Robot autonomous obstacle avoidance method and equipment for transformer substation control equipment

Abstract

The invention relates to the technical field of autonomous obstacle avoidance, in particular to a robot autonomous obstacle avoidance method and device for substation control equipment, and the method comprises the steps: projecting a coded grating pattern to the environment in front of a robot through a structured light measurement technology, collecting a grating image, carrying out the phase demodulation of the grating image, obtaining an absolute phase diagram and a modulation degree diagram, and obtaining a modulation degree diagram; calculating a depth value of each pixel point based on the grating image according to a triangular depth estimation principle so as to generate a depth image; calculating the adjustment coefficient of each pixel point, correcting the depth value of the pixel point in the depth image, determining the risk assessment value of each pixel point, projecting the risk assessment value to a two-dimensional grid map to generate an obstacle risk map, improving the total cost function of the A * algorithm, and planning the obstacle avoidance path of the robot in the transformer substation. The risk that the robot collides with the obstacle can be reduced.

Description

Technical Field

[0001] This application relates to the field of autonomous obstacle avoidance technology, specifically to a method and device for autonomous obstacle avoidance of robots for substation control equipment. Background Technology

[0002] Substations are crucial locations for voltage transformation, power distribution, and control, containing control equipment such as high-voltage switchgear, transformers, circuit breakers, and disconnectors. Substation inspection robots are intelligent robots capable of autonomously moving within the substation environment and performing inspection tasks such as equipment status monitoring, instrument readings, and infrared temperature measurement. Their autonomous obstacle avoidance capability is the core technology ensuring inspection safety and successful task execution.

[0003] Traditional obstacle avoidance methods for robots mainly rely on path planning based on a pre-set environmental map. Obstacle avoidance is achieved by sensing the environment and matching the map using a monocular RGB camera. However, in substation environments, equipment surfaces have diverse materials. During maintenance, obstacles not recorded in the map, such as toolboxes, ladders, and warning signs, may be temporarily placed. Additionally, new equipment or temporary wiring may be added. These obstacles and the metal surfaces of the equipment have strong specular reflection characteristics, which causes the traditional method of depth estimation based on texture features using a monocular RGB camera to fail. The reliability of depth measurement is low, and it is impossible to accurately determine the actual distance to obstacles, which can easily lead to collisions or overly conservative obstacle avoidance strategies. Summary of the Invention

[0004] To address the aforementioned technical problems, a method and device for autonomous obstacle avoidance by robots in substation control equipment are provided to solve the existing issues.

[0005] The solution to the technical problem in this application is to provide a method and device for autonomous obstacle avoidance by a robot for substation control equipment, including the following steps: In a first aspect, embodiments of this application provide a method for autonomous obstacle avoidance by a robot for substation control equipment, the method comprising the following steps: Using structured light measurement technology, an coded grating pattern is projected onto the environment in front of the robot, and the grating image is acquired. The phase is demodulated to obtain an absolute phase map and a modulation map. Based on the grating image and the principle of triangulation depth estimation, the depth value of each pixel is calculated to generate a depth image. For each pixel, the adjustment coefficient of each pixel is calculated by the gradient change of its absolute phase in the absolute phase map and the corresponding modulation degree of the same pixel in the modulation degree map. The depth value of the pixel in the depth image is corrected. The corrected depth value is used to assess the safety of the robot's distance from the obstacle. Combined with the corresponding modulation degree information in the modulation degree map, the danger assessment value of each pixel is determined and projected onto the two-dimensional grid map to generate an obstacle danger map. Based on the hazard status of different grids in the obstacle hazard map, the total cost function of the A* algorithm is improved. The improved total cost function of the A* algorithm is then used to plan the obstacle avoidance path of the robot in the substation.

[0006] Preferably, the process of obtaining the absolute phase map is as follows: based on the raster image, phase demodulation is performed through an N-step phase shift algorithm, the wrapped phase is calculated, the wrapped phase is expanded, the absolute phase of each pixel is calculated, and an absolute phase map is generated.

[0007] Preferably, the process of obtaining the modulation map is as follows: based on the raster image, the background light intensity and modulation light intensity of each pixel are calculated through an N-step phase shift algorithm, the ratio of modulation light intensity to background light intensity is used as the modulation index of each pixel, and a modulation map is generated.

