Robot-based cable duct detection and obstacle removal method and robot-based cable duct detection and obstacle removal system

Abstract

The invention relates to the technical field of pipeline obstacle clearance, and discloses a cable pipeline detection obstacle clearance method and system based on a robot, and the method comprises the steps: obtaining multi-source data in a cable pipeline, and carrying out the data processing and feature analysis, and obtaining obstacle information; generating a tool configuration adjustment instruction according to the tool parameters of the obstacle clearing tool and the obstacle information; according to the counter-acting force and the friction coefficient when the obstacle removing robot makes contact with the obstacle, operation force control parameters and a tool movement track are generated; and obtaining a particle distribution image in the cable pipeline, performing image processing to obtain a particle cleaning planning path, and generating a particle cleaning instruction. According to the invention, through multi-mode sensing fusion, dynamic optimization decision and output motion cooperative control, the efficiency of cable conduit obstacle clearing work is effectively improved, the obstacle clearing cost is reduced, the safety of the cable conduit in the obstacle clearing process is ensured, and an effective solution is provided for intelligent operation and maintenance of an urban pipe network.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline obstacle removal, and particularly to a cable pipeline detection and obstacle removal method and system based on a robot. Background Art

[0002] As a core component of urban infrastructure, cable pipelines undertake the key tasks of power and communication transmission. With the acceleration of urbanization, the types and complexity of obstacles in the pipelines are constantly increasing. Removing these obstacles has become an important issue for maintaining the pipeline function. The cleaning and maintenance of the interior of cable pipelines will affect the stability and safety of system operation.

[0003] Currently, the commonly used method for cleaning cable pipelines is pipeline robots. Since there may be obstacles with different hardnesses such as sand and gravel, and cement blocks inside the cable pipelines, the types of tools and the operating forces required for different obstacles vary. Most of the existing pipeline robots adopt fixed end tools. This fixed end tool method is difficult to adapt to obstacles with significant hardness differences and lacks the ability to quickly and accurately switch obstacle removal tools according to the actual obstacle situation. Frequent tool switching not only leads to low efficiency of obstacle removal work, but also easily causes scratches or secondary damage to the pipe wall. During the obstacle removal operation, the reaction force and the fluctuation of the friction coefficient are relatively large when the tool contacts the obstacle. The current robot control cannot adjust the operation parameters in real time according to the actual situation, resulting in unstable obstacle removal. In addition, the current pipeline robots lack intelligent planning of the cleaning path for the particles generated by crushing obstacles, further affecting the obstacle removal efficiency. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides a cable pipeline detection and obstacle removal method and system based on a robot, which can solve the problems of low tool switching efficiency and insufficient control accuracy of existing pipeline robots, achieve precise control of cable pipeline obstacle removal robots, and improve the cable pipeline obstacle removal efficiency.

[0005] In a first aspect, the present invention provides a cable pipeline detection and obstacle removal method based on a robot, the method comprising: Obtaining multi-source data inside the cable pipeline, and performing data processing and feature analysis on the multi-source data to obtain obstacle information, where the obstacle information includes obstacle type and obstacle position; According to the tool parameters of a preset obstacle removal tool and the obstacle information, using a multi-objective optimization algorithm to calculate an optimal tool switching sequence, and generating a tool configuration adjustment instruction according to the optimal tool switching sequence, and sending the instruction to the obstacle removal robot; In response to the obstacle clearing robot executing the tool configuration adjustment instruction to replace the obstacle clearing tool, generate an operation force control parameter and a tool movement trajectory based on the reaction force and friction coefficient when the obstacle clearing robot contacts the obstacle, generate a control instruction according to the operation force control parameter and the tool movement trajectory, and send it to the obstacle clearing robot; In response to the obstacle clearing robot completing the execution of the control instruction, acquire the particle distribution image inside the cable duct, perform image processing, obtain the particle cleaning planning path, generate a particle cleaning instruction according to the particle cleaning planning path, and send it to the obstacle clearing robot to execute the cleaning operation.

