within the robot's current field of view are included in subsequent degradation analysis, thus guaranteeing the representativeness and relevance of the features. A major advantage of using local feature map resampling to generate feature sets is that it does not require waiting for the feature extraction results of the current frame. Figure 7 The feature extraction and local map update processes in this invention can be processed in parallel with degradation analysis, or even after the SLAM algorithm has finished executing. This allows degradation analysis to begin immediately upon receiving the current frame point cloud data, without relying on the preprocessing steps of the SLAM algorithm, thus significantly reducing the interference of degradation analysis on the real-time performance of the SLAM algorithm. Simultaneously, the constructed feature map can be used to extract possible feature sets for viewpoints within a future range, providing a reference for applications such as active SLAM.

[0216] As can be seen, this optional embodiment can construct robust discriminant functions for candidate translational and spiral degenerate motion primitives to perform sensor degradation detection on the robot. By incorporating noise suppression factors into the robust discriminant functions, external interference can be effectively weakened, resolving the ambiguity of degradation boundaries caused by dynamic distortion and measurement noise. This improves the accuracy of state estimation in complex environments. Furthermore, the design of scale factors and proportional degradation thresholds weakens the sensitivity of the judgment criteria to changes in scene scale, thereby enhancing the adaptability and stability of state estimation in scenes of different scales. In addition, constructing dedicated discrimination logic for translational and spiral scenes can adapt to the differences in their motion characteristics, improving the targeting of degradation detection. This helps to accurately identify the degradation trend of the robot's scene, providing a solid basis for subsequent state estimation optimization and control decisions, and significantly improving the reliability of the robot's operation in complex dynamic environments.

[0217] Example 3

[0218] Please see Figure 3 , Figure 3 This is a schematic diagram of a robot sensor degradation analysis device based on spinor theory disclosed in an embodiment of the present invention. Figure 3 As shown, the robot sensor degradation analysis device based on spinor theory may include:

[0219] The determination module 301 is used to determine the equivalent mechanism model corresponding to the robot's target sensor;

[0220] The construction module 302 is used to construct the observation feature measurement model corresponding to the target sensor based on the equivalent mechanism model corresponding to the target sensor; construct the feature constraint model corresponding to the target sensor based on the observation feature measurement model corresponding to the target sensor; and construct the degradation analysis model corresponding to the robot based on the feature constraint model corresponding to the target sensor.

[0221] Analysis module 303 is used to perform numerical analysis on the degradation analysis model corresponding to the robot to obtain candidate motion primitives of the robot;

[0222] The detection module 304 is used to determine the robust discriminant function corresponding to the candidate motion primitive, and to perform degradation detection on the candidate motion primitive through the robust discriminant function to obtain the degradation detection result of the candidate motion primitive.

[0223] In this embodiment of the invention, the candidate motion primitives include candidate translational degenerate motion primitives and candidate spiral degenerate motion primitives; the degradation detection results of the candidate motion primitives are used to determine whether the application scenario in which the robot is located is degenerating towards the candidate motion primitives.

[0224] It is evident that implementation Figure 3 The described robot sensor degradation analysis device based on screw theory can construct sensor observation feature measurement and constraint models through screw analysis, and use unit screws to represent environmental constraints and motion primitives. Through equivalent mechanism modeling, feature constraint derivation, numerical analysis, and robust discrimination, the robot sensor degradation detection process is realized. In this way, it can solve the problems of high complexity and poor adaptability of state estimation in traditional methods without demand derivation or cost function dependence. It also reduces the influence of scene noise and environmental scale and improves modeling efficiency and versatility. At the same time, through the degradation analysis model and robust discrimination function, the translational and spiral degradation trends can be accurately located, which helps to improve the accuracy of robot sensor degradation detection. This provides a key basis for robot scene perception and state estimation, which helps the system to provide early warning and optimize control, and ensures the reliability and stability of operation in complex environments.

[0225] In an optional embodiment, the target sensor includes a lidar sensor, a distance measurement sensor, and an orientation measurement sensor. The observation feature measurement model corresponding to the target sensor includes a point-area feature measurement model and a point-edge feature measurement model corresponding to the lidar sensor, a distance feature measurement model corresponding to the distance measurement sensor, and an orientation feature measurement model corresponding to the orientation measurement sensor.