[0008] Preferably, the calculation of the adjustment coefficient for each pixel includes: Based on the absolute phase map, the magnitude of the phase gradient of each pixel is calculated, and its product with the preset attenuation coefficient is negatively mapped to serve as the penalty factor for each pixel. A predefined modulation reference range corresponding to the surface reflection characteristics of the object being measured in the substation is used to map the modulation of each pixel in the modulation graph to the reference range, thus obtaining the mapped modulation. The adjustment coefficient is the result of a positive fusion of a preset correction intensity coefficient, a mapped modulation index, and a penalty factor.

[0009] Preferably, the specific process of positive fusion is as follows: the product of the preset correction intensity coefficient, the mapped modulation degree, and the penalty factor is used as the adjustment coefficient.

[0010] Preferably, the pixel position in the depth image is The location corresponds to the corrected depth value. The calculation formula is: ,in, The pixel position in the depth image is Depth value at that location, The pixel position in the depth image is Adjustment coefficient at the location.

[0011] Preferably, determining the danger assessment value of each pixel includes: if the corrected depth value corresponding to each pixel in the depth image is greater than a preset safety distance, then the danger assessment value of each pixel is 0; otherwise, the calculation process of the danger assessment value is: performing a negative mapping on the ratio of the corrected depth value to the preset safety distance; and performing a positive fusion between the result of the negative mapping and the normalized modulation corresponding to the pixel, as the danger assessment value of each pixel.

[0012] Preferably, the obstacle hazard map generation process is as follows: the horizontal ground in the environment in front of the robot is divided into a two-dimensional grid map, and the hazard assessment values ​​of all pixels in the depth image are projected onto the two-dimensional grid map. The maximum hazard assessment value in each grid of the two-dimensional grid map is used as the hazard level of each grid to generate the obstacle hazard map.

[0013] The preferred and improved total cost function of the A* algorithm for: in, Indicates the distance from the starting point to the node. The actual cost, For nodes The heuristic cost to the destination, The preset risk weight coefficient, In a two-dimensional raster map, the coordinates are... The danger level of the grid, Indicates the distance from the starting point to the node. The full path.

[0014] Secondly, embodiments of this application also provide a robot autonomous obstacle avoidance device for substation control equipment, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-described robot autonomous obstacle avoidance methods for substation control equipment.

[0015] This application has at least the following beneficial effects: This application projects a four-step phase-shift sinusoidal grating pattern onto the substation environment using structured light measurement technology. Absolute phase map, modulation map, and depth image are obtained through phase demodulation calculation. The advantages include obtaining a modulation map reflecting surface reflection characteristics, providing a physical basis for distinguishing between specular and diffuse reflection from metal surfaces, thus quantifying the reflection characteristics of obstacle surfaces and increasing the perception capability of the substation's metallic environment. The absolute phase map quantifies the continuous and unambiguous phase information of the object's surface, accurately reflecting its three-dimensional shape, enabling the robot to acquire high-precision depth information. The adjustment coefficients of each pixel are calculated to correct the depth values ​​in the depth image. This effectively reduces depth measurement errors caused by differences in surface reflection characteristics by combining phase gradient information reflecting the sensitivity of contour edge abrupt changes with modulation information reflecting surface reflection characteristics, significantly reducing the error in depth estimation of metallic reflective surfaces. Based on this, the safety of the robot's distance from obstacles is assessed. Combined with the corresponding modulation information in the modulation map, the hazard assessment value of each pixel is determined and projected onto a two-dimensional grating. The grid-based obstacle hazard map generation method combines corrected depth information with surface reflectivity information, quantifying both distance threat and the reliability of depth measurements. This allows the robot to directly assess its relative position to obstacles, accurately determining the risk of collision. It also enables the robot to distinguish surfaces with different reflectivity, preventing depth measurement errors caused by surface characteristics from being misjudged as real obstacles, reducing false alarms and missed alarms, and thus accurately identifying and quantifying the hazard level of each location. Based on the hazard status of different grids in the obstacle hazard map, the total cost function of the A* algorithm is improved. Using this improved total cost function, obstacle avoidance paths for robots within substations are planned. The benefit lies in introducing a hazard penalty term into the total cost function, allowing the robot to perceive the hazard level of different areas and proactively avoid high-risk areas during path planning. This mechanism significantly improves path safety, reduces the risk of collisions between the robot and obstacles, and significantly enhances the autonomous obstacle avoidance safety and reliability of substation inspection robots in complex substation metal environments. Attached Figure Description

[0016] The following description, in conjunction with the accompanying drawings, provides a more detailed explanation of a robot-based autonomous obstacle avoidance method for substation control equipment according to this application.

