Six-axis patrol operation mechanical arm control method and system of dexterous mobile robot

Abstract

The application relates to the technical field of mechanical arm control, in particular to a six-axis patrol operation mechanical arm control method and system of a dexterous robot, which comprises the following steps: first, obtaining high-precision three-dimensional point clouds by using a laser radar, performing semantic labeling in combination with a public power inspection data set, and converting into a grid map format; obtaining an optimized patrol operation path by using an ant colony optimization algorithm; considering the negative influence caused by turning in the patrol operation path, constructing a turning reward item and a frequency reward item, reconstructing heuristic information of the ant colony optimization algorithm, and planning an optimal patrol operation path root; performing local obstacle avoidance operation between to-be-detected instruments by using an artificial potential field method according to the substation path obtained by using the ant colony optimization algorithm, and controlling the six-axis patrol operation mechanical arm of the dexterous robot to perform substation patrol operation according to the optimized path. The application aims to improve the patrol operation efficiency of the six-axis patrol operation mechanical arm of the dexterous robot.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mechanical arm control, in particular to a six-axis patrol and operation mechanical arm control method and system of a dexterous robot. BACKGROUND

[0002] Under the background of intelligent operation and maintenance of substations, six-axis dexterous robots are gradually becoming the core equipment for inspection and operation. In the current application of dexterous robots through six-axis mechanical arms for patrol and operation, such robots can replace manual work to complete device state inspection, instrument reading recording, bolt tightening and other complex operations by combining the precise control ability of high-degree-of-freedom mechanical arms with multi-modal perception technology, relying on human-like dexterous hand design, which significantly improves the safety and efficiency of work in high-voltage environments. The robot can realize intelligent understanding of the environment through a visual and force perception fusion system, capture device details through a visible light camera, locate overheating hazards through infrared thermal imaging, and construct a three-dimensional space map through a laser radar. Then, through real-time fusion of multi-sensor data by edge computing, it provides a basis for path planning and operation decision-making.

[0003] The use of such technology can not only solve the problems of high labor intensity, many blind spots and high risks in traditional manual inspection, but also provide a data cornerstone for predictive maintenance of substations through standardized operation processes of robots. Currently, when controlling the six-axis mechanical arm of a dexterous robot to patrol and operate the instruments of a substation, a reasonable inspection route is planned in the substation to provide a basis for subsequent detection. After reaching the instrument to be detected, the six-axis mechanical arm is controlled to detect and operate the instrument. Due to the non-uniform and clustered distribution of substations, the conventional ant colony optimization algorithm may have premature convergence and frequent switching of directions when optimizing the path, which can lead to an increase in inspection costs and further problems in ensuring the practicality of the mechanical arm for patrol and operation. SUMMARY

[0004] In view of the above, it is necessary to provide a six-axis patrol and operation mechanical arm control method and system of a dexterous robot to solve the above problems.

[0005] The first aspect of the present application provides a six-axis patrol and operation mechanical arm control method of a dexterous robot, which comprises:

[0006] Obtaining a three-dimensional point cloud map of a substation where the dexterous robot is located, annotating instrument equipment on the three-dimensional point cloud map, dividing the projection plane of the three-dimensional point cloud map according to the annotation result, and obtaining a grid map;

[0007] For each route chosen by an ant between any two instruments to be detected on a grid map in the ant colony algorithm, analyze the number of turns and the angle of each turn to determine the turning reward for each route between any two instruments to be detected.

[0008] For each iteration of the ant colony algorithm, analyze the number of times each route between any two instruments under test is selected by ants, and the number of iterations in which each route is selected, to determine the frequency reward for each route selected by ants between any two instruments under test;

[0009] Based on the turning reward and the frequency reward, the heuristic information of all routes of the ant colony algorithm is reconstructed and substituted into the probability selection formula of the algorithm to obtain the optimal patrol path of the six-axis patrol robot arm of the dexterous hand robot in the substation.

