Robot cable anti-interference wiring method and system based on reinforcement learning

Abstract

This invention provides a reinforcement learning-based method and system for robot cable anti-interference routing. This method involves obtaining the robot's three-dimensional structural data and electromagnetic field intensity distribution data, dividing the workspace into grid cells, determining the start and end coordinates and obstacle avoidance zones, generating candidate routing paths using a reinforcement learning algorithm, selecting the path with the minimum sum of electromagnetic field intensity as the target routing path, and installing fixtures at fixed points to complete the cable routing. This method effectively reduces electromagnetic interference and improves the robot's operational stability and reliability.

Description

Technical Field

[0001] The present invention relates to robot cable wiring technology, and in particular to a robot cable anti-interference wiring method and system based on reinforcement learning. Background Art

[0002] Robots are widely used in industrial production, and their operation is inevitably affected by various sources of electromagnetic interference. Cables in robotic systems carry out important functions such as signal transmission and power supply. In electromagnetic interference environments, they are prone to signal distortion, communication interruptions, and other faults, seriously affecting the robot's operational stability and control accuracy.

[0003] Currently, robot cable routing is primarily planned and installed based on manual experience, lacking systematic analysis and assessment of electromagnetic field distribution within the operating environment. Traditional routing methods often only consider factors like obstacle avoidance and cable length, failing to effectively avoid electromagnetic interference areas. This makes cables susceptible to interference during actual operation, impacting the reliability of the robot system.

[0004] With the increasingly complex electromagnetic environment in industrial sites and the continuous expansion of robot application scenarios, there is an urgent need for an intelligent wiring method that can automatically plan cable wiring paths and effectively avoid electromagnetic interference areas to improve the anti-interference ability and operational stability of the robot system. Summary of the Invention

[0005] The embodiments of the present invention provide a robot cable anti-interference wiring method and system based on reinforcement learning, which can solve the problems in the prior art.

[0006] A first aspect of an embodiment of the present invention provides a robot cable anti-interference wiring method based on reinforcement learning, comprising:

[0007] Acquire the three-dimensional structural data of the robot body, and at the same time collect electromagnetic field intensity data in the robot workspace to generate electromagnetic field intensity distribution data of the robot workspace;

[0008] The robot workspace is divided into multiple three-dimensional grid units, the electromagnetic field strength value of each three-dimensional grid unit is calculated according to the electromagnetic field strength distribution data, and the three-dimensional grid unit is obtained as the target interference area based on the electromagnetic field strength value;

[0009] The starting and ending coordinates of the cable routing are determined based on the robot's 3D structural data and mapped to 3D grid cells. The motion range coordinates of each joint during the robot's motion are calculated based on the robot's motion trajectory data, and the 3D grid cells where the motion range coordinates are located are marked as obstacle avoidance areas.

[0010] The target interference area and obstacle avoidance area are used as search constraints, and multiple candidate routing paths are generated in the three-dimensional grid cells through reinforcement learning algorithm.

[0011] The sum of the electromagnetic field strength values of the three-dimensional grid cells passed by each of the multiple candidate routing paths is calculated, and the candidate routing path with the smallest sum of the electromagnetic field strength values is selected as the target routing path; the coordinates of multiple fixed points are determined along the target routing path, fixtures are installed at the multiple fixed point coordinates, and the cable installation is completed along the target routing path.

[0012] In an optional embodiment,

[0013] Acquire the three-dimensional structural data of the robot body and collect the electromagnetic field intensity data in the robot workspace at the same time. Generate the electromagnetic field intensity distribution data of the robot workspace including:

[0014] Arranging characteristic marking strips on the surface of the robot body along the axis of the motion joint, wherein the characteristic marking strips include a periodic coding pattern; identifying the joint position and the connecting rod posture according to the spatial distribution characteristics of the coding pattern, and simultaneously collecting local electromagnetic field data through a probe in the coding pattern;

[0015] The robot is controlled to move along a preset trajectory, and the joint rotation angle is calculated based on the real-time position of the characteristic marking strip. When the overlapping fields of view of adjacent structured light cameras capture the same coded pattern, the robot's movement speed is adaptively adjusted according to the joint rotation angle, and the probe is triggered to collect electromagnetic field data.

[0016] The collected electromagnetic field data is spatially mapped according to the spatial position of the characteristic marker strips, and the mapped electromagnetic field data is interpolated and expanded based on the electromagnetic field propagation characteristics to obtain the electromagnetic field intensity distribution data of the working space.

[0017] In an optional embodiment,

[0018] The robot workspace is divided into multiple three-dimensional grid units, and the electromagnetic field strength value of each three-dimensional grid unit is calculated according to the electromagnetic field strength distribution data. The target interference area of the three-dimensional grid unit is obtained based on the electromagnetic field strength value.

[0019] Divide the robot workspace into three-dimensional grids and calculate the electromagnetic field intensity gradient of each three-dimensional grid unit; divide the three-dimensional grid units into densities based on the electromagnetic field intensity gradient to generate gradient adaptive grid units;

[0020] Decomposing the electromagnetic field intensity distribution data into three orthogonal components, performing interpolation calculation on each orthogonal component using a radial basis function, wherein the weight of the interpolation calculation is determined according to the electromagnetic field intensity gradient; applying a divergence constraint to the result of the interpolation calculation to obtain the electromagnetic field intensity value of each gradient adaptive grid cell;

[0021] Calculating the field strength characteristic parameters of the gradient adaptive grid unit based on the electromagnetic field strength, wherein the field strength characteristic parameters include the field strength amplitude and the directional change rate; performing wavelet transform on the field strength characteristic parameters to obtain characteristic components at different scales; establishing an interference degree judgment standard based on the characteristic components to determine the interference level of each gradient adaptive grid unit;

[0022] The electromagnetic field intensity value changes of the gradient adaptive grid cells are monitored in real time. When the electromagnetic field intensity value changes exceed the preset intensity value range, the corresponding gradient adaptive grid cells are marked as cells to be updated; only the density division, interpolation calculation and field strength characteristic parameter calculation are re-performed on the cells to be updated to update the interference level; and the target interference area is obtained based on the interference level of each updated gradient adaptive grid cell.

[0023] In an optional embodiment,

[0024] The starting and ending coordinates of the cable routing are determined based on the robot's 3D structural data and mapped to 3D grid cells. The range of motion coordinates of each joint during the robot's motion is calculated based on the robot's motion trajectory data. The 3D grid cells where the range of motion coordinates are located are marked as obstacle avoidance areas, including:

[0025] The robot's joint axis position data is obtained, and the endpoints of the joint axis are used as the starting and ending coordinates for cable routing. The initial routing curve of the cable is generated based on the starting and ending coordinates, and the tangential angle change rate of discrete points on the initial routing curve is calculated. The actual bending deformation path of the cable is determined based on the tangential angle change rate.

[0026] The actual bending deformation path is discretized into multiple path nodes, and the path nodes are mapped into a three-dimensional grid to obtain the wiring path grid unit; the robot's motion trajectory data is obtained, and the real-time motion angle of each joint is calculated based on the motion trajectory data. Preset motion redundant angles are added to the real-time motion angles;

[0027] The motion range coordinates of each joint are calculated based on the motion redundancy angle, and the motion range coordinates are mapped to the corresponding three-dimensional grid cells; a first type of safety buffer zone is constructed around the wiring path grid cells, and a second type of safety buffer zone is constructed around the three-dimensional grid cells corresponding to the motion range coordinates;

[0028] The grid cells are divided into obstacle avoidance levels according to the degree of overlap between the first and second safety buffer zones, and the three-dimensional grid cells whose obstacle avoidance levels are higher than the thresholds of the first and second safety buffer zones are marked as obstacle avoidance areas.

[0029] In an optional embodiment,

[0030] Using the target interference area and obstacle avoidance area as search constraints, a reinforcement learning algorithm is used to generate multiple candidate routing paths in the three-dimensional grid cells, including:

[0031] The octree data structure is used to divide the three-dimensional space into grids, and the grid density is adaptively adjusted according to the complexity of the area to obtain three-dimensional grid units;

[0032] Obtain the location information, interference area information, obstacle avoidance area information, and distance information from the starting and ending points of the three-dimensional grid unit;

[0033] Based on the acquired information, a reward function is constructed, including a target distance reward component, an interference area penalty component, and an obstacle avoidance area penalty component. The target distance reward component is used to guide the path to move toward the target point, while the interference area penalty component and the obstacle avoidance area penalty component generate negative reward values when the path passes through the corresponding area. A policy neural network is used to generate the probability of movement direction based on the three-dimensional grid cell information.

[0034] A temperature adjustment parameter is set on the moving direction probability output by the strategy neural network. The moving direction is sampled probabilistically by adjusting the temperature adjustment parameter to generate multiple initial wiring paths. The multiple initial wiring paths are subjected to repetitive analysis and screening to generate multiple candidate wiring paths.

