Quadruped Robot Dog-Assisted Attitude Correction Method and System for Cable Tunnel Inspection

Abstract

The present invention relates to a method and system for auxiliary attitude correction of a quadruped robot dog for cable tunnel inspection, including: measuring the distance to the wall surface using the wire harnesses at different included angles in the vertical direction of the lidar, calculating the parameters of attitude change using the distance data, and fusing the attitude change parameters based on space using the principal component analysis method; obtaining the distances from the wire harnesses at different included angles of the lidar to the wall surface at different times, substituting them into the distance update function to calculate the change rate and average change rate of the parameters, substituting the distance data and the average change rate of the parameters into the distance update function to iteratively calculate the distance of each group, obtaining the attitude change parameters through the new distance, and fusing the attitude change parameters based on time using the principal component analysis method; fusing the attitude change parameters based on space and time using the spatio-temporal adaptive fusion method to obtain the optimal estimated value of the attitude change parameters of the intelligent quadruped robot dog. The present invention can make up for the shortcoming that the intelligent quadruped robot dog cannot judge its own attitude change during driving.

Description

Technical Field

[0001] The present invention belongs to the technical field of robots, and particularly relates to a method and system for assisting attitude correction of a quadruped robot dog for cable tunnel inspection. Background Art

[0002] With the development of robot technology, various robots have been studied and applied in various fields. Quadruped and other legged robots have better passability compared to wheeled robots through various terrains and obstacles. In some narrow and enclosed tunnel environments where the working status of equipment needs to be frequently inspected, using an intelligent quadruped robot dog for inspection work can well complete the tasks, so there is a greater application demand in complex working scenarios.

[0003] Multiple sensors can be integrated on the intelligent quadruped robot dog, such as cameras, lidar, integrated navigation, etc. Through these sensors and corresponding algorithms, the surrounding environmental information can be sensed, a global or local map can be constructed, and then through the path planning algorithm, a global or local path can be planned, and the speed and attitude of the intelligent robot dog can be controlled in real time through corresponding strategies, so that the intelligent robot dog can actively avoid obstacles and at the same time move safely and reliably along the planned path to reach the destination and complete the corresponding tasks. When the intelligent robot dog is moving along the planned path, it will inevitably encounter obstacles. Therefore, the obstacle avoidance ability of the robot dog is also very important for completing tasks. During the obstacle avoidance process, it is necessary to judge the attitude of the intelligent robot dog in real time and make corresponding adjustments in order to safely and quickly avoid obstacles.

[0004] Currently, in order to obtain the position and attitude of the intelligent robot dog, precise information can be provided by the GPS positioning system and the inertial navigation system. However, in a narrow and enclosed tunnel environment, the GPS positioning system cannot work properly because satellite signals cannot be received, and the inertial navigation system has cumulative errors over time during long-term operation. Therefore, the attitude cannot be accurately obtained. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a method and system for assisting attitude correction of a quadruped robot dog for cable tunnel inspection to realize the application of the intelligent quadruped robot dog in narrow tunnel inspection. This method only uses the data collected by lidar to calculate the attitude change information of the intelligent robot dog, so as to assist the robot dog in correcting the attitude, provide more information for the control system of the intelligent robot dog, enhance the obstacle avoidance ability, and enable it to pass through obstacles safely and quickly.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method and system for correcting the posture of a four-legged robot dog for cable tunnel inspection includes the following steps:

[0008] S1: Use the laser radar to collect different angles and distances from the beam to the wall at multiple times, and then pre-process the collected data to eliminate invalid data;

[0009] S2: Calculate the space-based attitude change parameters using the distances measured by different laser radar angles and different beams at the same time, and fuse the space-based attitude change parameters through the principal component analysis algorithm;

[0010] S3: Use the distance update function to describe the parameter change rate of the distance between the robot dog and the wall at multiple moments, and calculate the average change rate of the parameter. Then iterate the average change rate of the parameter and the distance to recalculate the distance. Use the recalculated distance to find the time-based posture change parameters, and use the principal component analysis algorithm to fuse the time-based posture change parameters.