[0226] Specifically, the construction module 302 constructs the feature constraint model corresponding to the target sensor based on the observation feature measurement model corresponding to the target sensor in the following ways:

[0227] Based on the point-to-surface feature measurement model corresponding to the lidar sensor, the first degenerate space spiral system corresponding to the point-to-surface feature measurement model is determined, and the first degenerate space spiral system is subjected to anti-spiral analysis to obtain the planar feature constraint spiral system corresponding to the lidar sensor, which serves as the first feature constraint model corresponding to the lidar sensor.

[0228] Based on the point-edge feature measurement model corresponding to the lidar sensor, the second degenerate space spiral system corresponding to the point-edge feature measurement model is determined, and the second degenerate space spiral system is subjected to anti-spiral analysis to obtain the linear feature constraint spiral system corresponding to the lidar sensor, which serves as the second feature constraint model corresponding to the lidar sensor.

[0229] Based on the distance feature measurement model corresponding to the distance measurement sensor, the third degenerate space spiral system corresponding to the distance feature measurement model is determined, and the third degenerate space spiral system is subjected to anti-spiral analysis to obtain the distance measurement constraint spiral system corresponding to the distance measurement sensor, which serves as the third feature constraint model corresponding to the distance measurement sensor.

[0230] Based on the orientation feature measurement model corresponding to the orientation measurement sensor, the fourth degenerate space spiral system corresponding to the orientation feature measurement model is determined, and the fourth degenerate space spiral system is subjected to anti-spiral analysis to obtain the orientation measurement constraint spiral system corresponding to the orientation measurement sensor, which serves as the fourth feature constraint model corresponding to the orientation measurement sensor.

[0231] It is evident that implementation Figure 3The described robot sensor degradation analysis device based on screw theory first constructs equivalent mechanism models and observation feature measurement models for robot target sensors (LiDAR, distance measurement, and orientation measurement sensors), and determines the corresponding degradation space spiral system for different feature models of each sensor. Then, through anti-screw analysis, it obtains the feature constraint models of each sensor, and constructs the robot's degradation analysis model based on these models. The robot sensor degradation detection process is realized through numerical analysis and robust discrimination. In this way, accurate constraint modeling of LiDAR point-to-surface and point-to-edge features, as well as features of distance and orientation sensors, is achieved, improving the reliability and accuracy of feature constraints for each sensor. At the same time, through anti-screw analysis, the correspondence between degradation space and constraint spiral system is effectively established, making the degradation detection process more consistent with the characteristics of different sensors. This significantly improves the accuracy of the robot's judgment of sensor degradation trends in complex environments, providing a more reliable basis for robot state estimation and motion control.

[0232] In another optional embodiment, the construction module 302 constructs the degradation analysis model corresponding to the robot based on the feature constraint model corresponding to the target sensor in the following specific ways:

[0233] Based on the feature constraint model corresponding to the target sensor, construct the robot's system constraint spiral system;

[0234] Based on the preset translational motion primitives of the robot and the system constraint spiral system, the translational degenerate motion space of the robot is constructed, and based on the preset spiral motion primitives of the robot and the system constraint spiral system, the spiral degenerate motion space of the robot is constructed.

[0235] Based on the translational degenerate motion space, determine the constraint strength function of the robot's translational motion primitives, and based on the helical degenerate motion space, determine the constraint strength function of the robot's helical motion primitives.

[0236] The constraint strength functions of the translational motion primitive and the constraint strength functions of the helical motion primitive are determined as the corresponding degradation analysis models for the robot.

[0237] In this optional embodiment, the constraint strength function of the robot's translational motion primitives is further defined as follows:

[0238]

[0239] Where M1 is the number of translational environmental features of the robot obtained in advance. r S j Let be the principal vector of the j-th environmental constraint screw in the system constraint screw system. t S is the preset translation direction vector;

[0240] The constraint strength function of the robot's helical motion primitive is:

[0241]

[0242] Where M2 is the number of pre-acquired spiral environment features of the robot. Let be the follower vector of the j-th environmental constraint screw in the system constraint screw system. θ S is the preset rotation axis direction vector, r ξ is the preset target radius parameter, and h is the preset target pitch parameter.