[0035] In an optional embodiment,

[0036] Calculating the sum of the electromagnetic field strength values of the three-dimensional grid cells passed by each routing path among multiple candidate routing paths, and selecting the candidate routing path with the smallest sum of the electromagnetic field strength values as the target routing path includes:

[0037] Dividing the three-dimensional space into a plurality of grid cells, collecting electromagnetic field information of each grid cell, and extracting electric field components in three directions based on the electromagnetic field information; performing vector synthesis operations on the electric field components to obtain a comprehensive electromagnetic field strength of each grid cell;

[0038] An electromagnetic field distribution map is established based on the comprehensive electromagnetic field strength, and multiple candidate wiring paths are tracked using a grid scanning method. During the path tracking process, the comprehensive electromagnetic field strength of the grid cells passed by each candidate wiring path is accumulated to obtain the cumulative electromagnetic strength value of the path.

[0039] An exponential weighting method is used to calculate the comprehensive score of each candidate wiring path, where the comprehensive score is determined by the cumulative value of the path electromagnetic intensity and the path length; and the wiring path with the best comprehensive score is selected as the target wiring path.

[0040] In an optional embodiment,

[0041] Determining multiple fixed point coordinates along a target wiring path, installing fixtures at the multiple fixed point coordinates, and completing cable installation along the target wiring path includes:

[0042] Perform force analysis along the target wiring path to calculate the gravity distribution generated by the cable's own weight and the elastic stress distribution caused by bending deformation. Based on the gravity distribution and elastic stress distribution, use the superposition principle to calculate the comprehensive stress distribution at each point on the path. Determine the maximum stress threshold between fixed points based on the material's allowable stress.

[0043] Using a stress iteration calculation method, the maximum allowable spacing between adjacent fixed points is calculated under the condition that a maximum stress threshold constraint is satisfied. Curvature analysis is performed on the target wiring path, and at turns where the curvature exceeds a preset curvature threshold, the maximum allowable spacing is dynamically adjusted according to the curvature proportional coefficient.

[0044] Based on the adjusted maximum allowable spacing, the initial positions of the fixed points are marked on the target routing path using an evenly spaced distribution method. The path tangent vector and normal vector at the initial position of each fixed point are calculated. The initial position of the fixed point is optimized based on the tangent vector and normal vector to obtain the final position of the fixed point.

[0045] According to the spatial characteristics and force characteristics of the final position of the fixed point, the fixing device is divided into a straight segment fixing fixture, a corner reinforcement fixture and an end fixing device; the straight segment fixing fixture adopts a standard clamping force, the corner reinforcement fixture provides additional radial support force, and the end fixing device has an anti-falling mechanism; the installation direction of each fixing device is determined based on the path tangent vector, and the fixed point installation is completed.

[0046] A second aspect of an embodiment of the present invention provides a robot cable anti-interference wiring system based on reinforcement learning, comprising:

[0047] The first unit is used to obtain the three-dimensional structural data of the robot body and simultaneously collect the electromagnetic field intensity data in the robot workspace to generate the electromagnetic field intensity distribution data of the robot workspace;

[0048] The second unit is used to divide the robot workspace into multiple three-dimensional grid units, calculate the electromagnetic field strength value of each three-dimensional grid unit according to the electromagnetic field strength distribution data, and obtain the target interference area of the three-dimensional grid unit based on the electromagnetic field strength value;

[0049] The third unit is used to determine the starting and ending coordinates of the cable routing based on the three-dimensional structural data of the robot body and map them to the three-dimensional grid cells. The third unit is used to calculate the motion range coordinates of each joint of the robot during the motion process based on the robot motion trajectory data, and mark the three-dimensional grid cells where the motion range coordinates are located as obstacle avoidance areas.

[0050] The fourth unit is used to generate multiple candidate routing paths in the three-dimensional grid unit through a reinforcement learning algorithm using the target interference area and obstacle avoidance area as search constraints;

[0051] The fifth unit is used to calculate the sum of the electromagnetic field strength values of the three-dimensional grid cells passed by each of the multiple candidate wiring paths, select the candidate wiring path with the smallest sum of the electromagnetic field strength values as the target wiring path; determine the coordinates of multiple fixed points along the target wiring path, install fixtures at the coordinates of the multiple fixed points, and complete the cable installation along the target wiring path.

[0052] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0053] processor;

[0054] a memory for storing processor-executable instructions;

[0055] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0056] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0057] In this embodiment, by collecting electromagnetic field strength data within the workspace and establishing a three-dimensional grid model, electromagnetic interference areas can be accurately identified and quantified, providing reliable environmental information for cable routing. By using the robot's range of motion and electromagnetic interference areas as constraints and combining them with a reinforcement learning algorithm to automatically generate multiple candidate routing paths, it is possible to effectively avoid areas of high electromagnetic interference and robot motion areas, ensuring the feasibility and safety of the routing solution. By using the sum of electromagnetic field strength values as a path evaluation metric and selecting the path with the least electromagnetic interference as the final routing solution, the electromagnetic interference experienced by the cable can be significantly reduced, improving signal transmission quality and system stability. Furthermore, by automatically calculating fixed point coordinates and planning routing paths, the difficulty and workload of manual routing can be reduced, while improving routing efficiency and standardization. This achieves intelligent and automated robotic cable routing, enabling more scientific and efficient routing tasks compared to traditional manual experience-based routing methods. This effectively addresses the technical issue of robotic cables being susceptible to interference in complex electromagnetic environments at industrial sites, and improves the overall performance and reliability of the robotic system. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 Schematic diagram of the flow of a robot cable anti-interference wiring method based on reinforcement learning according to an embodiment of the present invention;

[0059] Figure 2 This is a comparative analysis diagram of reward function components according to an embodiment of the present invention;

[0060] Figure 3 This is a histogram of the cumulative distribution of electromagnetic intensity along the path of an embodiment of the present invention. DETAILED DESCRIPTION

[0061] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0062] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0063] Figure 1 FIG. 1 is a flow chart of a robot cable anti-interference wiring method based on reinforcement learning according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0064] Acquire the three-dimensional structural data of the robot body, and at the same time collect electromagnetic field intensity data in the robot workspace to generate electromagnetic field intensity distribution data of the robot workspace;

[0065] The robot workspace is divided into multiple three-dimensional grid units, the electromagnetic field strength value of each three-dimensional grid unit is calculated according to the electromagnetic field strength distribution data, and the three-dimensional grid unit is obtained as the target interference area based on the electromagnetic field strength value;

[0066] The starting and ending coordinates of the cable routing are determined based on the robot's 3D structural data and mapped to 3D grid cells. The motion range coordinates of each joint during the robot's motion are calculated based on the robot's motion trajectory data, and the 3D grid cells where the motion range coordinates are located are marked as obstacle avoidance areas.

[0067] The target interference area and obstacle avoidance area are used as search constraints, and multiple candidate routing paths are generated in the three-dimensional grid cells through reinforcement learning algorithm.

[0068] The sum of the electromagnetic field strength values of the three-dimensional grid cells passed by each of the multiple candidate routing paths is calculated, and the candidate routing path with the smallest sum of the electromagnetic field strength values is selected as the target routing path; the coordinates of multiple fixed points are determined along the target routing path, fixtures are installed at the multiple fixed point coordinates, and the cable installation is completed along the target routing path.

[0069] In an optional embodiment,

[0070] Acquire the three-dimensional structural data of the robot body and collect the electromagnetic field intensity data in the robot workspace at the same time. Generate the electromagnetic field intensity distribution data of the robot workspace including:

[0071] Arranging characteristic marking strips on the surface of the robot body along the axis of the motion joint, wherein the characteristic marking strips include a periodic coding pattern; identifying the joint position and the connecting rod posture according to the spatial distribution characteristics of the coding pattern, and simultaneously collecting local electromagnetic field data through a probe in the coding pattern;

[0072] The robot is controlled to move along a preset trajectory, and the joint rotation angle is calculated based on the real-time position of the characteristic marking strip. When the overlapping fields of view of adjacent structured light cameras capture the same coded pattern, the robot's movement speed is adaptively adjusted according to the joint rotation angle, and the probe is triggered to collect electromagnetic field data.

[0073] The collected electromagnetic field data is spatially mapped according to the spatial position of the characteristic marker strips, and the mapped electromagnetic field data is interpolated and expanded based on the electromagnetic field propagation characteristics to obtain the electromagnetic field intensity distribution data of the working space.

[0074] The present invention provides a method for obtaining three-dimensional structural data of a robot body and collecting electromagnetic field intensity distribution data. During implementation, characteristic marking strips are first arranged on the surface of the robot body along the axis of the motion joint. These characteristic marking strips contain a periodic coding pattern, which is composed of black and white stripes with a stripe width of 5mm and a spacing of 3mm between adjacent stripes. Each coding pattern contains a 32-bit binary code, which is used to uniquely identify different parts of the robot. The characteristic marking strips are made of flexible material with a thickness of 0.8mm and can be attached to the surface of the robot joint. Inside the strips, a miniature electromagnetic field probe is arranged every 50mm. The probe has a diameter of 2mm, a height of 1mm, a sampling frequency of 100Hz, and a measurement range of 0.1μT to 100μT.

[0075] Eight structured light cameras are positioned around the robot, with a resolution of 1920×1080 pixels, a frame rate of 60 fps, and a field of view of 85 degrees. The overlap between adjacent cameras' fields of view accounts for 30% of the total field of view. The cameras are connected to the central processing unit via Gigabit Ethernet, with a transmission latency of less than 10ms. The structured light emitted by the structured light cameras has a wavelength of 850nm, which is invisible to the human eye and does not interfere with normal operating conditions.