[0010] The six-axis robotic arm of the dexterous hand robot circulates the instrument to be inspected according to the optimal circulation path.

[0011] Specifically, the process of dividing the projection plane of the 3D point cloud map based on the annotation results to obtain a raster map involves:

[0012] The 3D point cloud map is projected along the Z-axis onto the XY plane, semantic labels are preserved, and grid division is performed. The ground is set as a walkable area. The substation infrastructure and instrument base on the 3D point cloud map are marked as obstacles, the instrument to be tested is marked as the point to be tested, and the obstacles are divided into grids at the same scale to obtain a complete grid map.

[0013] The specific formula for determining the turning bonus for each route between any two instruments to be tested is as follows: In the formula, This represents the turning reward for each path chosen by the ants between instruments m and n in the ant colony algorithm. It is an exponential function with the natural constant as its base. This represents the total number of turns made by each ant on the grid map for each route from instrument m to instrument n in the ant colony algorithm. This represents the angle of the i-th turn in the grid for each path taken by an ant from instrument m to instrument n in the ant colony algorithm. For the preset nonlinear weights, Pi is the mathematical constant of a circle.

[0014] Specifically, the turning angle is the angle between the direction of movement of the patrol machine before turning and the direction after turning.

[0015] Specifically, the frequency reward item for determining each route chosen by the ants between any two instruments to be detected is as follows:

[0016] The range of preset attenuation coefficient values ​​is used to determine the attenuation coefficient for the current iteration, taking into account the distribution of iteration counts.

[0017] The frequency reward for each route chosen by the ants between instrument m and instrument n to be tested is denoted as... Its formula is as follows: In the formula, This represents the decay coefficient for the current iteration. Indicates the current iteration number. This indicates the reward weighting based on the frequency of ant appearances. This represents the number of times each route between instrument m and instrument n to be tested is selected by the ants in the current iteration. This represents the total number of ants. This represents the number of iterations in the history of each route between instrument m and instrument n to be tested.

[0018] Specifically, confirming the decay coefficient of the current iteration involves:

[0019] Calculate the range of the attenuation coefficient values; calculate the ratio of the current iteration number to the total number of iterations; calculate the positive fusion result of the ratio and the range value, and add it to the minimum value of the range to obtain the attenuation coefficient of the current iteration.

[0020] Specifically, the process of reconstructing the heuristic information of all routes in the ant colony algorithm is as follows:

[0021] The constructed turning reward term and frequency reward term are normalized and then multiplied together. The result is then positively fused with the negative correlation mapping result of the Euclidean distance between the two corresponding instruments to be detected to obtain the heuristic information correction factor for each route.

[0022] The original heuristic information for each route in the ant colony algorithm is added together with the corresponding heuristic information correction factor to obtain the reconstructed heuristic information.

[0023] Specifically, the heuristic information correction factor for each route is the ratio of the multiplication result obtained for each route to the Euclidean distance between the two corresponding instruments to be tested.

[0024] Specifically, the six-axis roving robotic arm of the dexterous hand robot roves the instrument to be inspected according to the optimal roving path, which is as follows:

[0025] Upon reaching each instrument to be tested, a visible light camera scans the device, and real-time images are transmitted via the dexterous hand robot's communication module. Three-dimensional coordinates are generated using ultrasonic radar, and a hand-eye calibration algorithm is used for coordinate transformation to obtain the device's position in the robotic arm's base coordinate system. The six-axis robotic arm is operated using the dexterous hand robot's remote control handle, and the instruments inside the cabinet are scanned using dual-light cameras. Infrared scanning is used to measure temperature and obtain the test results. The dexterous hand's posture is calculated based on the instrument's surface normal vector, and the robotic arm's node angles are calculated using inverse kinematics to control the robotic arm to precisely operate the instruments and perform a patrol of the instruments to be tested.