[0017] Figure 1 A flowchart illustrating the steps of a robot autonomous obstacle avoidance method for substation control equipment provided in this application embodiment; Figure 2 A flowchart illustrating the steps of the method for obtaining the adjustment coefficients of each pixel as provided in the embodiments of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description, in conjunction with the accompanying drawings and implementation examples, provides a more comprehensive explanation of the robot autonomous obstacle avoidance method and equipment for substation control equipment proposed in this application. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0020] Please see Figure 1 The diagram illustrates a flowchart of a robot autonomous obstacle avoidance method for substation control equipment according to an embodiment of this application. The method includes the following steps: Step 1: Using structured light measurement technology, a coded grating pattern is projected onto the environment in front of the robot, and the grating image is acquired. The phase of the grating image is demodulated to obtain an absolute phase map and a modulation map. Based on the grating image and the principle of triangular depth estimation, the depth value of each pixel is calculated to generate a depth image.

[0021] With the advancement of intelligent power systems, substation inspection robots are gradually replacing manual inspections, offering significant advantages in improving safety and operational efficiency. Path planning, as a core component, requires generating smooth, safe, and collision-free optimal paths in environments with dense equipment and complex obstacles.

[0022] Due to the complex working environment of substations, which contain various equipment, cables, supports and other obstacles, in order to accurately obtain the three-dimensional depth of obstacles in the substation environment and effectively distinguish between the specular and diffuse reflection surfaces of metal on different equipment, structured light technology is used to project and encode grating patterns, and the deformation characteristics of the grating patterns on different surfaces are analyzed to achieve more accurate depth measurement.

[0023] Among them, the grating projection measurement method is a structured light method based on area array projection. Its principle is to project a grating pattern with specific encoded information onto the surface of the object being measured by a projector. When the grating is irradiated onto the surface of the object with depth changes, it will be deformed. Then, the deformed grating image is captured by a camera and encoded to restore the three-dimensional shape of the object.

[0024] Based on the above analysis, deploying a grating projection measurement system on a robot typically consists of an industrial camera and a DLP projector, wherein the baseline distance between the projector and the industrial camera is set to... The projector's focal length is The camera's focal length is The projector projects a coded four-step phase-shift sinusoidal grating pattern onto the environment in front of the robot, and the industrial camera simultaneously acquires the grating image. In this embodiment, the projector has a coverage angle of 90 degrees in the horizontal direction and a coverage angle of 60 degrees in the vertical direction.

[0025] In 3D measurement, to project a sinusoidal fringe grating with a certain phase shift, the light intensity generation distribution function of the corresponding pixels is: in, The position of a pixel in the k-th raster image is The light intensity at that location, The pixel position is Background light intensity at that location The pixel position is Modulated light intensity at that location The pixel position is The wrapping phase at the location, Indicates the first Phase shift of a grating image Let f(x) represent the cosine function, where f(x) = f(x) + ... ; It should be noted that the background light intensity represents the combined effect of ambient light and surface reflectivity, the modulated light intensity represents the contrast of the grating stripes, and the wrapping phase contains depth information.

[0026] In the above formula, the background light intensity Modulated light intensity and the wrap phase Since the light intensity distribution of the sinusoidal grating pattern is sinusoidal, at least three equations are needed to determine the wrapping phase. Therefore, a four-step phase-shifting algorithm is used for phase decomposition. The projector sequentially projects four images with a phase difference of... The sinusoidal grating patterns, the phases of these four patterns are 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 10, 1 ... , , The camera will generate four raster images in sequence, namely... , , and Therefore, based on the four-step phase shift algorithm, phase demodulation is performed using the four-step phase shift algorithm to calculate the wrap-around phase, specifically as follows: By solving the system of equations, we can obtain: It should be noted that the process of calculating the wrapper phase using the four-step phase shift algorithm is a well-known technique and will not be elaborated upon here; therefore, by analyzing the four phase differences... Trigonometric function operations are performed on the raster image to eliminate background luminance. and modulated light intensity This allows for the extraction of phase information.