[0006] Furthermore, the steps of acquiring multi-source data inside the cable duct, performing data processing and feature analysis on the multi-source data, and obtaining obstacle information, where the obstacle information includes the obstacle type and the obstacle position, include: Collect the laser point cloud data, acoustic wave data, and infrared data based on obstacles inside the cable duct, perform data processing and particle filtering, and obtain a three-dimensional obstacle feature map; According to the preset material matching rule and the three-dimensional obstacle feature map, perform obstacle recognition to obtain the obstacle type and obstacle position of each obstacle.

[0007] Furthermore, the steps of calculating the optimal tool switching sequence by using a multi-objective optimization algorithm according to the tool parameters of the preset obstacle clearing tool and the obstacle information include: According to the obstacle type, determine the obstacle hardness, and extract the tool hardness, tool remaining life, and tool expected life of each obstacle clearing tool from the preset tool database; Calculate the tool matching coefficient according to the obstacle hardness, the tool hardness, the tool remaining life, and the tool expected life; Taking the minimization of tool cost and the minimization of tool switching time as the objective functions, and taking the tool matching coefficient and the tool remaining life as the constraint conditions, establish a multi-objective optimization model; Solve the multi-objective optimization model to obtain the optimal tool switching sequence.

[0008] Furthermore, the tool matching coefficient is represented by the following formula: In the formula, represents the tool matching coefficient between the i-th obstacle clearing tool and the j-th obstacle type, S i represents the tool hardness of the i-th obstacle clearing tool, S j represents the obstacle hardness of the j-th obstacle type, α1 represents the hardness coefficient, M s,i represents the tool remaining life of the i-th obstacle clearing tool, Mi denotes the expected tool life of the i-th obstacle removal tool, and α2 denotes the life coefficient; The objective function is expressed by the following formula: In the formula, n represents the total number of obstacle removal tools, C i denotes the unit price of the i-th obstacle removal tool, N i denotes the number of uses of the i-th obstacle removal tool, T s,i denotes the switching time of the i-th obstacle removal tool, T o,i denotes the operation time of the i-th obstacle removal tool, C total denotes the tool cost, T total denotes the tool switching time; The constraint conditions are expressed by the following formula: In the formula, denotes the coefficient threshold, M t denotes the life threshold.

[0009] Further, the step of generating the operation force control parameter and the tool motion trajectory according to the reaction force and the friction coefficient when the obstacle removal robot contacts the obstacle includes: Determine the target force according to the obstacle type, and calculate the force error and the error change rate according to the reaction force when the obstacle removal robot contacts the obstacle and the target force; Perform fuzzy PID control on the force error and the error change rate to obtain the output power; Perform friction compensation on the output power according to the friction coefficient when the obstacle removal robot contacts the obstacle to obtain the final output power, and convert the final output power into the motor speed parameter and the motor torque parameter; Generate an initial path according to the obstacle position, and perform curvature smoothing on the initial path to obtain a smooth path; Adjust the initial motion speed according to the friction coefficient to obtain the motion speed; Obtain the tool motion trajectory of the obstacle removal robot according to the smooth path and the motion speed.

[0010] Further, the step of obtaining the particle distribution image inside the cable duct and performing image processing to obtain the particle cleaning planning path includes: Obtain the particle distribution image inside the cable duct, generate a particle density map through the region growing algorithm, and rasterize the particle density map to obtain a rasterized network; Calculate the particle density weight and the dynamic weight factor according to the rasterized network; Construct a cost function based on the path planning algorithm according to the particle density weight and the dynamic weight factor, and solve it to obtain the particle cleaning planning path.

[0011] Furthermore, the particle density weight is expressed by the following formula: In the formula, d(n) represents the particle density weight of node n, ρ(n) represents the grid density of node n, and ρ t represents the average grid density; The dynamic weight factor is expressed by the following formula: In the formula, w(t) represents the dynamic weight factor at time t, represents the particle density at time t, represents the maximum particle density; The cost function is expressed by the following formula: In the formula, f(n) represents the cost function of node n, g(n) represents the actual cost from the starting point to node n, h(n) represents the estimated cost from node n to the end point, w(t) represents the dynamic weight factor at time t, d(n) represents the particle density weight of node n, and β represents the density coefficient.