[0011] S4: Use the spatiotemporal adaptive fusion method to fuse the posture change parameters based on time and space, and finally obtain the optimal posture change parameters of the robot dog to assist in correcting the posture.

[0012] Further, in step S1, a 16-line laser radar is used to obtain data of different angles and distances from the beam to the wall. The laser radar position is fixed for acquisition and is installed in the middle position on the back of the intelligent quadruped robot dog (referred to as the robot dog) 2. The installation position is 0.5-0.7 meters from the ground, the detection angle is 360 degrees, the resolution is 0.18 degrees, and the distance from the wall is collected at a rate of 10 frames per second. Then, invalid data that is particularly large or particularly small in the detection distance data is eliminated. The specific data acquisition and processing include:

[0013] S11: The x-axis of the laser radar is the forward direction of the robot dog, the y-axis is perpendicular to the robot dog's body and the positive direction is from right to left, and the positive direction of the z-axis is from bottom to top, with the origin being o. Different angles refer to the horizontal plane with the vertical angle of the laser radar at 0° and the y-axis as the bisector. Let oa 30° With ob 30° The angle is 30°, oa 45° With ob 45° The angle is 45°, oa 60° With ob 60° The angle is 60°, oa 90° With ob 90° The angle is 90°, oa 120° With ob 120° The angle between the two is 120°, so as to collect data;

[0014] S12: Select wire harnesses at different vertical angles and different included angles according to the environment adaptive screening method, and collect data with the y-axis as the bisector for all of them;

[0015] S13: Find and eliminate the data in the collected data that do not conform to the narrow tunnel scenario, that is, eliminate the data with particularly large and particularly small distances.

[0016] Furthermore, the attitude change parameters for calculating and fusing data of different included angles and different wire harnesses at the same moment described in step S2 include:

[0017] S21: Measure the distances between the intersections of different included angles and different wire harnesses and (where α = 30°, 45°, 60°, 90°, 120°, β = 1°, 7°, 15°) and the wall, and the distances of two small angles (each small angle is 0.18°) before and after each intersection, and then use the distances of these 5 intersections to calculate the average distance to replace and the distances to the wall. The calculation formula is as follows:

[0018]

[0019] Among them, and represent the straight-line distance with a vertical direction angle of β and an included angle between the lidar wire harnesses oa and ob of α. oa -2 and ob -2 represent the distances from the first two small angles at the intersection of the lidar scan line and the wall to the wall; oa -1 and ob -1 represent the distances from the first small angle at the intersection of the lidar scan line and the wall to the wall; oa and ob represent the intersection distance of the midpoint between the lidar scan line and the wall; oa2 and ob2 represent the distances of the last two small angles at the intersection of the lidar scan line and the wall respectively, and oa1 and ob1 represent the distances of the last small angle at the intersection of the lidar scan line and the wall respectively.

[0020] S22: Substitute and into the difference over summation (DOS) function to calculate the attitude change parameters based on space to measure the attitude change of the quadruped robot. The specific calculation formula is as follows:

[0021]

[0022] Among them, θ represents the attitude change parameter of the robotic dog, which can reflect the attitude change of the robotic dog. If θ < 0, it means the head of the robotic dog is tilted to the left; if θ > 0, it means the head of the robotic dog is tilted to the right; if θ = 0, it means the robotic dog is parallel to the wall. The larger |θ| is, the greater the head deflection of the robotic dog is, and vice versa, the smaller the deflection is.

[0023] S23: Use the data fusion algorithm based on principal component analysis to fuse the attitude change parameters calculated at different angles and wire harnesses, so as to obtain the optimal attitude change parameter estimation based on space.