[0076] Before the robot begins movement, the system first calibrates the characteristic marker strips. During calibration, the robot's joints are controlled to move to five preset positions, remaining at each position for two seconds. A structured light camera captures images of the characteristic marker strips. The system uses an image processing algorithm to extract the coded pattern and identify the joint positions and linkage postures. This image processing utilizes an adaptive binarization method, with a threshold set at 85% of the average grayscale value of the local area. After extracting the coded pattern, the system establishes a mapping relationship between the coded pattern and the robot's joint positions, with a mapping accuracy of better than 0.5 mm.

[0077] As the robot moves along a preset trajectory, the system captures the position of the signature marking strips in real time. The preset trajectory consists of a series of path points, with a distance of 50 mm between adjacent path points. The system calculates the joint rotation angle based on the real-time position of the signature marking strips. This calculation method tracks the displacement changes of the coding pattern and combines it with the robot's kinematic model to obtain the joint rotation angle. The system records the angle data whenever the joint rotation angle changes by more than 5 degrees.

[0078] When the overlapping fields of view of adjacent structured light cameras capture the same coded pattern, the system adaptively adjusts the robot's movement speed based on the joint rotation angle. The specific adjustment method is: when the joint rotation angular velocity exceeds 30 degrees / second, the robot's movement speed is reduced to 70% of the original speed; when the joint rotation angular velocity is less than 10 degrees / second, the robot's movement speed is increased to 120% of the original speed, but not exceeding 500mm / second. The speed adjustment adopts a smooth transition method, and the acceleration is limited to 200mm / s. 2 Within.

[0079] While adjusting speed, the system triggers the probes in the signature strips to collect electromagnetic field data. Each probe collects triaxial electromagnetic field intensity data at a sampling rate of 100Hz. This data is transmitted wirelessly to the central processing unit via a microtransmission module built into the signature strips at 2.4GHz, with a transmission delay of less than 5ms. To reduce electromagnetic interference, the transmission power is kept below 5mW.

[0080] The system spatially maps the collected electromagnetic field data based on the spatial positions of the signature strips. During this mapping process, the probe's precise position in the global coordinate system is determined with an accuracy of better than 1mm. The collected electromagnetic field data is then correlated with the probe's position to form a discrete point cloud of the electromagnetic field intensity distribution. The point cloud data density reaches 20 sampling points per cubic decimeter near the robot joints, decreasing to 5 sampling points per cubic decimeter away from the joints.

[0081] Based on the electromagnetic field propagation characteristics, the mapped electromagnetic field data is interpolated and expanded to obtain the electromagnetic field intensity distribution data in the workspace. The interpolation method uses radial basis functions, with a Gaussian kernel selected as the interpolation kernel and a kernel width parameter set to 30 mm. Interpolation calculations are performed on a three-dimensional grid with a mesh resolution of 10 mm × 10 mm × 10 mm. For areas with drastic changes in electromagnetic field intensity, the system adaptively refines the mesh resolution to 5 mm × 5 mm × 5 mm. The interpolation calculations take into account the attenuation characteristics of the electromagnetic field. For areas farther from the probe, the interpolation weight decays inversely with the square of the distance.

[0082] The resulting electromagnetic field intensity distribution data is stored as a three-dimensional voxel grid, with each voxel containing triaxial electromagnetic field intensity values and their standard deviations. Data files are formatted in HDF5, supporting hierarchical storage and fast retrieval. The system also provides data visualization capabilities, generating electromagnetic field intensity isosurfaces and slice plots with a color mapping range of 0.1 μT to 100 μT, using a rainbow color spectrum to represent different intensity values. The visualization results have a resolution of up to 1920 × 1080 pixels and a refresh rate of 30 Hz, enabling real-time interactive exploration of electromagnetic field distribution characteristics.

[0083] In this embodiment, a design scheme of integrating electromagnetic field probes with feature marker strips is adopted to achieve the integration of robot structural feature recognition and electromagnetic field data acquisition, avoiding the interference and installation difficulty caused by the separate arrangement of sensors, and improving the efficiency and accuracy of data acquisition. By using periodic coding patterns in the strips, the joint position and connecting rod posture can be accurately identified, providing a reliable reference for the spatial positioning of electromagnetic field data. Combined with a multi-eye vision system and an adaptive speed control strategy, the system can dynamically adjust the robot's motion state according to the actual scene, optimizing the acquisition efficiency while ensuring sampling accuracy. In particular, in the area where the cameras overlap, the acquisition quality of electromagnetic field data at key locations is ensured by reducing the movement speed. An adaptive interpolation algorithm based on radial basis functions is adopted, which fully considers the propagation characteristics of the electromagnetic field and achieves high-precision mapping from discrete sampling points to continuous distribution.

[0084] In an optional embodiment,

[0085] The robot workspace is divided into multiple three-dimensional grid units, and the electromagnetic field strength value of each three-dimensional grid unit is calculated according to the electromagnetic field strength distribution data. The target interference area of the three-dimensional grid unit is obtained based on the electromagnetic field strength value.

[0086] Divide the robot workspace into three-dimensional grids and calculate the electromagnetic field intensity gradient of each three-dimensional grid unit; divide the three-dimensional grid units into densities based on the electromagnetic field intensity gradient to generate gradient adaptive grid units;

[0087] Decomposing the electromagnetic field intensity distribution data into three orthogonal components, performing interpolation calculation on each orthogonal component using a radial basis function, wherein the weight of the interpolation calculation is determined according to the electromagnetic field intensity gradient; applying a divergence constraint to the result of the interpolation calculation to obtain the electromagnetic field intensity value of each gradient adaptive grid cell;

[0088] Calculating the field strength characteristic parameters of the gradient adaptive grid unit based on the electromagnetic field strength, wherein the field strength characteristic parameters include the field strength amplitude and the directional change rate; performing wavelet transform on the field strength characteristic parameters to obtain characteristic components at different scales; establishing an interference degree judgment standard based on the characteristic components to determine the interference level of each gradient adaptive grid unit;

[0089] The electromagnetic field intensity value changes of the gradient adaptive grid cells are monitored in real time. When the electromagnetic field intensity value changes exceed the preset intensity value range, the corresponding gradient adaptive grid cells are marked as cells to be updated; only the density division, interpolation calculation and field strength characteristic parameter calculation are re-performed on the cells to be updated to update the interference level; and the target interference area is obtained based on the interference level of each updated gradient adaptive grid cell.

[0090] Exemplarily, when the robot workspace is divided into three-dimensional grids, an adaptive grid division strategy can be adopted. Specifically, the workspace is initially divided into initial grids with uniform intervals (such as 10 cm). The electromagnetic field intensity gradient at each grid point is calculated by the finite difference method, and the gradient calculation is obtained by dividing the difference in field strength values of adjacent grid points by the grid spacing. For example, for a grid point at position (x, y, z), its x-direction gradient can be calculated by [E(x+Δx, y, z)-E(x-Δx, y, z)] / (2Δx), where E represents the electromagnetic field intensity and Δx is the grid spacing in the x-direction.

[0091] Based on the calculated gradient values, the grid is adaptively refined. In areas where the gradient value exceeds a preset threshold (e.g., 0.5 V / m²), the original grid is subdivided into smaller subgrids (e.g., a 10 cm interval is refined to a 2 cm interval). In areas with smaller gradient values (e.g., less than 0.1 V / m²), the original grid size is maintained or increased appropriately (e.g., to 15 cm). This adaptive meshing strategy ensures a finer grid distribution in areas with drastic electromagnetic field fluctuations, improving computational accuracy while optimizing computing resources.

[0092] The electromagnetic field intensity distribution data is decomposed into three orthogonal components, Ex, Ey, and Ez, corresponding to the electric field components in the x, y, and z directions, respectively. The radial basis function is used to interpolate each component. In the specific implementation, the Gaussian radial basis function is selected as the interpolation kernel function. For any position p, its electric field component E(p) can be calculated by the weighted sum of the surrounding known sampling points pi. The weight coefficient wi is proportional to the electromagnetic field gradient. The larger the gradient, the higher the weight. For example, when the gradient value of a certain area is 1.2V / m 2 When the corresponding weight can be set to 0.8; and the gradient value is 0.3V / m 2 The weight can be set to 0.2.

[0093] To ensure the physical rationality of the interpolation results, a divergence constraint is imposed on the interpolation calculation results. In the passive area, the electric field divergence should be zero. By iteratively adjusting the interpolation results, the divergence error of each grid cell is controlled within 10^-6V / m 2 For example, for a grid cell with a volume of 8 cubic centimeters, the sum of the electric field fluxes on its six faces should be close to zero. If the initial calculation result is 0.002V, this error is distributed to the six faces through proportional distribution for correction.

[0094] Based on the corrected electromagnetic field strength values, the field strength characteristic parameters of each gradient adaptive grid cell are calculated. The field strength amplitude is calculated as the square root of the sum of the squares of the three orthogonal components. For example, if the three components of a grid cell are Ex = 2V / m, Ey = 1.5V / m, and Ez = 1V / m, then its field strength amplitude is 2.69V / m. The directional change rate is calculated by dividing the angle between the field strength direction vectors of adjacent grid cells by the grid spacing. For example, if the angle between the field strength directions of two adjacent cells is 15 degrees and the grid spacing is 5 cm, then the directional change rate is 3 degrees / cm.