[0243] It is evident that implementation Figure 3 The described spiral theory-based robot sensor degradation analysis device first constructs a system constraint spiral system for the robot based on the characteristic constraint model corresponding to the target sensor. Then, it constructs corresponding degradation motion spaces by combining preset translational and spiral motion primitives. It then determines the constraint strength functions corresponding to the translational and spiral motion primitives and uses both as the robot's degradation analysis model. By separating the translational and spiral degenerate motion primitives, the threshold setting problem caused by their dimensional differences is solved, reducing threshold confusion. Furthermore, the unit spiral constraint ensures consistent rotational measurements of the motion primitives, reducing the interference of scale differences on feature values ​​and unifying degradation evaluation standards across different scenarios. Simultaneously, by establishing spiral motion primitives, it can cover various complex spatial motion forms such as rotation, revolution, and spiral, significantly expanding the model's applicability. Furthermore, by applying the spinor reciprocity product, the constraint strength measurement does not depend on the specific form and coordinate system of the pose estimation cost function. It can accurately assess the degree of degradation at any position, and it does not require derivation and supports seamless fusion of multi-sensor information. It provides a flexible and convenient analysis tool for multi-sensor fusion systems to handle complex problems, and significantly improves the reliability and accuracy of robot state estimation in degradation scenarios.

[0244] In another optional embodiment, the analysis module 303 performs numerical analysis on the degradation analysis model corresponding to the robot to obtain the candidate motion primitives of the robot in the following specific ways:

[0245] For the constraint strength function of the translational motion primitive, the constraint strength function of the translational motion primitive is reconstructed to obtain the reconstructed constraint strength function of the translational motion primitive;

[0246] Singular value decomposition is performed on the translational degradation model information matrix in the reconstructed translational motion primitive constraint strength function to obtain multiple unit orthogonal eigenvectors corresponding to the translational motion primitive constraint strength function; each unit orthogonal eigenvector has a corresponding eigenvalue.

[0247] Based on the eigenvalues ​​corresponding to all orthogonal eigenvectors, target eigenvectors whose eigenvalues ​​are less than or equal to a preset eigenvalue threshold are determined from all orthogonal eigenvectors.

[0248] Based on the target feature vector, candidate translational degenerate motion primitives for the robot are constructed; and,

[0249] For the constraint strength function of the helical motion primitive, the constrained minimum value optimization function corresponding to the constraint strength function of the helical motion primitive is determined, and the singular value decomposition operation is performed on the constrained minimum value optimization function to obtain the solution function for the minimum eigenvalue of the unconstrained matrix.

[0250] The adaptive damped Newton algorithm is used to solve the minimum eigenvalue solution function of the unconstrained matrix, thereby obtaining the target pitch value and target radius value corresponding to the constraint strength function of the helical motion primitive. Based on the target pitch value and target radius value, candidate helical degenerate motion primitives of the robot are constructed.

[0251] It is evident that implementation Figure 3 The described robot sensor degradation analysis device based on screw theory can perform numerical analysis on robot degradation analysis models for two types of motion primitives: translational and helical. For translational motion primitives, the constraint strength function is reconstructed, and the information matrix of the translational degradation model is decomposed using singular value decomposition to select target eigenvectors with acceptable eigenvalues, thus constructing candidate translational degenerate motion primitives. For helical motion primitives, the constraint strength function is first transformed into a constrained minimum optimization function. After obtaining the minimum eigenvalue solution function of the unconstrained matrix through singular value decomposition, the target pitch value and target radius value are solved using an adaptive damped Newton algorithm, thereby constructing candidate helical degenerate motion primitives. In this way, for translational degradation analysis, the eigenvector with the minimum constraint strength can be accurately extracted to construct candidate primitives, adapting to the characteristics of linear space and improving the accuracy and efficiency of translational degradation scenario analysis. For helical degradation analysis, through optimization problem transformation and the adaptive damped Newton algorithm, high-dimensional constraint problems can be solved in a reduced-dimensional manner, efficiently obtaining key parameters and adapting to one-dimensional and three-dimensional helical degradation scenarios, thus broadening the model's applicability.

[0252] In another optional embodiment, the detection module 304 determines the robust discriminant function corresponding to the candidate motion primitive, and performs degradation detection on the candidate motion primitive using the robust discriminant function. The specific method for obtaining the degradation detection result of the candidate motion primitive includes:

[0253] For candidate translational degenerate motion primitives, based on the robot's translational motion primitives and the system constraint screw system, the environmental constraint strength function of the robot's translational motion primitives is determined, and the first maximum strength value of the environmental constraint strength function of the translational motion primitives is determined.