[0026] Secondly, embodiments of this application also provide a six-axis roving robotic arm control system for a dexterous hand robot, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0027] This application has at least the following beneficial effects:

[0028] This application obtains a 3D point cloud map of the substation where the dexterous robot is located and annotates the instruments and equipment on the map. This step provides basic data and spatial information for subsequent path planning and inspection tasks. Based on the annotation results, the projection plane of the 3D point cloud map is divided into a grid map to provide a discretized map representation, which facilitates algorithm calculation and path planning. In the ant colony algorithm, each path chosen by the ants is analyzed, and the number of turns and the angle of each turn are counted to determine the turning reward item for the path, encouraging the reduction of unnecessary turns and optimizing the inspection path. The number of times each route is selected in each iteration of the ant colony algorithm is analyzed to determine the frequently occurring paths, and then the frequency reward item is assigned to the frequently selected paths to avoid the path getting trapped in a local optimum and improve the global optimization effect. Combining the turning reward item and the frequency reward item, the heuristic information in the ant colony algorithm is reconstructed and substituted into the probability selection formula of the algorithm to calculate the optimal inspection path. Based on the optimal path, the six-axis inspection arm of the dexterous robot is controlled to inspect according to the planned path, ensuring efficient and low-cost completion of the substation instrument inspection task. Compared to traditional ant colony optimization algorithms, this application reconstructs heuristic information by combining turning reward terms and frequency reward terms. It rewards situations where local road segments with fewer turns occur less frequently, thereby suppressing the increased costs such as time and energy consumption caused by frequent turns and the high frequency of local road segment iterations. This avoids the path getting stuck in local optima while reducing patrol costs and further improving the efficiency of the six-axis patrol arm of the dexterous hand robot in reaching the instrument to be tested. Attached Figure Description

[0029] Figure 1A flowchart illustrating the steps of a six-axis roving manipulator control method for a dexterous hand robot according to an embodiment of this application;

[0030] Figure 2 This is a schematic diagram of the remote control handle and robot front view structure provided in one embodiment of this application. Detailed Implementation

[0031] In the description of the embodiments in this application, the words "exemplary," "or," and "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary," "or," and "for example" is intended to present the relevant concepts in a specific manner.

[0032] 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 belongs. The terminology used in this application's specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0033] It should also be noted that the terms "first" and "second" in this application and its accompanying drawings are used to distinguish similar objects, rather than to describe a specific order or sequence. The methods disclosed in the embodiments of this application or the methods shown in the flowcharts include one or more steps for implementing the method. Without departing from the scope of protection of this application, the execution order of multiple steps can be interchanged, and some steps can also be deleted.

[0034] 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.

[0035] The following description, in conjunction with the accompanying drawings, details the specific scheme of the six-axis roving manipulator control method and system for a dexterous hand robot provided in this application.

[0036] Please see Figure 1 The diagram illustrates a flowchart of a six-axis roving manipulator control method for a dexterous hand robot according to an embodiment of this application. The method includes the following steps:

[0037] The first step is to obtain a 3D point cloud map of the substation where the dexterous robot is located, label the 3D point cloud map with instruments and equipment, and divide the projection plane of the 3D point cloud map according to the labeling results to obtain a raster map.

[0038] Controlling dexterous robots for substation inspections and using six-axis robotic arms to inspect instruments is a common method for automated substation inspections. When controlling the six-axis robotic arm to inspect instruments, it's essential to first plan the robot's path. This allows the robot to move sequentially to the corresponding instrument, providing a foundation for subsequent control of the six-axis robotic arm to inspect and operate all instruments, ultimately achieving efficient inspections.

[0039] Firstly, in the environmental perception stage, it is necessary to obtain high-precision 3D point cloud data of the substation through LiDAR scanning. The obtained high-precision 3D point cloud data is used as input, and SLAM is used to construct a 3D point cloud map of the substation. SLAM is a well-known technology in this field, and the specific process will not be described in detail.