[0095] The field intensity characteristic parameters were subjected to wavelet transformation and decomposed into three levels using Daubechies wavelet basis functions to obtain characteristic components at different scales. The first level of decomposition captures high-frequency variations (such as field intensity variations at the 5-10 cm scale), the second level captures medium-frequency variations (10-20 cm scale), and the third level captures low-frequency variations (20-40 cm scale).

[0096] A standard for determining interference severity was established based on the characteristic components after wavelet transformation. A high-frequency component coefficient with an absolute value greater than 0.8 indicates strong interference, a coefficient between 0.5 and 0.8 indicates moderate interference, and a coefficient less than 0.5 indicates weak interference. Similar thresholds were used for mid- and low-frequency components (mid-frequency: 0.7 and 0.4; low-frequency: 0.6 and 0.3). Combining the results of the three scales, grid cells were divided into four levels of interference: severe interference (red), moderate interference (orange), mild interference (yellow), and negligible interference (green).

[0097] During real-time monitoring, electromagnetic field data is collected every 100 milliseconds. For each grid cell, the difference between the current electromagnetic field strength and the previous value is compared. If the difference exceeds a preset threshold (such as 0.5 V / m or a relative change of 20%), the cell is marked as a cell to be updated. For example, if the field strength of a cell was 3.2 V / m at the previous moment and the current measurement is 4.0 V / m, the rate of change is 25%, exceeding the threshold and marking it as a cell to be updated.

[0098] Only the cells to be updated are reprocessed, including: re-dividing the grid density (for example, a grid with an original 5 cm spacing may be refined to 2 cm); performing radial basis function interpolation calculations using the latest collected data; updating the field strength characteristic parameters and wavelet transform results; and re-evaluating the interference level. This incremental update strategy significantly improves system real-time performance. In a test environment, the calculation time was reduced from 120 milliseconds to 35 milliseconds compared to a global update.

[0099] Based on the updated interference levels of each gradient-adaptive grid cell, regions with severe and moderate interference are merged into target interference regions. This merging process uses a region growing algorithm, starting with the severely interfered cells and gradually incorporating adjacent moderately interfered cells to form a continuous interference region. The resulting target interference region is represented as a 3D geometry that can be directly used in robot path planning, allowing the robot to avoid these areas and maintain normal operation.

[0100] The above method achieves accurate representation of the electromagnetic field distribution in the robot's workspace and dynamic identification of interference areas. A gradient-adaptive meshing strategy improves space utilization while ensuring computational accuracy. Combining multi-scale analysis with fuzzy rule-based judgment, an objective interference assessment standard is established. Real-time monitoring and local update mechanisms ensure the system's rapid response to changes in the electromagnetic environment.

[0101] In an optional embodiment,

[0102] The starting and ending coordinates of the cable routing are determined based on the robot's 3D structural data and mapped to 3D grid cells. The range of motion coordinates of each joint during the robot's motion is calculated based on the robot's motion trajectory data. The 3D grid cells where the range of motion coordinates are located are marked as obstacle avoidance areas, including:

[0103] The robot's joint axis position data is obtained, and the endpoints of the joint axis are used as the starting and ending coordinates for cable routing. The initial routing curve of the cable is generated based on the starting and ending coordinates, and the tangential angle change rate of discrete points on the initial routing curve is calculated. The actual bending deformation path of the cable is determined based on the tangential angle change rate.

[0104] The actual bending deformation path is discretized into multiple path nodes, and the path nodes are mapped into a three-dimensional grid to obtain the wiring path grid unit; the robot's motion trajectory data is obtained, and the real-time motion angle of each joint is calculated based on the motion trajectory data. Preset motion redundant angles are added to the real-time motion angles;

[0105] The motion range coordinates of each joint are calculated based on the motion redundancy angle, and the motion range coordinates are mapped to the corresponding three-dimensional grid cells; a first type of safety buffer zone is constructed around the wiring path grid cells, and a second type of safety buffer zone is constructed around the three-dimensional grid cells corresponding to the motion range coordinates;

[0106] The grid cells are divided into obstacle avoidance levels according to the degree of overlap between the first and second safety buffer zones, and the three-dimensional grid cells whose obstacle avoidance levels are higher than the thresholds of the first and second safety buffer zones are marked as obstacle avoidance areas.

[0107] In this embodiment, the starting point coordinates and the end point coordinates of the cable wiring are determined according to the three-dimensional structural data of the robot body and mapped to the three-dimensional grid unit. The motion range coordinates of each joint of the robot during the movement are calculated according to the robot motion trajectory data. The method of marking the three-dimensional grid unit where the motion range coordinates are located as the obstacle avoidance area is as follows.

[0108] Obtain the robot's joint axis position data and use the endpoints of the joint axes as the starting and ending coordinates for cable routing. For example, for a six-axis industrial robot, the starting coordinates of its third joint axis are (100, 150, 200) mm, and the ending coordinates are (100, 150, 300) mm. These two coordinates can be used as the starting and ending points for cable routing, respectively. Based on the obtained starting and ending coordinates, generate the initial cable routing curve. This initial routing curve can be generated using the cubic spline interpolation method. Multiple control points are selected between the starting and ending points. For example, in the above example, (100, 160, 220), (110, 170, 250), and (105, 160, 280) can be selected as control points. These points are used to generate a smooth initial routing curve.

[0109] Calculate the rate of change of the tangential angle at discrete points on the initial routing curve and use this rate to determine the actual bending deformation path of the cable. Specifically, select discrete points every 5 mm on the initial routing curve and calculate the change in the tangential angle between two adjacent points. If the rate of change exceeds the maximum bend angle allowed by the cable material (e.g., no more than 30 degrees per 10 mm), adjust that curve segment to conform to the physical bending characteristics of the cable. The resulting curve represents the actual bending deformation path of the cable.

[0110] The actual bending deformation path is discretized into multiple path nodes. These path nodes are then mapped to a three-dimensional grid to obtain routing path grid cells. For example, the space is divided into a cubic grid with a side length of 10 mm. For each discrete point on the actual bending deformation path, the grid cell index is calculated. For example, the coordinates (103, 158, 225) correspond to the grid cell index (10, 15, 22). The set of grid cells after all path nodes are mapped is recorded as the routing path grid cell.

[0111] The robot's motion trajectory data is obtained and the real-time motion angle of each joint is calculated based on this data. For example, the motion trajectory data for the third joint indicates a 45-degree angle at a certain moment. Preset redundant motion angles are added to the real-time motion angles. For example, by adding ±10 degrees of redundant angles to each joint, the motion angle range of the third joint is 35 to 55 degrees.

[0112] The range of motion coordinates of each joint are calculated based on the motion redundancy angle. The joint angles are converted into spatial coordinates using the robot's forward kinematics model. For example, when the third joint angle is 35 degrees, the coordinates of its end effector are (250, 300, 150) mm; when the angle is 55 degrees, the coordinates are (270, 320, 170) mm. The spatial area covered by these coordinate points and the lines connecting them is the range of motion of the joint. The range of motion coordinates are mapped to the corresponding three-dimensional grid cells. For example, the grid cell index corresponding to the coordinates (250, 300, 150) is (25, 30, 15).

[0113] A first-class safety buffer is constructed around the routing path grid cells. For each routing path grid cell, the surrounding area with a radius of 3 grid cells is marked as a first-class safety buffer. For example, for grid cell (10, 15, 22), the grid cells with indexes in the range (7-13, 12-18, 19-25) are marked as a first-class safety buffer. A second-class safety buffer is constructed around the three-dimensional grid cells corresponding to the motion range coordinates. For each motion range grid cell, the surrounding area with a radius of 2 grid cells is marked as a second-class safety buffer.

[0114] Grid cells are classified into obstacle avoidance levels based on the degree of overlap between the first and second safety buffer zones. A scoring system for overlap is defined: if a grid cell belongs only to the first safety buffer zone, it receives a score of 1; if it belongs only to the second safety buffer zone, it receives a score of 2; and if it belongs to both safety buffer zones, it receives a score of 3. Three-dimensional grid cells with an obstacle avoidance level above a preset threshold are marked as obstacle avoidance areas. For example, if the threshold is set to 2, grid cells with scores of 2 and 3 are marked as obstacle avoidance areas.

[0115] In practical applications, parameters can be adjusted based on the specific robot structure and cable characteristics. For example, for highly flexible cables, the radius of the safety buffer zone can be reduced; for applications requiring high precision, the grid cell size can be reduced to improve accuracy. This method effectively determines the obstacle avoidance zone for robot cable routing, providing a foundation for subsequent cable routing path planning.