[0027] Since the mapping of the wrapped phase is not unique, in order to eliminate the periodic discontinuity of the wrapped phase and obtain the absolutely phase with a unique mapping, phase expansion is required to calculate the absolute phase, specifically: The multi-frequency heterodyne method is used to perform phase unwrapping on the wrapped phase and calculate the pixel position. The absolute phase at that point is denoted as It also outputs the absolute phase map; It should be noted that the multi-frequency heterodyne method is a well-known technique and will not be described in detail here. The specific process is as follows: multiple sets of grating patterns with different spatial frequencies are projected, wherein the fringe periods corresponding to the spatial frequencies are 70 pixels, 64 pixels and 59 pixels respectively. The wrapping phase of each set of grating patterns is calculated by a four-step phase shift algorithm. Then, by calculating the difference frequency of the wrapping phase between different frequencies, a composite phase with a longer equivalent wavelength is generated. The heterodyne principle is used to expand it step by step to finally obtain the absolute phase.

[0028] Modulation intensity is a physical quantity that reflects the sharpness and contrast of grating fringes. It is defined as the ratio of modulated light intensity to background light intensity, i.e. By using a system of simultaneous equations, the modulated light intensity and background light intensity are determined, and the pixel position is calculated. The adjustment system of the department, the specific derivation process is as follows: make , , ,because , ,but , ; because , ,but , ; + Therefore, we can obtain + ,but because , Therefore, the formula for calculating the modulation index is: It should be noted that the calculation of modulation degree is a well-known technique and will not be elaborated here. Modulation degree refers to the contrast between the light and dark areas of the grating fringes. For an ideal diffuse surface, such as a white wall, incident light is uniformly scattered, and the reflected light maintains the contrast of the grating pattern, resulting in clear grating fringes; therefore, a higher modulation degree is required. For a metallic specular surface, such as a stainless steel panel, incident light is reflected according to the law of specular reflection, and only light at a specific angle enters the camera. Furthermore, the high-gloss reflection from the metallic surface leads to increased background light intensity. Increase and modulate light intensity The smaller the relative size, the blurrier fringes become, indicating a smaller modulation depth.

[0029] Therefore, the pixel position is obtained as Adjustment system and output the modulation chart; It should be noted that the grayscale value of a pixel in the absolute phase image represents the absolute phase, while the grayscale value of a pixel in the modulation image represents the modulation degree.

[0030] Based on the principle of triangulation depth estimation, and the relationship between absolute phase and depth information, the pixel position is calculated as follows: The depth value at that location is used to generate a depth image; It should be noted that since the absolute phase map provides the phase value of each pixel in the raster image captured by the camera, this phase value needs to be converted into horizontal coordinates on the projector image, thus encoding the phase information into spatial points. The absolute phase uniquely identifies the direction of the light emitted by the projector. By comparing the horizontal coordinates on the projector image with the horizontal coordinates of the camera pixels, the parallax is calculated, and the depth value of the pixel is calculated using the triangulation formula. Specifically, the projector is treated as a virtual camera, thus forming a binocular vision system. Based on the principle of triangulation depth estimation, the pixel values ​​are... absolute phase value Converted to the corresponding coordinate information on the projector image, combined with the camera pixel coordinates Using the projector coordinate information, the disparity is calculated through the geometric relationship of similar triangles. Finally, the depth value of the pixel is calculated using the triangulation formula, where the specific formula is as follows: in, Indicates the position of the pixel. Depth value at that location, Indicates the baseline distance. Indicates focal length. Indicates parallax; It should be noted that the principle of triangulation depth estimation is a well-known technique and will not be elaborated here.

[0031] At this point, we have obtained a depth image, an absolute phase map, and a modulation map of the environment in front of the robot.

[0032] Step 2: For each pixel, calculate the adjustment coefficient of each pixel by considering the gradient change of its absolute phase in the absolute phase map and the corresponding modulation degree of the same pixel in the modulation map, and correct the depth value of the pixel in the depth image.

[0033] Furthermore, traditional structured light triangulation methods only utilize the relationship between phase and parallax for depth calculation, neglecting the influence of the reflective properties of different surfaces on the device. This leads to significant errors on metal surfaces. Additionally, lower modulation depth indicates lower grating fringe contrast, resulting in greater noise impact on phase calculation during demodulation and higher measurement errors. To address the unreliability of depth estimation caused by metal mirror reflection, an absolute phase map and modulation depth map are used to correct the depth values ​​of each pixel in the depth image.