[0040] Considering the presence of dust and other interference noise in the air inside the substation, bilateral filtering is used for noise reduction. Bilateral filtering adds spatial and color weights to Gaussian filtering, which can better preserve edges while reducing noise.

[0041] Substations contain numerous devices. During substation inspections using a six-axis robotic arm, specific inspections and operations are required for different devices, necessitating instrument segmentation. A publicly available power inspection dataset was acquired and divided into training, testing, and validation sets. A PointNet++ deep learning model was used, trained, tested, and validated on the training, testing, and validation sets respectively. The validated model was designated as the substation instrument segmentation model, which includes category labels for all substation instruments in the dataset. This model can be used to label instruments on the denoised 3D point cloud map. The 3D point cloud map was projected along the Z-axis onto the XY plane, retaining semantic labels. The projected plane was divided into grids. The ground was designated as a walkable area, and substation infrastructure and instrument bases were marked as obstacles. Instruments to be inspected were labeled as inspection points, each represented by a grid. Obstacles were also proportionally divided into grids to obtain a complete grid map.

[0042] The second step: For each route chosen by an ant between any two instruments to be detected on the grid map in the ant colony algorithm, analyze the number of turns and the angle of each turn to determine the turning reward for each route between any two instruments to be detected.

[0043] In the task planning section, considering the need to control the six-axis robotic arm of the dexterous hand robot to inspect instruments in the substation, reasonable path planning can improve inspection efficiency. Therefore, path optimization is required during task planning. Given that the inspection of substation instruments is a traveling salesman problem, requiring the inspection and operation of all instruments to be tested, this embodiment uses an ant colony optimization algorithm to solve the traveling salesman problem and optimize the selection of the inspection path.

[0044] In ant colony optimization algorithms, the core of patrol path planning lies in the probabilistic selection formula. This formula combines pheromone factors and heuristic pheromones to guide ants in path selection, calculating the selection probability between two nodes to filter for the optimal path. The pheromone factor represents the pheromone concentration in the path, reflecting the historical "experience" of ants; a higher concentration indicates that the path has been selected more frequently or is considered superior. The heuristic pheromone factor primarily guides ants to choose better paths. Conventional path planning mainly uses the reciprocal of the ant's travel distance as the heuristic pheromone factor. However, in path planning for a six-axis robotic arm controlling a dexterous hand robot to patrol a substation, using only the reciprocal of distance as the heuristic pheromone factor is not suitable. Therefore, the heuristic pheromone factor in the ant colony optimization algorithm needs to be reconstructed.

[0045] When using a six-axis manipulator arm of a dexterous hand robot for inspection operations in a substation, the inspection of instruments can be viewed as a traveling salesman problem, requiring the individual inspection and operation of each instrument to be tested. Because the instrument bases and their distribution in substations vary in shape, adjacent instruments may have paths that are short in distance but require frequent turning. For example, if two instruments are only 5 meters apart in a straight line, but need to navigate around bases or obstacles, the actual path requires multiple turns, and the total movement distance may exceed 10 meters. Even if this path is still the shortest, the frequent turning during machine movement leads to time loss (such as deceleration and acceleration), mechanical wear, and increased energy consumption, resulting in an overall cost higher than other paths. This means that the chosen path may not be the optimal one.

[0046] Therefore, to avoid the problem of circuitous path selection when simply using Euclidean distance as heuristic information, this application constructs a turning reward term to characterize the negative impact of turning on the selection of patrol paths. The specific formula is as follows: In the formula, This represents the turning reward for each route chosen by the ant between instrument m and instrument n under test. It is an exponential function with the natural constant as its base. This represents the total number of turns made by each ant on the grid map for each route from instrument m to instrument n in the ant colony algorithm. In the ant colony algorithm, this represents the angle of the i-th turn of the ant in the grid for each route from instrument m to instrument n. This angle is the angle between the direction of movement before the turn and the direction after the turn. The weight is a non-linear weight with a value range of [1.0, 4.0]. In this embodiment, it is set to 2 to further enhance the reward difference between large-angle steering and small-angle steering, while maintaining balance.