[0116] In a specific example, a six-axis robot requires routing between the second and third joints. The starting coordinates are (50, 100, 150) mm, and the ending coordinates are (200, 100, 150) mm. An initial routing curve is generated using cubic spline interpolation, with discrete points taken every 8 mm, for a total of 20 points. Calculations show that the tangential angle change rate at 64 mm from the starting point is 35 degrees per 10 mm, exceeding the maximum allowable bending angle of 30 degrees per 10 mm for the cable. Therefore, this curve segment is adjusted to conform to the physical characteristics of the cable. The actual bending deformation path after adjustment is discretized into 25 path nodes and mapped onto a 10 mm edge 3D grid, resulting in a routing path grid cell set. The robot's motion trajectory data shows that the second joint's angle range is 20 to 60 degrees. After adding a ±8 degree margin, the range of motion is 12 to 68 degrees. Forward kinematics is used to calculate the range of motion within this angle range and map it onto the 3D grid. A first-class safety buffer zone with a radius of 2 grid cells is constructed around the routing path grid cells, and a second-class safety buffer zone with a radius of 3 grid cells is constructed around the motion range grid cells. By setting the obstacle avoidance level threshold to 2, 120 obstacle avoidance area grid cells are ultimately marked, providing clear obstacle avoidance guidance for cable routing.

[0117] In this embodiment, by taking the endpoints of the robot joint axis as the starting and ending coordinates of the cable wiring, and combining the actual bending characteristics of the cable to generate the initial wiring path, the feasibility and rationality of the wiring scheme can be ensured. The path discretization and grid mapping methods are used to achieve accurate spatial positioning of the wiring path. By adding motion redundancy angles based on the actual motion angles, the fault tolerance and safety margin of the system are improved. Combining the construction methods of the two types of safety buffers, it not only ensures the safety distance requirements for cable wiring, but also meets the spatial requirements of robot movement. An obstacle avoidance level division mechanism based on the degree of buffer overlap has been established, and a scientific obstacle avoidance judgment standard has been established, which can effectively identify potential interference risk areas. By comprehensively considering the physical properties of the cable and the motion characteristics of the robot, this scheme significantly improves the accuracy and reliability of wiring planning and reduces the risk of interference between the cable and the moving parts of the robot.

[0118] In an optional embodiment,

[0119] Using the target interference area and obstacle avoidance area as search constraints, a reinforcement learning algorithm is used to generate multiple candidate routing paths in the three-dimensional grid cells, including:

[0120] The octree data structure is used to divide the three-dimensional space into grids, and the grid density is adaptively adjusted according to the complexity of the area to obtain three-dimensional grid units;

[0121] Obtain the location information, interference area information, obstacle avoidance area information, and distance information from the starting and ending points of the three-dimensional grid unit;

[0122] Based on the acquired information, a reward function is constructed, including a target distance reward component, an interference area penalty component, and an obstacle avoidance area penalty component. The target distance reward component is used to guide the path to move toward the target point, while the interference area penalty component and the obstacle avoidance area penalty component generate negative reward values when the path passes through the corresponding area. A policy neural network is used to generate the probability of movement direction based on the three-dimensional grid cell information.

[0123] A temperature adjustment parameter is set on the moving direction probability output by the strategy neural network. The moving direction is sampled probabilistically by adjusting the temperature adjustment parameter to generate multiple initial wiring paths. The multiple initial wiring paths are subjected to repetitive analysis and screening to generate multiple candidate wiring paths.

[0124] For example, the octree data structure is first used to grid the three-dimensional space. Octree is a tree-like data structure in which each node can have eight child nodes, which is suitable for representing three-dimensional space. In actual applications, the system adaptively adjusts the grid density according to the complexity of the area. Specifically, for parts of the space where obstacles are dense or interference areas are complex, the system divides the grid more finely, and the grid size can be as small as 0.1 meters; while for open areas, the grid size can be relaxed to 0.5 meters or larger to reduce the amount of calculation. For example, in a space of 10 meters × 10 meters × 5 meters, the system may use a grid size of 0.2 meters in areas close to electronic devices and a grid size of 0.8 meters in open areas. This adaptive grid division method can significantly reduce the computational complexity while ensuring path accuracy. Experiments have shown that the computational efficiency is improved by about 65%.

[0125] After the grid division is completed, the system obtains the key information of the three-dimensional grid unit, including location information, interference area information, obstacle avoidance area information, and distance information from the starting and ending points. The location information is represented by three-dimensional coordinates (x, y, z), and the center point of each grid unit is used as its location identifier. The interference area information is represented by a binary flag bit, where a value of 1 indicates that the grid unit is located in the interference area, and a value of 0 indicates that it is not in the interference area. Similarly, the obstacle avoidance area information is also represented by a binary flag bit. The Euclidean distance is calculated for the distance information from the starting point and the end point respectively, which is used as a reference for subsequent path planning. For example, for a grid unit with coordinates (3.2, 4.5, 2.1), if it is located in the interference area but not in the obstacle avoidance area, and the distance from the starting point is 5.7 meters and the distance from the end point is 8.3 meters, then its information is represented as {(3.2, 4.5, 2.1), 1, 0, 5.7, 8.3}.

[0126] Based on the acquired grid cell information, the system constructs a reward function to guide the reinforcement learning algorithm's path search. The reward function consists of three main components: a target distance reward, an interference area penalty, and an obstacle avoidance area penalty. The target distance reward is designed to be a negative function of the distance from the current position to the destination, with a higher reward for closer distances. In practice, this component can be set to the negative value of the distance multiplied by a weight coefficient. For example, with a weight of 0.5, the reward increases by 0.5 for every 1 meter decrease in distance. The interference area penalty generates a negative reward when the path passes through an interference area. The penalty value can range from -2 to -10 depending on the level of interference. The obstacle avoidance area penalty generates a larger negative reward when the path approaches or crosses an obstacle. It is typically set to -20 to -100 to ensure that the path does not cross impassable areas. For example, if a grid cell is 10 meters from the destination, within a moderate interference area but not within an obstacle avoidance area, its total reward might be: -10 × 0.5 - 5 + 0 = -10.

[0127] The system uses a policy neural network to generate movement direction probabilities based on information from three-dimensional grid cells. This neural network employs a multi-layer perceptron architecture. The input layer receives the feature vector of the grid cell, including information about the current position, obstacles and interference in 26 surrounding directions, and the distance to the starting and ending points. The hidden layer uses the Reluctant Unified Unit (ReLU) activation function and consists of two hidden layers of 64 neurons. The output layer uses the Softmax function to output a probability distribution over 26 possible movement directions (corresponding to 26 adjacent grid cells in three-dimensional space). The network is trained using stochastic gradient descent with a learning rate of 0.001 and a batch size of 128. The training data is derived from path exploration experience in a simulated environment and consists of approximately 100,000 state-action-reward sequences.

[0128] To generate a diverse set of candidate paths, the system applies a temperature adjustment parameter, τ, to the direction probabilities output by the policy neural network. This parameter controls the flatness of the probability distribution. A larger τ value results in a more uniform distribution of probabilities for each direction, making path exploration more random. A smaller τ value results in a network bias toward the direction with the highest probability, making the path more certain. In practice, the system uses 5-10 different τ values (e.g., 0.5, 0.8, 1.0, 1.2, and 1.5) to perform multiple path samplings. For example, when τ = 0.5, if the original probabilities for upward, rightward, and forward directions at a given location are 0.6, 0.3, and 0.1, the adjusted probability distribution might become 0.75, 0.2, and 0.05, favoring upward. When τ = 1.5, the adjusted probabilities might become 0.45, 0.35, and 0.2, reducing the variance in the probability of selecting the three directions.

[0129] Through the temperature regulation mechanism described above, the system generates multiple initial routing paths. These paths are analyzed and screened for duplication to ensure the diversity of the final candidate paths. The duplication analysis is based on the spatial overlap of the paths, calculating the proportion of grid cells shared by two paths to the total path length. If the overlap rate of two paths exceeds 70%, they are considered highly similar and only one is retained. In addition, the system also filters out paths with a total length exceeding 1.5 times that of the shortest path, as well as paths with excessive coverage of interference areas or obstacle avoidance areas. After screening, the system ultimately retains 3-5 representative candidate routing paths for subsequent optimization and decision-making.

[0130] In this embodiment, an octree data structure and an adaptive grid division strategy are used to achieve efficient expression of complex three-dimensional space, providing refined descriptions in key areas while reducing computational overhead. By obtaining multi-dimensional information of grid cells and constructing a reward function that includes a reward and penalty mechanism, the path planning process is effectively guided to avoid high-interference areas and obstacle avoidance areas. The intelligent decision-making mechanism based on a policy neural network, combined with a probability sampling method for temperature parameter adjustment, not only ensures the exploration capability of path search, but also avoids falling into local optimal solutions. Through path repeatability analysis and screening, the candidate paths finally generated not only meet the wiring constraint requirements, but also have good diversity, which significantly improves the reliability and adaptability of the wiring scheme. This solution realizes the intelligence and automation of robot cable wiring path planning, and effectively solves the wiring optimization problem in complex environments.

[0131] Figure 2 This is a comparative analysis diagram of the reward function components of an embodiment of the present invention. Figure 2 As shown in the figure, this figure shows the impact of different reward function designs on path planning. The horizontal axis represents the number of iterations, and the vertical axis represents the average cumulative reward value. This technical solution (diamond mark) adopts a composite reward function that integrates target distance reward, interference area penalty and obstacle avoidance area penalty. After 200 iterations, it reaches a stable reward value of 876.5, and the convergence speed is 35.2% faster than the traditional method. The distance priority method (circular mark, such as the A* algorithm) only considers the target distance and is prone to falling into local optimality. It eventually converges to a reward value of 651.2, and the pass rate in the interference area is only 72.3%. The obstacle avoidance priority method (triangle mark, such as the artificial potential field method) overemphasizes obstacle avoidance, resulting in an excessively long path, and eventually converges to a reward value of 728.9, with an average increase of 18.7% in path length. The data shows that this technical solution has significant advantages in complex environments and can achieve highly efficient path planning while ensuring safety.