[0254] Based on the environmental constraint strength function of the translational motion primitive, the first maximum strength value, and the preset first feature vector angle error robust threshold, the first robust discrimination function corresponding to the candidate translational degenerate motion primitive is determined;

[0255] Based on the first robust discriminant function and the pre-acquired parameters of the robot's translational environment features, a robust discriminant function for translational degradation corresponding to candidate translational degradation motion primitives is constructed. Then, based on the robust discriminant function and a preset translational degradation ratio threshold, degradation detection is performed on the candidate translational degradation motion primitives to obtain the degradation detection results; and...

[0256] For candidate spiral degenerate motion primitives, based on the robot's spiral motion primitives and the system's constrained spiral system, the environmental constraint strength function of the robot's spiral motion primitives is determined, and the second maximum strength value of the environmental constraint strength function of the spiral motion primitives is determined.

[0257] Based on the environmental constraint strength function of the helical motion primitive, the second maximum strength value, and the preset robust threshold for the second feature vector angle error, the second robust discrimination function corresponding to the candidate helical degenerate motion primitive is determined;

[0258] Based on the second robust discriminant function, the position of the robot's spiral environment feature points, the pre-acquired number of spiral environment feature parameters of the robot, and the target spiral feature parameters, a spiral degradation robust discriminant function corresponding to the candidate spiral degradation motion primitive is constructed.

[0259] Based on the robust discriminant function for spiral degradation and the preset spiral degradation ratio threshold, degradation detection is performed on candidate spiral degradation motion primitives to obtain the degradation detection results of candidate spiral degradation motion primitives.

[0260] In this optional embodiment, the target helical eigenvalues ​​include the target pitch parameter and the target radius parameter of the helical degradation model information matrix in the helical motion primitive constraint strength function.

[0261] Furthermore, module 301 is also used for:

[0262] When the degradation detection result of the candidate translational degenerate motion primitive meets the preset first condition, the degradation detection result of the candidate translational degenerate motion primitive is determined to indicate that the application scenario in which the robot is located is degenerating towards the candidate translational degenerate motion primitive:

[0263] When the degradation detection result of the candidate spiral degenerate motion primitive meets the preset second condition, the degradation detection result of the candidate spiral degenerate motion primitive is determined to indicate that the application scenario in which the robot is located is degenerating toward the candidate spiral degenerate motion primitive;

[0264] The first condition is:

[0265]

[0266] Let tξ be a candidate translational degenerate motion primitive, and μ be a translational motion primitive of the robot. t The translation degradation ratio threshold is given by, where, t c k ( t ξ) is the translational degeneracy robustness discriminant function and M1 is the quantitative parameter of the translational environment features. t e j The intensity function of the translational motion primitive under environmental constraints and r ξ j Let σ1 be the j-th environmental constraint screw in the system constraint screw system, σ1 be the robust threshold of the first eigenvector angle error, and ρ be the environmental constraint screw. n ( t e j ,σ1) is the first robust discriminant function and U1 is the first maximum strength value;

[0267] The second condition is:

[0268]

[0269] As a candidate spiral degenerate motion unit, θ ξ is the translational motion primitive of the robot, μ θ The threshold value for spiral degradation is denoted as , where θ c k ( θ ξ) is the robust discriminant function for spiral degeneration and M2 is a quantitative parameter representing the characteristics of the spiral environment. θ e j The helical motion primitive is subject to environmental constraint intensity function and σ² is the robust threshold for the angle error of the second eigenvector, ρ n ( θ e j ,σ2) is the second robust discriminant function and U2 is the second maximum intensity value, r j r represents the location of the helical environmental feature point of the robot. ξ is the target radius parameter, and h is the target pitch parameter.

[0270] It is evident that implementation Figure 3The described spinor-based robot sensor degradation analysis device can construct robust discriminant functions for candidate translational and helical degenerate motion primitives to detect robot sensor degradation. By incorporating noise suppression factors into the robust discriminant functions, external interference can be effectively weakened, resolving the ambiguity of degradation boundaries caused by dynamic distortion and measurement noise. This improves the accuracy of state estimation in complex environments. Furthermore, the design of scale factors and proportional degradation thresholds mitigates the sensitivity of the judgment criteria to changes in scene scale, enhancing the adaptability and stability of state estimation across different scales. In addition, constructing dedicated discriminant logic for translational and helical scenes adapts to the differences in their motion characteristics, improving the targeting of degradation detection. This helps to accurately identify the degradation trend of the robot's scene, providing a solid basis for subsequent state estimation optimization and control decisions, significantly improving the reliability of robot operations in complex dynamic environments.