[0012] Furthermore, after the step of generating a control instruction according to the operation force control parameter and the tool movement trajectory and sending it to the obstacle clearing robot, the following steps are further included: In response to the obstacle clearing robot executing the control instruction, obtain the laser scan data of the cable duct, filter and denoise the laser scan data, and generate a pipe wall contour image through three-dimensional reconstruction; Input the pipe wall contour image into a preset damage classification model to obtain a damage classification result, and the damage classification model is constructed based on a convolutional neural network model; Extract the damage features of the corresponding area from the laser scan data according to the damage classification result, and the damage features include scratch depth and damage area; Determine the damage level according to the damage area, and adjust the tool movement trajectory according to the damage level; Adjust the output power of the obstacle clearing robot according to the scratch depth.

[0013] Furthermore, after the step of generating a control instruction according to the operation force control parameter and the tool movement trajectory and sending it to the obstacle clearing robot, the following steps are further included: Calculate the actual stress based on the reaction force and the tool contact area; Calculate the safety factor based on the allowable stress of the cable duct and the actual stress; Adjust the output power of the obstacle removal robot according to the safety factor; Among them, the following formula is used to represent the actual stress: In the formula, represents the actual stress, F ac represents the reaction force, A represents the tool contact area, and K represents the stress coefficient; The following formula is used to represent the safety factor: In the formula, G represents the safety factor, represents the allowable stress.

[0014] In a second aspect, the present invention provides a cable duct detection and obstacle removal system based on a robot, and the system includes: An obstacle classification module, configured to obtain multi-source data inside the cable duct, perform data processing and feature analysis on the multi-source data, and obtain obstacle information, where the obstacle information includes obstacle types and obstacle positions; A tool instruction generation module, configured to calculate an optimal tool switching sequence according to tool parameters of a preset obstacle removal tool and the obstacle information, generate a tool configuration adjustment instruction according to the optimal tool switching sequence, and send it to the obstacle removal robot; An obstacle removal instruction generation module, configured to generate an operation force control parameter and a tool movement trajectory according to the reaction force and the friction coefficient when the obstacle removal robot contacts the obstacle in response to the obstacle removal robot executing the tool configuration adjustment instruction to replace the obstacle removal tool, generate an obstacle removal control instruction according to the operation force control parameter and the tool movement trajectory, and send it to the obstacle removal robot; A cleaning instruction generation module, configured to obtain a particle distribution image inside the cable duct and perform image processing to obtain a particle cleaning planning path in response to the obstacle removal robot executing the control instruction, generate a particle cleaning instruction according to the particle cleaning planning path, and send it to the obstacle removal robot to perform a cleaning operation.

[0015] The present invention provides a method and system for cable pipeline detection and obstacle clearance based on a robot. Through multi-sensor data fusion modeling, the present invention can improve the accuracy of obstacle recognition inside the cable pipeline; through a multi-objective optimization algorithm to optimize the dynamic scheduling of tools, the tool switching efficiency of the obstacle clearance robot can be improved, and the obstacle clearance cost can be reduced; through fuzzy PID control to dynamically regulate the output power of the obstacle clearance robot, precise control of the obstacle clearance robot can be achieved, and the obstacle clearance efficiency can be improved; through intelligent particle cleaning path planning, the cleaning efficiency can be improved; and through a closed-loop protection mechanism for pipe wall damage, the safety of the cable pipeline during the obstacle clearance process can be ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a schematic flowchart of a method for cable pipeline detection and obstacle clearance based on a robot according to an embodiment of the present invention; Figure 2 is a schematic structural diagram of a system for cable pipeline detection and obstacle clearance based on a robot according to an embodiment of the present invention; Reference Signs: 10, obstacle classification module; 20, tool instruction generation module; 30, obstacle clearance instruction generation module; 40, cleaning instruction generation module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0018] Please refer to Figure 1 , a method for cable pipeline detection and obstacle clearance based on a robot proposed in the first embodiment of the present invention, including steps S10 to S40: Step S10, obtain multi-source data inside the cable pipeline, perform data processing and feature analysis on the multi-source data to obtain obstacle information, where the obstacle information includes obstacle type and obstacle position; Step S20, according to the tool parameters of the preset obstacle clearance tool and the obstacle information, use a multi-objective optimization algorithm to calculate the optimal tool switching sequence, generate a tool configuration adjustment instruction according to the optimal tool switching sequence, and send it to the obstacle clearance robot; Step S30: In response to the obstacle removal robot executing the tool configuration adjustment instruction to replace the obstacle removal tool, generate an operation force control parameter and a tool movement trajectory based on the reaction force and friction coefficient when the obstacle removal robot contacts the obstacle, generate a control instruction according to the operation force control parameter and the tool movement trajectory, and send it to the obstacle removal robot; Step S40: In response to the obstacle removal robot completing the execution of the control instruction, obtain the particle distribution image inside the cable duct, perform image processing to obtain the particle cleaning planning path, generate a particle sweeping instruction according to the particle cleaning planning path, and send it to the obstacle removal robot to perform the sweeping operation.