[0132] In an optional embodiment,

[0133] Calculating the sum of the electromagnetic field strength values of the three-dimensional grid cells passed by each routing path among multiple candidate routing paths, and selecting the candidate routing path with the smallest sum of the electromagnetic field strength values as the target routing path includes:

[0134] Dividing the three-dimensional space into a plurality of grid cells, collecting electromagnetic field information of each grid cell, and extracting electric field components in three directions based on the electromagnetic field information; performing vector synthesis operations on the electric field components to obtain a comprehensive electromagnetic field strength of each grid cell;

[0135] An electromagnetic field distribution map is established based on the comprehensive electromagnetic field strength, and multiple candidate wiring paths are tracked using a grid scanning method. During the path tracking process, the comprehensive electromagnetic field strength of the grid cells passed by each candidate wiring path is accumulated to obtain the cumulative electromagnetic strength value of the path.

[0136] An exponential weighting method is used to calculate the comprehensive score of each candidate wiring path, where the comprehensive score is determined by the cumulative value of the path electromagnetic intensity and the path length; and the wiring path with the best comprehensive score is selected as the target wiring path.

[0137] For example, a three-dimensional space is first divided into multiple grid cells. The size of the grid cells can be customized based on the actual application scenario and accuracy requirements. For example, the three-dimensional space can be divided into cubic grid cells with a side length of 5 cm. For each grid cell, electromagnetic field information is collected using an electromagnetic field sensor array. The electromagnetic field sensor array includes multiple sensor nodes distributed at different locations, each capable of measuring the electric and magnetic field components at a specific location.

[0138] After acquisition, the three electric field components are extracted from the electromagnetic field information: the x-direction electric field component Ex, the y-direction electric field component Ey, and the z-direction electric field component Ez. During the extraction process, digital filtering techniques can be used to remove noise interference and improve the accuracy of the electric field components. For example, for the grid cell at coordinates (3, 4, 5), the measured electric field components might be Ex = 2.5 V / m, Ey = 1.8 V / m, and Ez = 3.2 V / m.

[0139] A vector synthesis operation is performed on the extracted electric field components to obtain the comprehensive electromagnetic field strength for each grid cell. Vector synthesis uses the principle of vector synthesis in three-dimensional space to combine the electric field components in three directions to obtain the composite electric field strength. Specifically, the composite electromagnetic field strength is equal to the square root of the sum of the squares of the electric field components in the three directions. For the above grid cell, its comprehensive electromagnetic field strength is 4.47 V / m.

[0140] Based on the calculated comprehensive electromagnetic field intensity for each grid cell, an electromagnetic field distribution map of the entire three-dimensional space is created. This map is stored in the form of a three-dimensional matrix, with each element in the matrix corresponding to the comprehensive electromagnetic field intensity value of a grid cell. For example, within a 10×10×10 spatial region, a three-dimensional electromagnetic field intensity matrix containing 1000 elements can be obtained.

[0141] Based on the electromagnetic field distribution map, a grid scanning method is used to trace multiple candidate routing paths. This path tracing utilizes the A* algorithm, which combines the advantages of the Dijkstra algorithm with the heuristic properties of best-first search. Starting from the starting point, the path search gradually explores adjacent grid cells until the destination is reached. For each explored grid cell, its location coordinates and corresponding integrated electromagnetic field strength are recorded.

[0142] During path tracing, the integrated electromagnetic field strength of each grid cell passed by each candidate routing path is accumulated to obtain the path's cumulative electromagnetic strength value. For example, if a candidate path passes through five grid cells with integrated electromagnetic field strengths of 4.47 V / m, 3.85 V / m, 5.12 V / m, 2.93 V / m, and 4.26 V / m, the path's cumulative electromagnetic strength value is 20.63 V / m.

[0143] To comprehensively consider the electromagnetic field strength and length of the path, an exponentially weighted approach is used to calculate the comprehensive score of each candidate routing path. The comprehensive score is determined by the cumulative electromagnetic strength value of the path and the path length. Specifically, the comprehensive score is equal to the sum of the cumulative electromagnetic strength value of the path multiplied by the weight coefficient α and the path length multiplied by the weight coefficient β. The weight coefficients α and β range from 0 to 1, with α + β = 1, and can be adjusted according to actual application requirements. For example, if the electromagnetic interference sensitivity is higher than the path length requirement, α = 0.7 and β = 0.3 can be set; conversely, α = 0.3 and β = 0.7 can be set.

[0144] To illustrate this with a specific example, consider three candidate routing paths: A, B, and C. Path A has a cumulative electromagnetic strength value of 20.63 V / m and a length of 15 grid units; Path B has a cumulative electromagnetic strength value of 18.75 V / m and a length of 17 grid units; and Path C has a cumulative electromagnetic strength value of 22.41 V / m and a length of 14 grid units. Setting the weight coefficients α = 0.6 and β = 0.4, the comprehensive scores for the three paths are: Path A = 18.38, Path B = 17.65, and Path C = 19.05. Since Path B has the lowest comprehensive score, it is selected as the target routing path.

[0145] In practical applications, parameters such as grid cell size and weight coefficient can be adjusted to optimize routing paths based on different routing environments and requirements. For example, in areas with severe electromagnetic interference, the grid cell size can be reduced to improve path planning accuracy. In applications particularly sensitive to electromagnetic interference, the weight coefficient of the accumulated electromagnetic intensity can be increased to make the path more inclined to avoid areas with high electromagnetic field intensity.

[0146] In this embodiment, by dividing the three-dimensional space into grid units and collecting electromagnetic field information, combined with vector synthesis operations, an accurate characterization of the electromagnetic field distribution in the workspace is achieved. Path tracking is performed based on the electromagnetic field distribution map and grid scanning method, and the degree of electromagnetic interference to each candidate path is accurately obtained. An exponentially weighted scoring mechanism is adopted, which comprehensively considers the two key indicators of the cumulative value of the electromagnetic intensity of the path and the length of the path, thereby ensuring the anti-interference performance of the wiring scheme while taking into account the wiring cost. By selecting the wiring path with the best comprehensive score as the target path, a scientific path evaluation system is established. This solution realizes the optimal selection of the robot cable wiring path, significantly improves the cable's anti-electromagnetic interference ability, reduces the risk of signal transmission distortion, and at the same time ensures the practicality and feasibility of the wiring path, providing a guarantee for the reliable operation of the industrial robot.

[0147] Figure 3 This is a cumulative distribution histogram of electromagnetic intensity along the path of the embodiment of the present invention, as shown in FIG. Figure 3 The figure shows a comparison of the accumulated electromagnetic intensity values of four algorithms along five different routing paths. This solution (diamonds) demonstrated significant advantages across all tested paths, particularly in path 4, where the accumulated electromagnetic intensity was only 19.5 μW / cm², 78.3% lower than the Dijkstra algorithm (squares). The graph shows that the traditional Dijkstra algorithm focuses on the shortest geometric distance and tends to traverse high-intensity areas in the electromagnetic field environment, resulting in a generally accumulated electromagnetic intensity range of 87-96 μW / cm², the highest among all algorithms. The simulated annealing algorithm (triangles) avoids some high-intensity areas by probabilistically accepting suboptimal solutions, but its randomness leads to unstable results, with accumulated values fluctuating between 52-59 μW / cm². The ant colony algorithm (circles), which uses a pheromone mechanism to optimize paths, outperformed both the Dijkstra and simulated annealing algorithms. However, due to its complex parameter adjustment and sensitivity to initial conditions, its accumulated values ranged from 61-68 μW / cm². This technical solution combines electromagnetic field vector synthesis and an exponentially weighted scoring mechanism to accurately identify and avoid high-intensity areas, while also achieving a good balance in length control. Its cumulative value of path 3 is only 24.7μW / cm², which is 62.1% lower than that of the ant colony algorithm, while the path length increases by only 8.3%, demonstrating its superior performance in complex electromagnetic environments.

[0148] In an optional embodiment,

[0149] Determining multiple fixed point coordinates along a target wiring path, installing fixtures at the multiple fixed point coordinates, and completing cable installation along the target wiring path includes:

[0150] Perform force analysis along the target wiring path to calculate the gravity distribution generated by the cable's own weight and the elastic stress distribution caused by bending deformation. Based on the gravity distribution and elastic stress distribution, use the superposition principle to calculate the comprehensive stress distribution at each point on the path. Determine the maximum stress threshold between fixed points based on the material's allowable stress.

[0151] Using a stress iteration calculation method, the maximum allowable spacing between adjacent fixed points is calculated under the condition that a maximum stress threshold constraint is satisfied. Curvature analysis is performed on the target wiring path, and at turns where the curvature exceeds a preset curvature threshold, the maximum allowable spacing is dynamically adjusted according to the curvature proportional coefficient.

[0152] Based on the adjusted maximum allowable spacing, the initial positions of the fixed points are marked on the target routing path using an evenly spaced distribution method. The path tangent vector and normal vector at the initial position of each fixed point are calculated. The initial position of the fixed point is optimized based on the tangent vector and normal vector to obtain the final position of the fixed point.