[0034] Furthermore, the flowchart of the method for obtaining the adjustment coefficients of each pixel provided in the embodiments of this application is as follows: Figure 2 As shown.

[0035] First, the penalty factor is calculated by analyzing the changes in the phase gradient in the absolute phase map, specifically: Based on the absolute phase map, calculate the magnitude of the phase gradient of each pixel; In this embodiment, since the grayscale value of a pixel in the absolute phase image represents the absolute phase, the phase gradient is calculated based on the grayscale gradient calculation process in the image. The grayscale gradient calculation is a well-known technique and will not be elaborated upon here. The phase gradient is expressed as: The formula for calculating its modulus is: , express right The partial derivatives, express right The partial derivatives; The product of the modulus and the preset attenuation coefficient is negatively mapped and used as a penalty factor for each pixel. In this embodiment, to control the suppression strength of the phase gradient on the correction amplitude, the preset attenuation coefficient is set to 5 pixels / rad. A larger attenuation coefficient results in greater sensitivity to gradient changes and a more severe penalty, while a smaller coefficient is more tolerant of gradient changes and a milder penalty. Secondly, the specific process of negative mapping is as follows: negative mapping is performed using an exponential function. Let's assume the product of the magnitude and the preset attenuation coefficient is denoted as... ,but The result is used as the result of the negative mapping, where, This represents an exponential function with the natural constant as the base. As another implementation, the implementer can use the reciprocal of the product of the modulus and the preset attenuation coefficient as the result of the negative mapping. To avoid the denominator being 0 when calculating the reciprocal, an adjustment parameter is added to the denominator. In this embodiment, the adjustment parameter is set to 1. This embodiment does not impose any special restrictions on this.

[0036] It should be noted that the magnitude of the phase gradient reflects the degree of phase change between adjacent pixels in the absolute phase map. When a pixel is located at the edge of an object, the phase changes abruptly. The larger the magnitude of the phase gradient, the closer the pixel is to the edge of the object, and the smaller the penalty factor. That is, the more the correction amplitude needs to be reduced. This is because at the edge of an object, the light signal of the pixel does not come from a single surface point, but is a mixture of the foreground object and the background. If the correction amplitude is too large, it will amplify the error.

[0037] Secondly, based on the adjustment system chart, and combined with the penalty factor, the adjustment coefficient is calculated as follows: A predefined modulation reference range corresponding to the surface reflection characteristics of the object being measured in the substation is used to map the modulation of each pixel in the modulation graph to the reference range, thus obtaining the mapped modulation. In this embodiment, the lower limit of the modulation reference range represents the typical modulation of the metallic specular reflective surface. The specific setting process is as follows: after multiple measurements and statistical analysis of common substation metal materials, such as stainless steel panels, the average value is taken as the lower limit of the reference range. The upper limit of the reference range represents the typical modulation of the diffuse reflective surface. The specific setting process is as follows: after multiple measurements and statistical analysis of diffuse reflective materials within the substation, such as white walls, the average value is taken as the upper limit of the reference range. Based on experimental results, in this embodiment, the lower limit of the reference range is set to 0.2, and the upper limit of the reference range is set to 0.8. In other implementation methods, the implementer can set these values ​​according to actual conditions. Secondly, mapping is performed using the maximum-minimum normalization method. The specific mapping process is as follows: ,in, This indicates the modulation index after mapping. The pixel position in the modulation diagram is The adjustment system of the department, This is the lower limit of the reference range. This is the upper limit of the reference range.

[0038] It should be noted that the normalization method for the maximum and minimum values ​​is used to map the adjustment mechanism to... ,when When the mapped modulation is 0, it represents pure specular reflection. When the mapped modulation is 1, it represents pure diffuse reflection.

[0039] The product of the preset correction intensity coefficient, the mapped modulation degree, and the penalty factor is used as the adjustment coefficient for each pixel. In this embodiment, the process of setting the correction intensity coefficient is as follows: for a diffuse reflective surface, the average value of the depth measurement error generated by the grating projection measurement system through multiple tests is used as the correction intensity coefficient. In this embodiment, for a diffuse reflective surface, the average error of multiple measurements is 30%. Therefore, the correction intensity coefficient is set to 0.3 to compensate for system errors. As for other implementation methods, the implementer can set it according to the actual situation.