[0024] Optionally, the specific steps of S23 are as follows:

[0025] ① Regard the attitude change parameters θ based on space calculated at different angles and different wire harnesses as a whole Θ = (θ1, θ2,..., θ n ), where n is the number of data. Let the covariance matrix of Θ be D, and obtain the eigenvalues as λ1 ≥ λ2 ≥... ≥ λ n and the corresponding unitary orthogonal eigenvectors as e1, e2,..., e n , where the unitary orthogonal eigenvector of the i-th principal component is e i = (e 1i , e 2i ,..., e ni ). Therefore, the i-th principal component is:

[0026] f i = e′ i Θ = e 1i θ1 +... + e ni θ n (3)

[0027] ② Calculate the variance contribution rate i of the principal component f and the cumulative variance contribution rate of the first m principal components. The calculation formulas are as follows:

[0028]

[0029]

[0030] Among them, when the value of m makes the cumulative variance contribution rate reach a proportion suitable for the current situation, other principal components can be omitted.

[0031] ③ Calculate the comprehensive support degree of different angles and wire harnesses. The specific formula is as follows:

[0032]

[0033] Among them, f i *For each θ i The principal component most relevant to it, and determine ρ(θ i , f j ) through the correlation coefficient between them, and select the f with the largest value of ρ(θ i , f j ) as the principal component most relevant to θ j , and the specific formula is as follows: i where σ

[0034]

[0035] is the standard deviation of θ i , and e i is the j-th component of the eigenvector e ij . i

[0036] ④Finally, use the dynamic fusion method to calculate the result of the attitude change based on space, and the specific formula is as follows:

[0037]

[0038] where η i is the binary indicator variable of the i-th group of data, and its value is determined according to the specific working environment. When the amount of input data is small, let m = n. Conversely, if the amount of input data is large, take the cumulative contribution rate to reach the proportion suitable for the current environment for fusion.

[0039] Furthermore, in step S3, use the distance update function to describe the change trend of the distance from the robotic dog to the wall at different times, and then calculate the parameters of the attitude change, specifically including:

[0040] S31: Save the data in as, and then substitute each group of data into the distance update function to calculate the parameters and and save these parameters, where the calculation function is as follows:

[0041]

[0042] where w k-1 and v k-1 respectively represent the measurement noise of the lidar at the (k - 1)-th moment, where k ≥ 5; represents represents represents is the compensation parameter for the distance change from the (k - 3)-th moment to the (k - 2)-th moment in represents is the compensation parameter for the distance change from the (k - 4)-th moment to the (k - 3)-th moment in represents ​The compensation parameter for the distance change from the (k - 3)-th moment to the (k - 2)-th moment in denotes the compensation parameter for the distance change from the (k - 4)-th moment to the (k - 3)-th moment in

[0043] S32: Using the parameters at each moment and calculate the change rate of each group of scan lines at each moment. The calculation formula is as follows:

[0044]

[0045] where, △k represents the time difference before and after the k-th moment; denotes the first parameter of parameter a at the k-th moment, with the horizontal angle being α and the vertical angle being β; denotes the second parameter of parameter a at the k-th moment, with the horizontal angle being α and the vertical angle being β, denotes the first parameter of parameter b at the k-th moment, with the horizontal angle being α and the vertical angle being β; denotes the second parameter of parameter b at the k-th moment, with the horizontal angle being α and the vertical angle being β. Since the change rate of the parameters between two consecutive moments in continuous motion is not too large, the wire bundles with a smaller parameter change rate are retained, and the data with a larger change rate are excluded. Then, the average change rate of the parameters is calculated using the retained wire bundles. The calculation formula is as follows:

[0046]

[0047] where, and respectively represent the average change rates of parameters and . Then, substitute the average change rate into formula (9) to calculate and

[0048] S33: Substitute the and calculated in step S32 into formula (2) to calculate the attitude change parameters for different angles and different wire bundles. Then, based on the data fusion algorithm of principal component analysis, fuse the attitude change parameters θ calculated for different angles and wire bundles to finally obtain the attitude change parameters based on time The specific process is the same as step S23.