[0153] According to the spatial characteristics and force characteristics of the final position of the fixed point, the fixing device is divided into a straight segment fixing fixture, a corner reinforcement fixture and an end fixing device; the straight segment fixing fixture adopts a standard clamping force, the corner reinforcement fixture provides additional radial support force, and the end fixing device has an anti-falling mechanism; the installation direction of each fixing device is determined based on the path tangent vector, and the fixed point installation is completed.

[0154] For example, firstly, a force analysis is performed on the target wiring path to calculate the force conditions of the cable on the path, then the position of the fixing point is determined, and finally the corresponding fixing device is installed according to the position characteristics.

[0155] When performing a force analysis on the target cabling path, it's necessary to consider the distribution of gravity generated by the cable's own weight. For example, for a 100-meter cable weighing 2.5 kg per meter, the unit weight on a horizontal path is 24.5 Newtons per meter. When the path is non-horizontal, the gravity component varies depending on the path's inclination. For example, on a path segment with a 30-degree inclination, the gravity component perpendicular to the path is 21.2 Newtons per meter. Furthermore, elastic stress is generated at bends in the cable, and this stress is related to the cable's elastic modulus and bend radius. For a copper-core cable with a cross-sectional area of 50 square millimeters and an elastic modulus of 110 GPa, the elastic stress generated at a bend radius of 0.5 meters is approximately 11 MPa.

[0156] The calculation of the combined stress distribution uses the superposition principle, adding the stress caused by gravity to the elastic stress. In the above example, the combined stress at a certain bend point can reach 15 MPa. Based on the allowable stress of the cable material (for example, 20 MPa for copper-core cable), the maximum stress threshold between fixed points can be determined to be 18 MPa, leaving a 10% safety margin.

[0157] The iterative stress calculation method is used to determine the maximum allowable spacing between adjacent fixed points. In practice, starting from the path's starting point, assuming an initial spacing of 1 meter, the maximum stress value at that spacing is calculated. If the stress value falls below the threshold, the spacing is increased to 1.2 meters and the calculation is repeated. If the stress value exceeds the threshold, the spacing is reduced to 0.8 meters and the calculation is repeated. Through multiple iterations, when the calculated stress value approaches but does not exceed the threshold, the maximum allowable spacing at that point is determined. In practice, for the example cable described above, the maximum allowable spacing calculated under standard conditions is 3.5 meters.

[0158] Analyzing the curvature of the target routing path is a critical step in ensuring secure cable anchoring around bends. When the path curvature exceeds a preset threshold (e.g., a curvature of 2.0 for a 0.5-meter bend radius), the spacing between anchor points needs to be dynamically adjusted. This adjustment uses a curvature scaling factor: the actual spacing equals the maximum allowable spacing divided by the ratio of the curvature to the baseline curvature. For example, if the curvature at a location is 4.0, the spacing between anchor points at that location should be adjusted to 1.75 meters (3.5 meters divided by 2).

[0159] The evenly spaced placement method is used to mark the initial locations of fixed points along the target routing path. Starting from the path's starting point, the fixed points are marked sequentially at the adjusted spacing. For a 100-meter path with an average spacing of 3 meters, approximately 33 fixed points are initially marked.

[0160] For each initial fixed point position, calculate the path's tangent vector and normal vector at that location. The tangent vector indicates the path's forward direction, while the normal vector indicates the direction perpendicular to the path. For example, at a fixed point, the path's tangent vector is (0.866, 0.5, 0), indicating that the path points upward at a 30-degree angle to the horizontal; the normal vector is (-0.5, 0.866, 0), indicating that the path is perpendicular to the path.

[0161] Fixing point position optimization takes into account installation convenience and uniform stress distribution. If the initial position is in a difficult-to-install area (such as a pipe joint), the fixing point is appropriately moved along the path, typically within 10% of the maximum allowable spacing. At the same time, stress distribution between adjacent fixing points is ensured to avoid stress concentration. After optimization, the final fixing point position is more suitable for actual installation requirements.

[0162] Select the appropriate fixture based on the spatial and force characteristics of the fixing point. Straight-section fixtures are suitable for straight sections of the path and feature a standard clamping force design, typically three times the cable weight, approximately 75 Newtons. Bend reinforcement fixtures provide additional radial support, approximately 112.5 Newtons, or 1.5 times the standard clamping force, to prevent cable slippage around bends. End fixtures feature a pull-out prevention mechanism capable of withstanding the full cable tension, typically designed to withstand 500 Newtons.

[0163] The fixture's installation orientation is determined by the path tangent vector. The fixture's opening should be aligned with the path normal vector to ensure the cable is securely positioned within the fixture. For example, when the path tangent vector is (0.866, 0.5, 0), the fixture's major axis should be parallel to this direction, with the opening aligned with the normal vector (-0.5, 0.866, 0).

[0164] After installing the fixing points, proceed with cable laying. Starting at one end, place the cable into each fixture in sequence, ensuring that the cable is centered in the fixture. Tighten the fixing bolts to the specified torque (typically 5-8 Nm). At bends, ensure the cable bends naturally without twisting or excessive stress. Finally, inspect all fixing points to ensure the cable is securely installed and free of significant stress concentrations.

[0165] In this embodiment, the scientific determination of the spacing between fixed points is achieved through precise analysis of the cable gravity distribution and elastic stress distribution, combined with the stress superposition principle and iterative calculation method. Based on curvature analysis and dynamic adjustment mechanism, the distribution of fixed points is made more reasonable, effectively preventing excessive deformation of the cable. The accuracy of the fixed point layout is improved by calculating the path characteristic vector and optimizing the position of the fixed point. Different types of fixing devices are used to solve the different fixing requirements of straight sections, turns and ends. Among them, the reinforced clamps at the turns and the anti-fall-off mechanism at the ends significantly improve the fixing reliability of key positions. The solution of determining the installation direction based on the tangential vector ensures the correct installation of the fixing device and realizes stable support of the cable. The overall solution improves the reliability and service life of the industrial robot cable wiring, reduces the risk of cable damage, and provides a guarantee for the continuous and stable operation of the robot.

[0166] A second aspect of an embodiment of the present invention provides a robot cable anti-interference wiring system based on reinforcement learning, the system comprising:

[0167] The first unit is used to obtain the three-dimensional structural data of the robot body and simultaneously collect the electromagnetic field intensity data in the robot workspace to generate the electromagnetic field intensity distribution data of the robot workspace;

[0168] The second unit is used to divide the robot workspace into multiple three-dimensional grid units, calculate the electromagnetic field strength value of each three-dimensional grid unit according to the electromagnetic field strength distribution data, and obtain the target interference area of the three-dimensional grid unit based on the electromagnetic field strength value;

[0169] The third unit is used to determine the starting and ending coordinates of the cable routing based on the three-dimensional structural data of the robot body and map them to the three-dimensional grid cells. The third unit is used to calculate the motion range coordinates of each joint of the robot during the motion process based on the robot motion trajectory data, and mark the three-dimensional grid cells where the motion range coordinates are located as obstacle avoidance areas.

[0170] The fourth unit is used to generate multiple candidate routing paths in the three-dimensional grid unit through a reinforcement learning algorithm using the target interference area and obstacle avoidance area as search constraints;

[0171] The fifth unit is used to calculate the sum of the electromagnetic field strength values of the three-dimensional grid cells passed by each of the multiple candidate wiring paths, select the candidate wiring path with the smallest sum of the electromagnetic field strength values as the target wiring path; determine the coordinates of multiple fixed points along the target wiring path, install fixtures at the coordinates of the multiple fixed points, and complete the cable installation along the target wiring path.

[0172] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0173] processor;

[0174] a memory for storing processor-executable instructions;

[0175] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0176] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0177] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0178] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A robot cable anti-interference wiring method based on reinforcement learning, characterized in that: include: Acquire the three-dimensional structural data of the robot body, and at the same time collect electromagnetic field intensity data in the robot workspace to generate electromagnetic field intensity distribution data of the robot workspace; The robot workspace is divided into multiple three-dimensional grid units, the electromagnetic field strength value of each three-dimensional grid unit is calculated according to the electromagnetic field strength distribution data, and the three-dimensional grid unit is obtained as the target interference area based on the electromagnetic field strength value; The starting and ending coordinates of the cable routing are determined based on the robot's 3D structural data and mapped to 3D grid cells. The motion range coordinates of each joint during the robot's motion are calculated based on the robot's motion trajectory data, and the 3D grid cells where the motion range coordinates are located are marked as obstacle avoidance areas. The target interference area and obstacle avoidance area are used as search constraints, and multiple candidate routing paths are generated in the three-dimensional grid cells through reinforcement learning algorithm. Calculating the sum of electromagnetic field strength values of three-dimensional grid cells passed by each of the multiple candidate routing paths, and selecting the candidate routing path with the smallest sum of electromagnetic field strength values as the target routing path; Determining coordinates of a plurality of fixed points along a target wiring path, installing fixtures at the coordinates of the plurality of fixed points, and completing cable installation along the target wiring path; Using the target interference area and obstacle avoidance area as search constraints, a reinforcement learning algorithm is used to generate multiple candidate routing paths in the three-dimensional grid cells, including: The octree data structure is used to divide the three-dimensional space into grids, and the grid density is adaptively adjusted according to the complexity of the area to obtain three-dimensional grid units; Obtain the location information, interference area information, obstacle avoidance area information, and distance information from the starting and ending points of the three-dimensional grid unit; Based on the acquired information, a reward function is constructed, including a target distance reward component, an interference area penalty component, and an obstacle avoidance area penalty component. The target distance reward component is used to guide the path to move toward the target point, while the interference area penalty component and the obstacle avoidance area penalty component generate negative reward values when the path passes through the corresponding area. A policy neural network is used to generate the probability of movement direction based on the three-dimensional grid cell information. A temperature adjustment parameter is set on the moving direction probability output by the strategy neural network. The moving direction is sampled probabilistically by adjusting the temperature adjustment parameter to generate multiple initial wiring paths. The multiple initial wiring paths are subjected to repetitive analysis and screening to generate multiple candidate wiring paths.