[0047] It should be understood that when the ant's route between two instruments does not involve turning, i.e., when the ant travels in a straight line from instrument m to instrument n, the turning reward value is set to 1. In the patrol path, the smaller the turning angle and the fewer the turns in the section between two instruments, the closer the calculated turning reward value A is to 1. Conversely, the larger the turning angle and the more turns in the section between two instruments, the greater the calculated turning reward value A is and much less than 1. In other words, the larger the turning angle and the more turns in the section between instruments, the more severe the negative impact on the patrol path. Combining heuristic information can further reduce the selection weight of the path.

[0048] The third step: For each iteration of the ant colony algorithm, analyze the number of times each route between any two instruments under test is selected by ants, and the number of iterations in which each route is selected, to determine the frequency reward for each route selected by ants between any two instruments under test.

[0049] Within a substation area, due to the unique distribution of instruments, they may be concentrated in certain areas (such as the distribution cabinet area), resulting in a non-uniform, multi-cluster distribution. Using distance as the heuristic function alone can lead to paths guided by distance repeatedly traversing dense areas. In different iterations, the ant colony may repeatedly visit paths in dense areas due to pheromone monopoly, easily getting trapped in local cluster loops. This reduces the likelihood of the ant colony exploring new routes from dense to sparse areas, leading to a local optimum. For example, if a dense area contains adjacent instruments 1, 2, and 3, and another area at a certain distance contains adjacent instruments 4 and 5, then after multiple iterations, the path may repeatedly appear as instrument 1 → instrument 2 → instrument 3 → instrument 4 → instrument 5. The ant colony gathers on this path too early, making it difficult to escape the loop and obtain other path solutions, causing the path to get trapped in a local optimum.

[0050] Based on the above analysis, this application constructs a frequency reward term to characterize the overall negative impact of combining the number of ants appearing in the corresponding road segment in the ant colony and the frequency of the path containing the corresponding road segment in the iteration on path selection. Specifically: a preset range of attenuation coefficient values ​​is defined, and the range of attenuation coefficient values ​​is calculated; the ratio of the current iteration number to the total number of iterations is calculated; the positive fusion result of the ratio and the range value is calculated and added to the minimum value of the range to obtain the attenuation coefficient of the current iteration. In this embodiment, the positive fusion of multiple variables is performed using a multiplication calculation method.

[0051] The frequency reward for each route chosen by the ants between instrument m and instrument n to be tested is denoted as... Its formula is as follows: In the formula, This represents the decay coefficient for the current iteration, controlling the sensitivity of the iteration frequency reward. Its value ranges from [0.1, 1.0] and is dynamically adjusted based on the iteration count. Indicates the current iteration number. This represents the reward weight that adjusts the frequency of ant appearances, with a value range of [0.1, 0.5]. In this embodiment, the value is 0.3. This represents the number of times each route between instrument m and instrument n to be tested is selected by the ants in the current iteration. This represents the total number of ants. This represents the number of iterations in the history of each route between instrument m and instrument n being tested, where each iteration is considered as one when any number of ants select each route between instrument m and instrument n.

[0052] It should be understood that when the number of ants passing through the section between the two instruments being tested is greater, and the frequency of that section appearing in the iteration is higher, the frequency reward term B value obtained is smaller, and the result will be greater than 0 and much less than 1. When the number of ants passing through the section between the two instruments being tested is less, and the frequency of that section appearing in the iteration is lower, the frequency reward term B value obtained is larger, and the result is closer to 1.