[0049] Furthermore, in the quadruped robot-aided attitude correction method and system for cable tunnel inspection described in step S4, it is characterized in that: step S4 is specifically:

[0050] Calculate the optimal attitude change parameters using the spatio-temporal adaptive fusion function The specific formula is as follows:

[0051]

[0052]

[0053] Among them, and are the optimal estimates based on the spatial and temporal attitude change parameters respectively, and σ1 and σ2 are the adaptive weights based on the spatial and temporal attitude changes respectively, and θ space and θ time represent the maximum values of the current spatial and temporal attitude changes respectively.

[0054] The beneficial effects of the present invention are as follows:

[0055] 1. The present invention only uses lidar to collect the distance data of different angles and beam bundles from the wall for assisting in correcting the attitude of the intelligent quadruped robot dog in a narrow tunnel, thus reducing the implementation costs such as installing an inertial navigation system, and at the same time avoiding the processing of huge data when using the point set registration algorithm.

[0056] 2. Compared with the existing auxiliary attitude correction methods, the present invention can measure the attitude change more accurately and stably by combining the distance update function and the principal component analysis method.

[0057] 3. The attitude change is fused with the adaptive weight, so that the attitude change situation can be better obtained according to the environment.

[0058] 4. The present invention uses the spatial and temporal detection of the attitude change of the intelligent quadruped robot dog, which can make up for the defect of the lidar scanning the non-wall distance, and improve the stability, accuracy and robustness of the detection result when scanning an obstacle. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to make the objectives, technical solutions and beneficial effects of the present invention clearer, the following drawings are provided for illustration:

[0060] Figure 1 is the main flow chart of the auxiliary attitude correction method described in the embodiment of the present invention;

[0061] Figure 2 is the installation position diagram of the lidar described in the embodiment of the invention;

[0062] Figure 3 is the flow chart of calculating the attitude change based on space and time described in the embodiment of the invention;

[0063] Figure 4 is the model diagram of scanning the narrow tunnel wall described in the embodiment of the invention. DETAILED DESCRIPTION

[0064] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0065] In this embodiment, a 16-line laser radar is selected as a data acquisition sensor, and an algorithm is written in the ROS system to implement a quadruped robot dog-assisted posture correction method and system for cable tunnel inspection.

[0066] In this embodiment, Figure 1 As shown, a method and system for correcting the posture of a quadruped robot dog assisted by a cable tunnel inspection includes the following steps:

[0067] S1: Use the laser radar 1 to collect different angles and distances from the beam to the wall at multiple times, and then pre-process the collected data to eliminate invalid data;

[0068] S2: Calculate the space-based attitude change parameters using the distances measured by different laser radar angles and different beams at the same time, and fuse the space-based attitude change parameters through the principal component analysis algorithm;

[0069] S3: Use the distance update function to describe the parameter change rate of the distance between the robot dog and the wall at multiple moments, then calculate the average change rate of the parameter, and then iterate the average change rate of the parameter and the distance to recalculate the distance, and use the recalculated distance to find the time-based posture change parameters, and finally use the principal component analysis algorithm to fuse the time-based posture change parameters;

[0070] S4: Use the spatiotemporal adaptive fusion method to fuse the posture change parameters based on time and space, and finally obtain the optimal posture change parameters of the robot dog to assist in correcting the posture.

[0071] In step S1, the laser radar 1 is located as follows Figure 2 As shown, it should be fixed and installed in the middle of the back of the intelligent robot dog 2, 0.5-0.7 meters, with a detection angle of 360 degrees and a resolution of 0.18 degrees. The distance to the wall is collected at a rate of 10 frames per second, and then invalid data that is particularly large or small in the detection distance data is eliminated. In narrow tunnel collection scenarios such as Figure 4 As shown, the specific data collection and processing include:

[0072] S11: The x-axis of the laser radar is the forward direction of the robot dog, the y-axis is perpendicular to the robot dog's body and the positive direction is from right to left, and the positive direction of the z-axis is from bottom to top, with the origin being o. Different angles refer to the horizontal plane with the vertical angle of the laser radar 0° and the y-axis as the bisector. Let oa 30° With ob 30° The angle is 30°, oa45° The included angle with ob 45° is 45°, oa 60° and ob 60° is 60°, oa 90° and ob 90° is 90°, oa 120° and ob 120° is 120°, and data is collected accordingly;

[0073] S12: Select wire harnesses at different vertical angles and different included angles according to the environment adaptive screening method, and collect data with the axis as the bisector;

[0074] S13: Screen and eliminate the data that does not conform to the narrow tunnel scenario in the data, such as eliminating the data with particularly large and particularly small distances.