[0040] Furthermore, based on the adjustment coefficient, the depth values ​​in the depth image are corrected, specifically as follows: The calculation process for the corrected depth value corresponding to each pixel in the depth image is as follows: in, The pixel position in the depth image is The location corresponds to the corrected depth value. The pixel position in the depth image is Depth value at that location, The pixel position in the depth image is Adjustment factor at the location; It should be noted that the modulation degree reflects the systematic bias characteristics of depth measurement. A larger modulation degree indicates higher contrast and sharpness of the grating fringes at that pixel, and more accurate phase demodulation. Systematic errors can be corrected through statistical compensation; a larger correction magnitude, i.e., a higher adjustment coefficient, is needed. Conversely, a smaller modulation degree results in blurrier grating fringes, larger and more unstable phase demodulation errors. In such cases, conservative corrections should be made to avoid further errors, hence a smaller correction magnitude, i.e., a lower adjustment coefficient. The correction range is further constrained by the phase gradient. At object edges, there are abrupt phase changes and depth discontinuities; in these cases, the correction magnitude should be reduced to avoid introducing edge artifacts.

[0041] At this point, the corrected depth values ​​corresponding to each pixel in the depth image are obtained.

[0042] Step 3: Using the corrected depth value, assess the safety of the robot's distance from the obstacle, and combine it with the corresponding modulation information in the modulation map to determine the danger assessment value of each pixel, and project it onto the two-dimensional grid map to generate an obstacle danger map.

[0043] To achieve safe obstacle avoidance, a hazard assessment value is constructed based on the corrected depth value, combined with the voltage level and safety distance requirements of the substation control equipment, to generate an obstacle hazard map, and dynamic path planning is performed based on this map.

[0044] Secondly, traditional obstacle avoidance methods mainly determine whether obstacle avoidance is needed based on the distance to the obstacle. When the distance between the robot and the obstacle is less than the safe distance, obstacle avoidance is triggered, and otherwise it is ignored. However, since the reliability difference of depth measurement is not considered, it is easy to treat the low-quality depth measurement of the metal mirror surface and the high-quality measurement of the diffuse reflection surface as the same. This leads to the depth error caused by metal reflection being misjudged as a real obstacle, resulting in false alarms and conservative obstacle avoidance.

[0045] Based on the above analysis, by judging the corrected depth value and the safe distance, and using the measurement reliability reflected by the modulation index, the hazard assessment value is calculated as follows: If the corrected depth value corresponding to each pixel in the depth image is greater than the preset safety distance, then the danger assessment value of each pixel is 0; otherwise, the danger assessment value is calculated as follows: The ratio of the corrected depth value to the preset safety distance is negatively mapped; the result of the negative mapping is positively fused with the normalized modulation of the corresponding pixel to serve as the risk assessment value for each pixel. In this embodiment, the modulation normalization process is as follows: the ratio of the modulation of each pixel to the upper limit of the reference range is used as the normalized modulation. Therefore, the formula for calculating the risk assessment value of each pixel is: in, The pixel position in the depth image is Hazard assessment values ​​at the location, The pixel position is The location corresponds to the corrected depth value. To establish a safe distance, The pixel position is The adjustment system of the department, This is the upper limit of the reference range; secondly, the highest voltage level of the substation is 220kV, so the safety distance is set at [missing value]. This refers to the minimum permissible distance from the robot's center to the obstacle surface. As an alternative implementation method, the implementer can set this distance according to the actual situation.

[0046] in, To represent the result of negative mapping, as another implementation method, the implementer can use an exponential function to perform negative mapping, The result is used as the result of the negative mapping, where, This represents an exponential function with the natural constant as its base.