[0271] Example 4

[0272] Please see Figure 4 , Figure 4 This is a schematic diagram of another robot sensor degradation analysis device based on spinor theory disclosed in an embodiment of the present invention. Figure 4 As shown, the robot sensor degradation analysis device based on spinor theory may include:

[0273] Memory 401 storing executable program code;

[0274] Processor 402 coupled to memory 401;

[0275] The processor 402 calls the executable program code stored in the memory 401 to execute the steps in the robot sensor degradation analysis method based on spinor theory described in Embodiment 1 or Embodiment 2 of the present invention.

[0276] Example 5

[0277] This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute the steps in the robot sensor degradation analysis method based on spinor theory described in Embodiment 1 or Embodiment 2 of this invention.

[0278] Example 6

[0279] This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the robot sensor degradation analysis method based on spinor theory described in Embodiment 1 or Embodiment 2.

[0280] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0281] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0282] Finally, it should be noted that the robot sensor degradation analysis method and apparatus based on spinor theory disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for analyzing the degradation of robot sensors based on spinor theory, characterized in that, The method includes: Determine the equivalent mechanism model corresponding to the target sensor of the robot, and construct the observation feature measurement model corresponding to the target sensor based on the equivalent mechanism model corresponding to the target sensor; Based on the observation feature measurement model corresponding to the target sensor, a feature constraint model corresponding to the target sensor is constructed, and based on the feature constraint model corresponding to the target sensor, a degradation analysis model corresponding to the robot is constructed. Numerical analysis is performed on the degradation analysis model corresponding to the robot to obtain candidate motion primitives of the robot; the candidate motion primitives include candidate translational degenerate motion primitives and candidate spiral degenerate motion primitives; A robust discriminant function is determined for the candidate motion primitive, and degradation detection is performed on the candidate motion primitive using the robust discriminant function to obtain the degradation detection result of the candidate motion primitive; the degradation detection result of the candidate motion primitive is used to determine whether the application scenario in which the robot is located is degrading towards the candidate motion primitive.

2. The robot sensor degradation analysis method based on spinor theory according to claim 1, characterized in that, The target sensor includes a lidar sensor, a distance measurement sensor, and an orientation measurement sensor. The observation feature measurement model corresponding to the target sensor includes the point-area feature measurement model and the point-edge feature measurement model corresponding to the lidar sensor, the distance feature measurement model corresponding to the distance measurement sensor, and the orientation feature measurement model corresponding to the orientation measurement sensor. The step of constructing a feature constraint model corresponding to the target sensor based on the observation feature measurement model corresponding to the target sensor includes: Based on the point-to-surface feature measurement model corresponding to the lidar sensor, the first degenerate space spiral system corresponding to the point-to-surface feature measurement model is determined, and the first degenerate space spiral system is subjected to anti-spiral analysis to obtain the planar feature constraint spiral system corresponding to the lidar sensor, which is used as the first feature constraint model corresponding to the lidar sensor. Based on the point-edge feature measurement model corresponding to the lidar sensor, the second degenerate space spiral system corresponding to the point-edge feature measurement model is determined, and the second degenerate space spiral system is subjected to anti-spiral analysis to obtain the linear feature constraint spiral system corresponding to the lidar sensor, which is used as the second feature constraint model corresponding to the lidar sensor. Based on the distance feature measurement model corresponding to the distance measurement sensor, the third degenerate space spiral system corresponding to the distance feature measurement model is determined, and the third degenerate space spiral system is subjected to anti-spiral analysis to obtain the distance measurement constraint spiral system corresponding to the distance measurement sensor, which serves as the third feature constraint model corresponding to the distance measurement sensor. Based on the orientation feature measurement model corresponding to the orientation measurement sensor, the fourth degenerate space spiral system corresponding to the orientation feature measurement model is determined, and the fourth degenerate space spiral system is subjected to anti-spiral analysis to obtain the orientation measurement constraint spiral system corresponding to the orientation measurement sensor, which serves as the fourth feature constraint model corresponding to the orientation measurement sensor.