[0075] In this embodiment, the calculation and fusion of the attitude change parameters at the same moment under different included angles and different wire harnesses in step S2 include:

[0076] S21: Use different included angles and different wire harnesses and (where α = 30°, 45°, 60°, 90°, 120°, β = 1°, 7°, 15°) to measure the intersection points with the wall and the distances at two small angles before and after each intersection point, and use the average distance of these 5 intersection points to replace and the distances to the wall. The calculation formula is as follows:

[0077]

[0078] Among them, and represent the straight-line distance with a vertical direction angle of β and an included angle between the lidar wire harness oa and ob of α. oa -2 and ob -2 represent the distances from the two small angles before the intersection point of the lidar scan line and the wall to the wall; oa -1 and ob -1 represent the distance from the previous small angle at the intersection point of the lidar scan line and the wall to the wall. oa and ob represent the intersection point distance of the midpoint between the lidar scan line and the wall; similarly, oa2 and ob2 respectively represent the distances at the two small angles after the intersection point of the lidar scan line and the wall, and oa1 and ob1 respectively represent the distances at the previous small angle at the intersection point of the lidar scan line and the wall.

[0079] S22: Take and Substitute into the Difference Over Summation (DOS) function to calculate the spatial-based attitude change parameters to measure the attitude change of the robotic dog. The specific calculation formula is as follows:

[0080]

[0081] Among them, θ represents the attitude change parameter of the robotic dog, which can reflect the attitude change of the robotic dog. If θ < 0, it means the head of the robotic dog is tilted to the left; if θ > 0, it means the head of the robotic dog is tilted to the right; if θ = 0, it means the robotic dog is parallel to the wall. The larger |θ| is, the greater the head deflection of the robotic dog is, and vice versa, the smaller the deflection is.

[0082] S23: Use the data fusion algorithm based on principal component analysis to fuse the attitude change parameters calculated from different angles and wire harnesses, so as to obtain the optimal spatial-based attitude change parameter estimation, as follows:

[0083] ①Regard the spatial-based attitude change parameters θ calculated from different angles and different wire harnesses as a whole Θ = (θ1, θ2,..., θ n ), where n is the number of data. Let the covariance matrix of Θ be D, and obtain the eigenvalues as λ1 ≥ λ2 ≥... ≥ λ n and the corresponding unitized orthogonal eigenvectors as e1, e2,..., e n , where the i-th principal component unitized orthogonal eigenvector is e i = (e 1i , e 2i ,..., e ni ). Therefore, the i-th principal component is:

[0084] f i = e i ′Θ = e 1i θ1 +... + e ni θ n (3)

[0085] ②Calculate the variance contribution rate i of the principal component f and the cumulative variance contribution rate of the first m principal components. The calculation formulas are as follows:

[0086]

[0087]

[0088] Among them, when the value of m makes the cumulative variance contribution rate reach a proportion suitable for the current situation, other principal components can be omitted.

[0089] ③ Calculate the comprehensive support degrees of different included angles and beam bundles. The specific formula is as follows:

[0090]

[0091] Among them, f i * is the principal component most relevant to each θ i , and ρ(θ i , f j ) is determined through the correlation coefficient between them. Select the f i with the largest value of ρ(θ j , fj) as the principal component most relevant to θ i . The specific formula is as follows:

[0092]

[0093] Among them, σ i is the standard deviation of θ i , and e ij is the j-th component of the eigenvector e i .

[0094] ④ Finally, use the dynamic fusion method to calculate the result of the attitude change based on space. The specific formula is as follows:

[0095]

[0096] Among them, η i is the binary indicator variable of the i-th group of data, and its value is determined according to the specific working environment. When the amount of input data is small, m = n can be set. Conversely, if the amount of input data is large, the cumulative contribution rate is taken to reach an appropriate proportion for the current environment for fusion.