[0047] It should be noted that when the corrected depth value is greater than the safe distance, the obstacle is outside the safe range, the hazard assessment value is 0, and the robot does not need to avoid the obstacle; when the corrected depth value is less than or equal to the safe distance, the robot enters the danger zone, and the hazard assessment value needs to be calculated based on the distance and the reliability of the measurement. The distance term quantifies the distance threat posed by obstacles. When the value is 0, meaning the vehicle has just entered the safety boundary, the hazard assessment value is at its lowest. When the time is 1, representing a zero-distance collision, the risk assessment value is the highest. By employing a squared relationship to enhance the response to close-range obstacles, it reflects the safety requirement of more urgent obstacle avoidance for close-range obstacles; secondly... This represents a reliability weighting term, quantifying the credibility of depth measurements. Normalization is performed and used as a reliability weight. For areas with high-profile systems, the depth measurement is reliable, and this item is 1. The hazard assessment value retains its full weight, ensuring accurate detection and timely obstacle avoidance of real obstacles. For areas with low-profile systems, the depth measurement error is large, and the hazard assessment value is significantly reduced to avoid false depth jumps caused by metal reflections being misjudged as obstacles, thus reducing false alarms.

[0048] Then, based on the hazard assessment values, an obstacle hazard map is generated, specifically as follows: The horizontal ground in the environment in front of the robot is divided into a two-dimensional grid map. The hazard assessment values ​​of all pixels in the depth image are projected onto the two-dimensional grid map. The maximum hazard assessment value in each grid of the two-dimensional grid map is used as the hazard level of each grid to generate an obstacle hazard map. In this embodiment, the horizontal ground refers to the XOY plane. Secondly, the resolution of the grid in the two-dimensional grid map is 50mm×50mm. As for other implementation methods, the implementer can set it according to the actual situation.

[0049] At this point, the obstacle and hazard map is obtained.

[0050] Step 4: Based on the hazard situation of different grids in the obstacle hazard map, the total cost function of the A* algorithm is improved, and the obstacle avoidance path of the robot in the substation is planned using the improved total cost function of the A* algorithm.

[0051] Then, based on the hazard assessment values, an obstacle hazard map is generated, specifically as follows: The horizontal ground in the environment in front of the robot is divided into a two-dimensional grid map. The hazard assessment values ​​of all pixels in the depth image are projected onto the two-dimensional grid map. The maximum hazard assessment value in each grid of the two-dimensional grid map is used as the hazard level of each grid to generate an obstacle hazard map. In this embodiment, the horizontal ground refers to the XOY plane. Secondly, the resolution of the grid in the two-dimensional grid map is 50mm×50mm. As for other implementation methods, the implementer can set it according to the actual situation.

[0052] Based on an obstacle hazard map, the A* algorithm is used for path planning. The A* algorithm is a classic heuristic search algorithm used to find the minimum-cost path from the start point to the end point in a graph or state space. By introducing a hazard penalty term into the A* algorithm, the total cost function of the improved A* algorithm is: in, Represents the total cost function. Indicates the distance from the starting point to the node. The actual cost, For nodes The heuristic cost to the destination, The preset risk weight coefficient, In a two-dimensional raster map, the coordinates are... The danger level of the grid, Indicates the distance from the starting point to the node. The complete path; In this embodiment, The term "hazard penalty" is represented by a preset hazard weighting coefficient, which reflects the importance of hazard in the total cost. Its purpose is to make the hazard penalty and path length terms roughly equivalent in magnitude. In this embodiment, assuming a path length of 5000mm, a hazard level of 0.5, and a path containing 100 grids, the total hazard level is approximately 50. For the hazard level to have a significant impact on path selection, it is necessary to... Therefore, it is set to As another implementation method, the implementer can set it according to the actual situation; the cost function enables the planned path to obtain the shortest distance while avoiding high-risk areas in the substation, thus achieving a balance between safety and efficiency.

[0053] Using the A* algorithm, the robot's obstacle avoidance path is output. The robot can safely bypass unknown obstacles and complete the inspection task by moving along the path.

[0054] It should be noted that the A* algorithm is a well-known technique and will not be elaborated upon here.

[0055] Based on the same inventive concept as the above method, this application embodiment also provides a robot autonomous obstacle avoidance device for substation control equipment, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described robot autonomous obstacle avoidance methods for substation control equipment.

[0056] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0057] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0058] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application. Therefore, any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of this application, without departing from the content of the technical solution of this application, shall fall within the protection scope of the technical solution of this application.