2. The method according to claim 1, characterized in that Acquire the three-dimensional structural data of the robot body and collect the electromagnetic field intensity data in the robot workspace at the same time. Generate the electromagnetic field intensity distribution data of the robot workspace including: Arranging characteristic marking strips on the surface of the robot body along the axis of the motion joint, wherein the characteristic marking strips include a periodic coding pattern; identifying the joint position and the connecting rod posture according to the spatial distribution characteristics of the coding pattern, and simultaneously collecting local electromagnetic field data through a probe in the coding pattern; The robot is controlled to move along a preset trajectory, and the joint rotation angle is calculated based on the real-time position of the characteristic marking strip. When the overlapping fields of view of adjacent structured light cameras capture the same coded pattern, the robot's movement speed is adaptively adjusted according to the joint rotation angle, and the probe is triggered to collect electromagnetic field data. The collected electromagnetic field data is spatially mapped according to the spatial position of the characteristic marker strips, and the mapped electromagnetic field data is interpolated and expanded based on the electromagnetic field propagation characteristics to obtain the electromagnetic field intensity distribution data of the working space.

3. The method according to claim 1, characterized in that The robot workspace is divided into multiple three-dimensional grid units, and the electromagnetic field strength value of each three-dimensional grid unit is calculated according to the electromagnetic field strength distribution data. The target interference area of the three-dimensional grid unit is obtained based on the electromagnetic field strength value. Divide the robot workspace into three-dimensional grids and calculate the electromagnetic field intensity gradient of each three-dimensional grid unit; divide the three-dimensional grid units into densities based on the electromagnetic field intensity gradient to generate gradient adaptive grid units; Decomposing the electromagnetic field intensity distribution data into three orthogonal components, performing interpolation calculation on each orthogonal component using a radial basis function, wherein the weight of the interpolation calculation is determined according to the electromagnetic field intensity gradient; applying a divergence constraint to the result of the interpolation calculation to obtain the electromagnetic field intensity value of each gradient adaptive grid cell; Calculating the field strength characteristic parameters of the gradient adaptive grid unit based on the electromagnetic field strength, wherein the field strength characteristic parameters include the field strength amplitude and the directional change rate; performing wavelet transform on the field strength characteristic parameters to obtain characteristic components at different scales; establishing an interference degree judgment standard based on the characteristic components to determine the interference level of each gradient adaptive grid unit; The electromagnetic field intensity value changes of the gradient adaptive grid cells are monitored in real time. When the electromagnetic field intensity value changes exceed the preset intensity value range, the corresponding gradient adaptive grid cells are marked as cells to be updated; only the density division, interpolation calculation and field strength characteristic parameter calculation are re-performed on the cells to be updated to update the interference level; and the target interference area is obtained based on the interference level of each updated gradient adaptive grid cell.

4. The method according to claim 1, wherein The starting and ending coordinates of the cable routing are determined based on the robot's 3D structural data and mapped to 3D grid cells. The range of motion coordinates of each joint during the robot's motion is calculated based on the robot's motion trajectory data. The 3D grid cells where the range of motion coordinates are located are marked as obstacle avoidance areas, including: The robot's joint axis position data is obtained, and the endpoints of the joint axis are used as the starting and ending coordinates for cable routing. The initial routing curve of the cable is generated based on the starting and ending coordinates, and the tangential angle change rate of discrete points on the initial routing curve is calculated. The actual bending deformation path of the cable is determined based on the tangential angle change rate. The actual bending deformation path is discretized into multiple path nodes, and the path nodes are mapped into a three-dimensional grid to obtain the wiring path grid unit; the robot's motion trajectory data is obtained, and the real-time motion angle of each joint is calculated based on the motion trajectory data. Preset motion redundant angles are added to the real-time motion angles; The motion range coordinates of each joint are calculated based on the motion redundancy angle, and the motion range coordinates are mapped to the corresponding three-dimensional grid cells; a first type of safety buffer zone is constructed around the wiring path grid cells, and a second type of safety buffer zone is constructed around the three-dimensional grid cells corresponding to the motion range coordinates; The grid cells are divided into obstacle avoidance levels according to the degree of overlap between the first and second safety buffer zones, and the three-dimensional grid cells whose obstacle avoidance levels are higher than the thresholds of the first and second safety buffer zones are marked as obstacle avoidance areas.

5. The method according to claim 1, wherein Calculating the sum of the electromagnetic field strength values of the three-dimensional grid cells passed by each routing path among multiple candidate routing paths, and selecting the candidate routing path with the smallest sum of the electromagnetic field strength values as the target routing path includes: Dividing the three-dimensional space into a plurality of grid cells, collecting electromagnetic field information of each grid cell, and extracting electric field components in three directions based on the electromagnetic field information; performing vector synthesis operations on the electric field components to obtain a comprehensive electromagnetic field strength of each grid cell; An electromagnetic field distribution map is established based on the comprehensive electromagnetic field strength, and multiple candidate wiring paths are tracked using a grid scanning method. During the path tracking process, the comprehensive electromagnetic field strength of the grid cells passed by each candidate wiring path is accumulated to obtain the cumulative electromagnetic strength value of the path. An exponential weighting method is used to calculate the comprehensive score of each candidate wiring path. The comprehensive score is determined by the cumulative value of the path electromagnetic intensity and the path length. The wiring path with the best comprehensive score is selected as the target wiring path.

6. The method according to claim 1, characterized in that Determining multiple fixed point coordinates along a target wiring path, installing fixtures at the multiple fixed point coordinates, and completing cable installation along the target wiring path includes: Perform force analysis along the target wiring path to calculate the gravity distribution generated by the cable's own weight and the elastic stress distribution caused by bending deformation. Based on the gravity distribution and elastic stress distribution, use the superposition principle to calculate the comprehensive stress distribution at each point on the path. Determine the maximum stress threshold between fixed points based on the material's allowable stress. Using a stress iteration calculation method, the maximum allowable spacing between adjacent fixed points is calculated under the condition that a maximum stress threshold constraint is satisfied. Curvature analysis is performed on the target wiring path, and at turns where the curvature exceeds a preset curvature threshold, the maximum allowable spacing is dynamically adjusted according to the curvature proportional coefficient. Based on the adjusted maximum allowable spacing, the initial positions of the fixed points are marked on the target routing path using an evenly spaced distribution method. The path tangent vector and normal vector at the initial position of each fixed point are calculated. The initial position of the fixed point is optimized based on the tangent vector and normal vector to obtain the final position of the fixed point. According to the spatial characteristics and force characteristics of the final position of the fixed point, the fixing device is divided into a straight segment fixing fixture, a corner reinforcement fixture and an end fixing device; the straight segment fixing fixture adopts a standard clamping force, the corner reinforcement fixture provides additional radial support force, and the end fixing device has an anti-falling mechanism; the installation direction of each fixing device is determined based on the path tangent vector, and the fixed point installation is completed.

7. A robot cable anti-interference wiring system based on reinforcement learning, used to implement the method according to any one of claims 1 to 6, characterized in that: include: The first unit is used to obtain the three-dimensional structural data of the robot body and simultaneously collect the electromagnetic field intensity data in the robot workspace to generate the electromagnetic field intensity distribution data of the robot workspace; The second unit is used to divide the robot workspace into multiple three-dimensional grid units, calculate the electromagnetic field strength value of each three-dimensional grid unit according to the electromagnetic field strength distribution data, and obtain the target interference area of the three-dimensional grid unit based on the electromagnetic field strength value; The third unit is used to determine the starting and ending coordinates of the cable routing based on the three-dimensional structural data of the robot body and map them to the three-dimensional grid cells. The third unit is used to calculate the motion range coordinates of each joint of the robot during the motion process based on the robot motion trajectory data, and mark the three-dimensional grid cells where the motion range coordinates are located as obstacle avoidance areas. The fourth unit is used to generate multiple candidate routing paths in the three-dimensional grid unit through a reinforcement learning algorithm using the target interference area and obstacle avoidance area as search constraints; The fifth unit is used to calculate the sum of the electromagnetic field strength values of the three-dimensional grid cells passed by each routing path among the multiple candidate routing paths, and select the candidate routing path with the smallest sum of the electromagnetic field strength values as the target routing path; A plurality of fixed point coordinates are determined along a target wiring path, fixtures are installed at the plurality of fixed point coordinates, and the cable installation is completed along the target wiring path.

8. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 6.

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

Images