[0097] In this embodiment, the use of the distance update function to depict the change trend of the distance from the robotic dog to the wall at different times in step S3, and then calculate the parameters of the attitude change specifically includes:

[0098] S31: Save the data within as (for example: 5s), and then substitute each group of data into the distance update function to calculate the parameters and and save these parameters. The calculation function is as follows:

[0099]

[0100] Among them, w k-1 and v k-1 respectively represent the measurement noise of the lidar at the (k - 1)-th moment, where k ≥ 5; represents represents Represents The compensation parameter for the distance change from the (k - 3)-th moment to the (k - 2)-th moment in Represents The compensation parameter for the distance change from the (k - 4)-th moment to the (k - 3)-th moment in Represents The compensation parameter for the distance change from the (k - 3)-th moment to the (k - 2)-th moment in Represents The compensation parameter for the distance change from the (k - 4)-th moment to the (k - 3)-th moment in

[0101] S32: Using the parameters at each moment And Calculate the change rate of each group of wire harnesses at each moment. The calculation formula is as follows:

[0102]

[0103] Where, △k represents the time difference before and after the k-th moment; Represents the first parameter of parameter a at the k-th moment, with the horizontal angle α and the vertical angle β; Represents the second parameter of parameter a at the k-th moment, with the horizontal angle α and the vertical angle β. Similarly, the meaning of can be known. Since the change rate of the parameters between two consecutive moments in continuous motion is not too large, the wire harnesses with a smaller parameter change rate are retained, the data with a larger change rate are removed, and then the average change rate of the parameters is calculated using the retained wire harnesses. The calculation formula is as follows:

[0104]

[0105] Where, And Represent the average change rates of parameters And respectively. Then substitute the average change rate into formula (9) to calculate And

[0106] S33: Substitute the And calculated in step S32 into formula (2) to calculate the attitude change parameters under different angles and different wire harnesses, and then fuse the attitude change parameters θ calculated for different angles and wire harnesses based on the data fusion algorithm of principal component analysis to finally obtain the attitude change parameters based on time The specific process is the same as step S23.

[0107] In this embodiment, in step S4, the optimal attitude change parameters are calculated using the spatio-temporal adaptive fusion function The specific formula is as follows:

[0108]

[0109]

[0110] Wherein, and are the optimal estimates based on the spatial and temporal attitude change parameters respectively, σ1 and σ2 are the adaptive weights of the attitude changes based on space and time respectively, and θ space and θ time represent the maximum values of the current attitude changes based on space and time respectively.

[0111] In this embodiment, a quadruped robot dog assisted attitude correction system for cable tunnel inspection includes a memory and a controller. A computer-readable program is stored in the memory. When the computer-readable program is called by the controller, it can execute the steps of the quadruped robot dog assisted attitude correction method and system for cable tunnel inspection as described in this embodiment. This method realizes the application of the intelligent quadruped robot dog 2 in narrow tunnel inspection. This method only uses the data collected by the lidar 1 to calculate the attitude change information of the intelligent quadruped robot dog 2, so as to assist the robot dog to correct its attitude, provide more information for the control system of the intelligent robot dog, enhance the obstacle avoidance ability, and enable it to safely and quickly pass through the obstacle 3.

[0112] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made in form and details without departing from the scope defined by the claims of the present invention.

Claims

1. A method for auxiliary attitude correction of a quadruped robot dog for cable tunnel inspection, characterized in that, It includes the following steps: S1: Use lidar to collect the distances from different angles and beam bundles to the wall at multiple moments, and then preprocess the collected data to eliminate invalid data; S2: Calculate the spatial attitude change parameters using the distances measured by the lidar at different angles and beam bundles at the same moment, and fuse the spatial attitude change parameters through the principal component analysis algorithm; S3: Use the distance update function to characterize the parameter change rate of the distance from the quadruped robot to the wall at multiple moments, calculate the average change rate of the parameters, then iteratively recalculate the distance with the average change rate of the parameters and the distance, use the recalculated distance to obtain the time-based attitude change parameters, and fuse the time-based attitude change parameters through the principal component analysis algorithm; S4: Use the spatio-temporal adaptive fusion method to fuse the time-based and space-based attitude change parameters to obtain the optimal attitude change parameters of the quadruped robot to assist in attitude correction.