3. The robot sensor degradation analysis method based on spinor theory according to claim 2, characterized in that, The step of constructing a degradation analysis model for the robot based on the feature constraint model corresponding to the target sensor includes: Based on the feature constraint model corresponding to the target sensor, construct the system constraint spiral system of the robot; Based on the preset translational motion primitives of the robot and the system constraint spiral system, the translational degenerate motion space of the robot is constructed, and based on the preset spiral motion primitives of the robot and the system constraint spiral system, the spiral degenerate motion space of the robot is constructed. Based on the translational degenerate motion space, the translational motion primitive constraint strength function of the robot is determined, and based on the helical degenerate motion space, the helical motion primitive constraint strength function of the robot is determined. The constraint strength functions of the translational motion primitives and the constraint strength functions of the helical motion primitives are determined as the degradation analysis model corresponding to the robot.

4. The robot sensor degradation analysis method based on spinor theory according to claim 3, characterized in that, The constraint strength function of the translational motion primitive of the robot is: Where M1 is the number of translational environmental features of the robot obtained in advance. r S j Let be the principal vector of the j-th environmental constraint screw in the system constraint screw system. t S is the preset translation direction vector; The constraint strength function of the helical motion primitive of the robot is: Where M2 is the number of spiral environment features of the robot obtained in advance. Let be the follower vector of the j-th environmental constraint screw in the system constraint screw system. θ S is the preset rotation axis direction vector, r ξ is the preset target radius parameter, and h is the preset target pitch parameter.

5. The robot sensor degradation analysis method based on spinor theory according to claim 3, characterized in that, The step of performing numerical analysis on the degradation analysis model corresponding to the robot to obtain candidate motion primitives of the robot includes: For the translational motion primitive constraint strength function, the translational motion primitive constraint strength function is reconstructed to obtain the reconstructed translational motion primitive constraint strength function; Singular value decomposition is performed on the translational degradation model information matrix in the reconstructed translational motion primitive constraint strength function to obtain multiple unit orthogonal eigenvectors corresponding to the translational motion primitive constraint strength function; each unit orthogonal eigenvector has a corresponding eigenvalue. Based on the eigenvalues ​​corresponding to all the unit orthogonal eigenvectors, a target eigenvector whose eigenvalue is less than or equal to a preset eigenvalue threshold is determined from all the unit orthogonal eigenvectors; Based on the target feature vector, candidate translational degenerate motion primitives of the robot are constructed; and, For the constraint strength function of the spiral motion primitive, the constrained minimum value optimization function corresponding to the constraint strength function of the spiral motion primitive is determined, and the singular value decomposition operation is performed on the constrained minimum value optimization function to obtain the unconstrained matrix minimum eigenvalue solution function. The adaptive damped Newton algorithm is used to solve the minimum eigenvalue solution function of the unconstrained matrix to obtain the target pitch value and target radius value corresponding to the constraint strength function of the helical motion primitive. Based on the target pitch value and the target radius value, the candidate helical degenerate motion primitive of the robot is constructed.