[0053] The fourth step: Based on the turning reward and the frequency reward, the heuristic information of all routes in the ant colony algorithm is reconstructed and substituted into the probability selection formula of the algorithm to obtain the optimal patrol path of the six-axis patrol robot arm of the dexterous hand robot in the substation.

[0054] The heuristic information of the ant colony optimization algorithm is reconstructed using two normalized metrics. Specifically, the constructed turning reward term and frequency reward term are normalized separately, multiplied together, and then divided by the Euclidean distance between the two instruments to be tested to obtain the heuristic information correction factor. This factor is then added to the original heuristic information from the ant colony algorithm to obtain the heuristic information for each route of the two instruments to be tested. The original heuristic information from the ant colony algorithm is specifically the reciprocal of the Euclidean distance between the two instruments to be tested.

[0055] It should be understood that using the normalized turning reward item and frequency reward item road segment length for rewarding, and suppressing road segments with many turning situations and high frequency of occurrence, can more accurately plan reasonable patrol routes.

[0056] After reconstructing the heuristic information in the ant colony optimization algorithm, in the ant colony optimization algorithm, when ants choose a path, they use a probability selection formula to determine the direction of their next move. The probability is determined by the pheromone concentration and heuristic information. The higher the pheromone accumulated on the path, the greater the probability of being selected. The heuristic information is inversely proportional to the cost. The greater the heuristic information value of the path, the greater the probability of being selected.

[0057] After reconstructing the heuristic information, the new heuristic information is substituted into the algorithm's probability selection formula. Simultaneously, pheromone weights and heuristic information weights are set for the algorithm. This application mainly solves the path planning problem for controlling a six-axis manipulator arm of a dexterous hand robot during substation patrols. This is a traveling salesman problem, therefore the heuristic function is the focus, and the pheromone weights are... The value range is [1,2], and it is set to 1, which is the heuristic information weight. The value range is [2,5], and it is set to 3, which is the pheromone evaporation rate. The value range is [0.1, 0.5]. It is set to 0.3. The number of ants is related to the number of instruments to be tested in the substation. Based on the number of instruments to be tested in the substation, the number of ants is set to 10 times the number of instruments to be tested in the substation to ensure that the data volume will not be deviated due to the insufficient number of ants. At the same time, the computational load is balanced. The number of iterations is set to 300. The optimal patrol path is calculated and output by using the configured ant colony optimization algorithm on the substation grid map with the instruments to be tested calibrated. The calculation process of the ant colony optimization algorithm is a well-known technology, and the specific calculation process will not be described in detail here.

[0058] The fifth step: Control the six-axis roving robotic arm of the dexterous hand robot to roving along the optimal roving path to inspect the instrument.

[0059] After planning the path for the six-axis manipulator arm controlling the dexterous hand robot to perform substation inspection operations, considering that there may be obstacles in the path when moving from one instrument to another, obstacle avoidance is required. The robot's lidar and depth camera are used to acquire information between two inspection points, and an artificial potential field method is used for local obstacle avoidance. The gravity function is set to linear gravity, the target gravity gain is set within the range of [0.5, 2.0] (1.0 in this embodiment), the repulsion gain is set within the range of [0.5, 3.0] (1.5 in this embodiment), the repulsion influence distance is set within the range of [0.5, 2.0] meters (1.0 in this embodiment), and the minimum safe distance is set to the robot radius plus 0.3 meters.