2. The method for auxiliary attitude correction of a quadruped robot dog for cable tunnel inspection according to claim 1, characterized in that, In the step S1, the acquisition and processing of lidar data include: S11: Set the x-axis of the lidar as the forward driving direction of the quadruped robot, the y-axis as the direction perpendicular to the body of the quadruped robot with the right-to-left direction as the positive direction, and the z-axis with the bottom-to-top direction as the positive direction, and the origin as o; different angles refer to the horizontal plane with the lidar vertical angle of 0° and the y-axis as the bisector; let oa 30° and ob 30° have an included angle of 30°, oa 45° and ob 45° have an included angle of 45°, oa 60° and ob 60° have an included angle of 60°, oa 90° and ob 90° have an included angle of 90°, oa 120° and ob 120° have an included angle of 120°, and collect data accordingly; S12: Select beam bundles at different vertical angles and different angles according to the environment adaptive screening method, and collect data with the y-axis as the bisector; S13: Screen and eliminate the data that does not conform to the narrow tunnel scenario in the data.

3. The method for auxiliary attitude correction of a quadruped robot dog for cable tunnel inspection according to claim 2, characterized in that, In the step S2, calculating and fusing the attitude change parameters at different angles and beam bundles at the same moment includes: S21: Using and where α = 30°, 45°, 60°, 90°, 120° and β = 1°, 7°, 15°, measure the distances of the intersection points on the wall surface and the distances of two small angles before and after each intersection point. The small angle is 0.18°. Then use the distances of these 5 intersection points to calculate the average distance to replace and the distance to the wall surface. The calculation formula is as follows: Among them, and represent the straight-line distance with a vertical angle of β and an included angle of α between the lidar beams oa and ob. oa -2 and ob -2 represent the distances from the first two small angles at the intersection of the lidar scan line and the wall to the wall; oa -1 and ob -1 represent the distance from the previous small angle at the intersection of the lidar scan line and the wall to the wall. oa and ob represent the intersection distance at the midpoint between the lidar scan line and the wall; oa2 and ob2 respectively represent the distances of the last two small angles at the intersection of the lidar scan line and the wall, and oa1 and ob1 respectively represent the distance of the last small angle at the intersection of the lidar scan line and the wall. S22: Substitute and into the difference-sum-ratio function to calculate the spatial-based attitude change parameters for measuring the attitude change of the robotic dog. The specific calculation formula is as follows: Among them, θ represents the attitude change parameter of the quadruped robot, which can reflect the attitude change of the quadruped robot. If θ < 0, it means that the head of the quadruped robot is tilted to the left; if θ > 0, it means that the head of the quadruped robot is tilted to the right; if θ = 0, it means that the quadruped robot is parallel to the wall, and the larger |θ| is, the greater the head deflection of the quadruped robot, and vice versa, the smaller the deflection; S23: Use the data fusion algorithm based on principal component analysis to fuse the attitude change parameters calculated from different angles and beam bundles, so as to obtain the optimal estimated attitude change parameters based on space.