6. The robot sensor degradation analysis method based on spinor theory according to any one of claims 3-5, characterized in that, The step of determining the robust discriminant function corresponding to the candidate motion primitive and performing degradation detection on the candidate motion primitive using the robust discriminant function to obtain the degradation detection result of the candidate motion primitive includes: For the candidate translational degenerate motion primitive, based on the translational motion primitive of the robot and the system constraint screw system, the environmental constraint strength function of the translational motion primitive of the robot is determined, and the first maximum strength value of the environmental constraint strength function of the translational motion primitive is determined. Based on the environmental constraint strength function of the translational motion primitive, the first maximum strength value, and the preset first feature vector angle error robustness threshold, the first robust discrimination function corresponding to the candidate translational degenerate motion primitive is determined; Based on the first robust discriminant function and the pre-acquired parameters of the robot's translational environment features, a robust discriminant function for translational degradation corresponding to the candidate translational degradation motion primitive is constructed. Then, based on the robust discriminant function and a preset translational degradation ratio threshold, degradation detection is performed on the candidate translational degradation motion primitive to obtain the degradation detection result of the candidate translational degradation motion primitive; and... For the candidate helical degenerate motion primitive, based on the helical motion primitive of the robot and the system constraint helical system, the environmental constraint strength function of the helical motion primitive of the robot is determined, and the second maximum strength value of the environmental constraint strength function of the helical motion primitive is determined. Based on the environmental constraint strength function of the spiral motion primitive, the second maximum strength value, and the preset second feature vector angle error robust threshold, the second robust discrimination function corresponding to the candidate spiral degenerate motion primitive is determined; Based on the second robust discriminant function, the position of the spiral environment feature points of the robot, the pre-acquired spiral environment feature quantity parameters of the robot, and the target spiral feature parameters, a spiral degradation robust discriminant function corresponding to the candidate spiral degradation motion primitive is constructed; the target spiral feature value includes the target pitch parameter and the target radius parameter of the spiral degradation model information matrix in the constraint strength function of the spiral motion primitive; Based on the spiral degradation robust discrimination function and the preset spiral degradation ratio threshold, degradation detection is performed on the candidate spiral degradation motion primitives to obtain the degradation detection results of the candidate spiral degradation motion primitives.

7. The robot sensor degradation analysis method based on spinor theory according to claim 6, characterized in that, The method further includes: When the degradation detection result of the candidate translational degenerate motion primitive meets a preset first condition, the degradation detection result of the candidate translational degenerate motion primitive is determined to indicate that the application scenario in which the robot is located is degenerating towards the candidate translational degenerate motion primitive. When the degradation detection result of the candidate spiral degenerate motion primitive meets the preset second condition, it is determined that the degradation detection result of the candidate spiral degenerate motion primitive is used to indicate that the application scenario in which the robot is located is degenerating towards the candidate spiral degenerate motion primitive; The first condition is: For the candidate translational degenerate motion primitive, t ξ is the translational motion element of the robot, μ t The translation degradation ratio threshold is given, where, t c k ( t ξ) is the translational degeneracy robustness discriminant function and M1 is the quantity parameter of the translational environment features. t e j The translational motion primitive is subject to environmental constraint intensity function and r ξ j Let σ1 be the j-th environmental constraint spiral in the system constraint spiral system, σ1 be the robust threshold of the first eigenvector angle error, and ρ be the environmental constraint spiral. n ( t e j ,σ1) is the first robust discriminant function and U1 is the first maximum strength value; The second condition is: Let θξ be the candidate helical degenerate motion primitive, and μ be the translational motion primitive of the robot. θ The spiral degradation ratio threshold is given, where, θ c k ( θ ξ) is the robust discriminant function for spiral degeneration and M2 is the numerical parameter of the spiral environment characteristics. θ e j The helical motion primitive is subject to environmental constraint intensity function and σ² is the robust threshold for the angle error of the second eigenvector, ρ n ( θ e j ,σ2) is the second robust discriminant function and U2 is the second maximum intensity value, r j Let r be the location of the spiral environmental feature point of the robot. ξ Let h be the target radius parameter and h be the target pitch parameter.

8. A robot sensor degradation analysis device based on spinor theory, characterized in that, The device includes: The determination module is used to determine the equivalent mechanism model corresponding to the target sensor of the robot; The construction module is used to construct an observation feature measurement model corresponding to the target sensor based on the equivalent mechanism model corresponding to the target sensor; construct a feature constraint model corresponding to the target sensor based on the observation feature measurement model corresponding to the target sensor; and construct a degradation analysis model corresponding to the robot based on the feature constraint model corresponding to the target sensor. The analysis module is used to perform numerical analysis on the degradation analysis model corresponding to the robot to obtain candidate motion primitives of the robot; the candidate motion primitives include candidate translational degradation motion primitives and candidate spiral degradation motion primitives; The detection module is used to determine the robust discriminant function corresponding to the candidate motion primitive, and to perform degradation detection on the candidate motion primitive through the robust discriminant function to obtain the degradation detection result of the candidate motion primitive; the degradation detection result of the candidate motion primitive is used to determine whether the application scenario in which the robot is located is degrading towards the candidate motion primitive.

9. A robot sensor degradation analysis device based on spinor theory, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the robot sensor degradation analysis method based on spinor theory as described in any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, are used to execute the robot sensor degradation analysis method based on spinor theory as described in any one of claims 1-7.

Images