[0060] After setting parameters, obstacle avoidance planning is performed for each detection point. The robot then performs motion inspection operations using a path optimized by an ant colony optimization algorithm. Upon reaching each detection point, the robot's built-in visible light camera scans and captures images of the switchgear and protection devices. The real-time images are transmitted to the display via the robot's mainboard, communication module, and remote control mainboard. The robot generates the corresponding 3D coordinates of the targets identified by the camera using ultrasonic radar. The Tsai-Lenz algorithm is used for coordinate transformation, converting the output into the instrument and equipment position coordinates in the 3D robotic arm's base coordinate system. The Tsai-Lenz algorithm is a well-known technology and will not be elaborated upon here. The six-axis robotic arm is operated using a remote control handle. A schematic diagram of the remote control handle and the robot's main view structure is shown below. Figure 2 As shown in the schematic diagram of the robot's main view structure, it includes a mobile component 1, a horizontal transfer platform 101, a brake wheel 102, a SLAM laser navigation system 103, an ultrasonic radar 104, a visible light camera 105, a control component 2, a housing 201, a remote control motherboard 202, a control handle 203, a display 204, a communication module 205, a lifting component 3, a vertical telescopic mechanism 301, a carbon fiber telescopic cylinder 302, a bottom mounting plate 303, an operating component 4, a 6-axis robotic arm 401, a dual-light camera 402, and a bionic dexterous hand 403.

[0061] A dual-light camera positioned above the robotic arm is aimed at the observation window of the light-emitting cabinet. The camera captures images of the instruments inside and uses infrared radiation to scan and measure their temperature, obtaining the detection results and identifying the parts to be manipulated. A control handle is used to control the six-axis robotic arm to operate the instruments. The dexterous hand's posture is set by default to a rotation matrix (Euler angles in XYZ order, calculated based on the surface normal vector of the part to be manipulated) aligned with the Z-axis of the target instrument. To facilitate subsequent manipulation by the bionic dexterous hand, the coordinates of the dexterous hand's vertex, the center position coordinates of the instrument to be manipulated, and the dexterous hand's posture are used as inputs. Inverse kinematics calculations are performed on the bending angle of each node of the robotic arm, outputting the angle to be transformed at each node. Inverse kinematics is a well-known technique, and the specific calculation process will not be elaborated further. The angle corresponding to each node is used as input to control the robotic arm to reach the instrument to be manipulated. The dexterous hand is then used to operate the instrument. Through the above operations, the six-axis robotic arm performs a roving operation on all the instruments to be tested.

[0062] Based on the same inventive concept as the above method, this application embodiment also provides a six-axis roving manipulator control system for a dexterous hand robot, 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 methods for controlling a six-axis roving manipulator for a dexterous hand robot.

[0063] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description; sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0064] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from its essential characteristics. Therefore, the embodiments described above should be considered exemplary and non-limiting in all respects; modifications to the technical solutions described in the foregoing embodiments, or equivalent substitutions of some technical features, without causing the essence of the corresponding technical solutions to deviate from the scope of the technical solutions in the embodiments of this application, should all be included within the protection scope of this application.

Claims

1. A control method for a six-axis roving robotic arm of a dexterous hand robot, characterized in that, The method includes the following steps: Obtain a 3D point cloud map of the substation where the dexterous robot is located, annotate the 3D point cloud map with instruments and equipment, and divide the projection plane of the 3D point cloud map according to the annotation results to obtain a raster map. For each route chosen by an ant between any two instruments to be detected on a grid map in the ant colony algorithm, analyze the number of turns and the angle of each turn to determine the turning reward for each route between any two instruments to be detected. For each iteration of the ant colony algorithm, analyze the number of times each route between any two instruments under test is selected by ants, and the number of iterations in which each route is selected, to determine the frequency reward for each route selected by ants between any two instruments under test; Based on the turning reward and the frequency reward, the heuristic information of all routes of the ant colony algorithm is reconstructed and substituted into the probability selection formula of the algorithm to obtain the optimal patrol path of the six-axis patrol robot arm of the dexterous hand robot in the substation. The six-axis robotic arm of the dexterous hand robot circulates the instrument to be tested according to the optimal circulation path. The specific formula for determining the turning bonus for each route between any two instruments to be tested is as follows: In the formula, This represents the turning reward for each path chosen by the ants between instruments m and n in the ant colony algorithm. It is an exponential function with the natural constant as its base. This represents the total number of turns made by each ant on the grid map for each route from instrument m to instrument n in the ant colony algorithm. This represents the angle of the i-th turn in the grid for each path taken by an ant from instrument m to instrument n in the ant colony algorithm. For the preset nonlinear weights, Pi; The frequency reward item for determining each route chosen by the ants between any two instruments to be detected is specifically as follows: The range of preset attenuation coefficient values ​​is used to determine the attenuation coefficient for the current iteration, taking into account the distribution of iteration counts. The frequency reward for each route chosen by the ants between instrument m and instrument n to be tested is denoted as... Its formula is as follows: In the formula, This represents the decay coefficient for the current iteration. Indicates the current iteration number. This indicates the reward weighting based on the frequency of ant appearances. This represents the number of times each route between instrument m and instrument n to be tested is selected by the ants in the current iteration. This represents the total number of ants. This represents the number of iterations in the history of each route between instrument m and instrument n to be tested.