Claims

1. A method for autonomous obstacle avoidance by robots for substation control equipment, characterized in that, The method includes the following steps: Using structured light measurement technology, an coded grating pattern is projected onto the environment in front of the robot, and the grating image is acquired. The phase is demodulated to obtain an absolute phase map and a modulation map. Based on the grating image and the principle of triangulation depth estimation, the depth value of each pixel is calculated to generate a depth image. For each pixel, the adjustment coefficient of each pixel is calculated by the gradient change of its absolute phase in the absolute phase map and the corresponding modulation degree of the same pixel in the modulation degree map. The depth value of the pixel in the depth image is corrected. The corrected depth value is used to assess the safety of the robot's distance from the obstacle. Combined with the corresponding modulation degree information in the modulation degree map, the danger assessment value of each pixel is determined and projected onto the two-dimensional grid map to generate an obstacle danger map. Based on the hazard status of different grids in the obstacle hazard map, the total cost function of the A* algorithm is improved. The improved total cost function of the A* algorithm is then used to plan the obstacle avoidance path of the robot in the substation.

2. The method for autonomous obstacle avoidance by a robot for substation control equipment as described in claim 1, characterized in that, The process of obtaining the absolute phase map is as follows: based on the raster image, phase demodulation is performed through an N-step phase shift algorithm, the wrapped phase is calculated, the wrapped phase is expanded, the absolute phase of each pixel is calculated, and the absolute phase map is generated.

3. The method for autonomous obstacle avoidance by a robot for substation control equipment as described in claim 1, characterized in that, The process of obtaining the modulation map is as follows: based on the raster image, the background light intensity and modulation light intensity of each pixel are calculated through an N-step phase shift algorithm. The ratio of the modulation light intensity to the background light intensity is used as the modulation index of each pixel, and a modulation map is generated.

4. The method for autonomous obstacle avoidance by a robot for substation control equipment as described in claim 1, characterized in that, The calculation of the adjustment coefficient for each pixel includes: Based on the absolute phase map, the magnitude of the phase gradient of each pixel is calculated, and its product with the preset attenuation coefficient is negatively mapped to serve as the penalty factor for each pixel. A predefined modulation reference range corresponding to the surface reflection characteristics of the object being measured in the substation is used to map the modulation of each pixel in the modulation graph to the reference range, thus obtaining the mapped modulation. The adjustment coefficient is the result of a positive fusion of a preset correction intensity coefficient, a mapped modulation index, and a penalty factor.

5. The method for autonomous obstacle avoidance by a robot for substation control equipment as described in claim 4, characterized in that, The specific process of positive fusion is as follows: the product of the preset correction intensity coefficient, the mapped modulation degree, and the penalty factor is used as the adjustment coefficient.

6. The method for autonomous obstacle avoidance by a robot for substation control equipment as described in claim 1, characterized in that, The position of a pixel in a depth image is The location corresponds to the corrected depth value. The calculation formula is: ,in, The pixel position in the depth image is Depth value at that location, The pixel position in the depth image is Adjustment coefficient at the location.

7. The method for autonomous obstacle avoidance by a robot for substation control equipment as described in claim 1, characterized in that, The process of determining the danger assessment value of each pixel includes: if the corrected depth value corresponding to each pixel in the depth image is greater than the preset safety distance, then the danger assessment value of each pixel is 0; otherwise, the calculation process of the danger assessment value is: negatively mapping the ratio of the corrected depth value to the preset safety distance; positively fusing the result of the negative mapping with the normalized modulation of the pixel to obtain the danger assessment value of each pixel.

8. The method for autonomous obstacle avoidance by a robot for substation control equipment as described in claim 1, characterized in that, The process of generating the obstacle hazard map is as follows: the horizontal ground in the environment in front of the robot is divided into a two-dimensional grid map, and the hazard assessment values ​​of all pixels in the depth image are projected onto the two-dimensional grid map. The maximum hazard assessment value in each grid of the two-dimensional grid map is used as the hazard level of each grid to generate the obstacle hazard map.

9. A robot autonomous obstacle avoidance method for substation control equipment as described in claim 8, characterized in that, The total cost function of the improved A* algorithm for: in, Indicates the distance from the starting point to the node. The actual cost, For nodes The heuristic cost to the destination, The preset risk weight coefficient, In a two-dimensional raster map, the coordinates are... The danger level of the grid, Indicates the distance from the starting point to the node. The full path.

10. A robotic autonomous obstacle avoidance device for substation control equipment, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the robot autonomous obstacle avoidance method for substation control equipment as described in any one of claims 1-9.

Images