4. The method for auxiliary attitude correction of a quadruped robot dog for cable tunnel inspection according to claim 3, characterized in that: The step S23 is specifically: ①Regarding the space-based attitude change parameters θ calculated with different angles and different wire bundles as a whole Θ = (θ1, θ2,..., θ n ), where n is the number of data; let the covariance matrix of Θ be D, and the obtained eigenvalues be λ1 ≥ λ2 ≥... ≥ λ n and the corresponding unitized orthogonal eigenvectors be e1, e2,..., e n , where the unitized orthogonal eigenvector of the i-th principal component is e i = (e 1i , e 2i ,..., e ni ), so the i-th principal component is: f i = e i 'Θ = e 1i θ1 +... + e ni θ n (3) ② Calculate the variance contribution rate of the principal component f i and the cumulative variance contribution rate of the first m principal components The calculation formulas are as follows: The calculation formulas are as follows: ③ Calculate the comprehensive support degree of different angles and beam bundles, and the specific formula is as follows: Among them, f i * is the principal component most relevant to each θ i , and ρ(θ i , f j ) is determined by the correlation coefficient between them. Select the f i with the largest value of ρ(θ j , f j ) as the principal component most relevant to θ i . The specific formula is as follows: Among them, σ i is the standard deviation of θ i , and e ij is the j-th component of the eigenvector e i ; ④ Finally, use the dynamic fusion method to calculate the result of the attitude change situation based on space, and the specific formula is as follows: where η i is the binary indicator variable of the i-th group of data, and its value is determined according to the specific working environment. When the input data volume is small, let m = n. On the contrary, if the input data volume is large, the cumulative contribution rate is taken to reach a proportion suitable for the current environment for fusion.

5. The method for auxiliary attitude correction of a quadruped robot dog for cable tunnel inspection according to claim 4, characterized in that: In step S3, use the distance update function to characterize the change trend of the distance from the quadruped robot to the wall at different moments, and then calculate the parameters of the attitude change, specifically including: S31: Save the data in as, and then substitute the data of each group into the distance update function to calculate the parameters and Save these parameters, and the calculation function is as follows: where, w k-1 and v k-1 respectively represent the measurement noise of the lidar at the (k - 1)-th moment, where k ≥ 5; represents represents represents the compensation parameter for the distance change from the (k - 3)-th moment to the (k - 2)-th moment in represents the compensation parameter for the distance change from the (k - 4)-th moment to the (k - 3)-th moment in represents the compensation parameter for the distance change from the (k - 3)-th moment to the (k - 2)-th moment in represents the compensation parameter for the distance change from the (k - 4)-th moment to the (k - 3)-th moment in S32: Utilize the parameters at each moment and Calculate the change rate of each group of scan lines at each moment. The calculation formula is as follows: where Δk represents the time difference before and after time k; represents the first parameter of parameter a at time k with a horizontal angle of α and a vertical angle of β; represents the second parameter of parameter a at time k with a horizontal angle of α and a vertical angle of β, represents the first parameter of parameter b at time k with a horizontal angle of α and a vertical angle of β; represents the second parameter of parameter b at time k with a horizontal angle of α and a vertical angle of β; Since the change rate of parameters between two consecutive times in continuous motion is not too large, the wire harness with a smaller change rate of parameters is retained, the data with a larger change rate is eliminated, and then the average change rate of the parameters is calculated using the retained wire harness. The calculation formula is as follows: Among them, and respectively represent the average rates of change of the parameters and Substitute the average rate of change into formula (9) to calculate and S33: Substitute the and calculated in step S32 into formula (2) to calculate the attitude change parameters under different angles and different wire harnesses, and then fuse the attitude change parameters θ calculated for different angles and wire harnesses based on the data fusion algorithm of principal component analysis to finally obtain the attitude change parameter 6. The auxiliary attitude correction method for a quadruped robot dog for cable tunnel inspection according to claim 5, characterized in that, The step S4 is specifically: Calculate the optimal attitude change parameters using the spatio-temporal adaptive fusion function The specific formula is as follows: Among them, and are the optimal estimates based on the spatial and temporal attitude change parameters respectively, and σ1 and σ2 are the adaptive weights of the attitude change based on space and time respectively, and θ space and θ time represent the maximum values of the current attitude change based on space and time respectively.

7. An auxiliary attitude correction system for a quadruped robot dog for cable tunnel inspection, characterized in that: It includes a memory and a controller. The memory stores a computer-readable program. When the computer-readable program is called by the controller, it can execute the steps of the quadruped robot-assisted attitude correction method for cable tunnel inspection as described in any one of claims 1 to 6.

Images