2. The control method for a six-axis roving robotic arm of a dexterous hand robot as described in claim 1, characterized in that, The process of dividing the projection plane of the 3D point cloud map based on the annotation results to obtain a raster map is as follows: The 3D point cloud map is projected along the Z-axis onto the XY plane, semantic labels are preserved, and grid division is performed. The ground is set as a walkable area. The substation infrastructure and instrument base on the 3D point cloud map are marked as obstacles, the instrument to be tested is marked as the point to be tested, and the obstacles are divided into grids at the same scale to obtain a complete grid map.

3. The control method for a six-axis roving robotic arm of a dexterous hand robot as described in claim 1, characterized in that, The turning angle is specifically the angle between the direction of movement of the patrol machine before turning and the direction after turning.

4. The control method for a six-axis roving robotic arm of a dexterous hand robot as described in claim 1, characterized in that, The confirmation of the decay coefficient for the current iteration is specifically as follows: Calculate the range of the decay coefficient values; calculate the ratio of the current iteration number to the total number of iterations; The positive fusion result of the ratio and the range is calculated and added to the minimum value of the range to obtain the decay coefficient of the current iteration.

5. The control method for a six-axis roving robotic arm of a dexterous hand robot as described in claim 1, characterized in that, The process of reconstructing the heuristic information of all routes in the ant colony algorithm is as follows: The constructed turning reward term and frequency reward term are normalized and then multiplied together. The result is then positively fused with the negative correlation mapping result of the Euclidean distance between the two corresponding instruments to be detected to obtain the heuristic information correction factor for each route. The original heuristic information for each route in the ant colony algorithm is added together with the corresponding heuristic information correction factor to obtain the reconstructed heuristic information.

6. The control method for a six-axis roving robotic arm of a dexterous hand robot as described in claim 5, characterized in that, The heuristic information correction factor for each route is specifically the ratio of the product of the results obtained for each route to the Euclidean distance between the two corresponding instruments to be tested.

7. The control method for a six-axis roving robotic arm of a dexterous hand robot as described in claim 1, characterized in that, The six-axis robotic arm controlling the dexterous hand robot traverses the instrument to be inspected according to the optimal traversal path, specifically as follows: Upon reaching each instrument to be tested, a visible light camera scans the device, and real-time images are transmitted via the dexterous hand robot's communication module. Three-dimensional coordinates are generated using ultrasonic radar, and a hand-eye calibration algorithm is used for coordinate transformation to obtain the device's position in the robot's base coordinate system. The six-axis robot is operated using the robot's remote control handle, and the instruments inside the cabinet are scanned using dual-light cameras. Infrared scanning is used to measure temperature and obtain the test results. The dexterous hand's posture is calculated based on the instrument's surface normal vector, and the robot's node angles are calculated using inverse kinematics. This allows the robot to precisely operate the instruments, completing the inspection of all instruments to be tested.

8. A six-axis roving robotic arm control system for a dexterous hand robot, 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 method as described in any one of claims 1-7.

Images