[0019] In the present invention, for the obstacles inside the cable duct, multiple sensors are used for data collection, and through multi-data fusion and analysis, the types of obstacles inside the duct are determined. The specific steps include: Collect the laser point cloud data, acoustic data, and infrared data based on the obstacles inside the cable duct, and perform data processing and particle filtering to obtain a three-dimensional obstacle feature map; According to the preset material matching rules and the three-dimensional obstacle feature map, perform obstacle recognition to obtain the obstacle types and obstacle positions of each obstacle.

[0020] In this embodiment, the obstacle information inside the duct is collected through laser scanning, acoustic detection, and infrared thermal imaging technology. Among them, for the laser point cloud data, the concave and convex regions are distinguished by calculating the local point cloud density, so as to obtain the surface structure features, and the texture features are extracted by calculating the adjacent point spacing. For example, a spacing less than 1 mm is a smooth feature, and a spacing greater than 5 mm is a rough feature, etc.; for the acoustic data, since the reflection time of sound waves for materials with different hardnesses is different, the material hardness of the obstacle can be calculated through the reflection time of ultrasonic waves; for the infrared data, since the heat conduction curves of different materials are different, the overall heat conduction curve in a preset band is analyzed, and the infrared material data is matched from the preset material database.

[0021] For the surface structure features, texture features, acoustic wave hardness values, and infrared material data after data processing, the particle filter method is used to correct the non-linear errors. For the filtered multi-source data, feature vectors are generated through calculation. The feature vectors include shape feature vectors and material feature vectors. Among them, the shape feature vectors include curvature, normal vectors, and contour curves, and the material feature vectors include hardness values, thermal conductivity coefficients, and reflectivity, etc. Finally, based on these feature vectors, a three-dimensional obstacle feature map is generated. The feature map contains fusion data such as position, shape, hardness, and material labels. For the three-dimensional obstacle feature map, through material matching rules and classification algorithms, such as decision tree classification algorithms, the obstacle material classification is carried out, and finally, obstacle information data such as the type, position coordinates, and size parameters of each obstacle are output.

[0022] After determining the relevant information such as the type and position of the obstacles inside the cable duct, it is also necessary to select the corresponding obstacle removal tools according to the obstacle information, and generate a tool switching sequence and corresponding tool configuration adjustment instructions according to different obstacle removal tools. The specific steps include: According to the type of the obstacle, determine the hardness of the obstacle, and extract the tool hardness, tool remaining life, and tool expected life of each obstacle removal tool from the preset tool database; According to the obstacle hardness, the tool hardness, the tool remaining life, and the tool expected life, calculate the tool matching coefficient; Taking the minimum tool cost and the minimum tool switching time as the objective functions, and taking the tool matching coefficient and the tool remaining life as the constraint conditions, establish a multi-objective optimization model; Solve the multi-objective optimization model to obtain the optimal tool switching sequence.

[0023] In this embodiment, according to the type of the obstacle in the obstacle information, the hardness data of the obstacle can be determined. At the same time, a plurality of obstacle removal tools required for the obstacle removal work of the obstacle removal robot and the corresponding tool parameters are stored in the preset tool database. Among them, these obstacle removal tools are used as the end effectors of the obstacle removal robot to contact the obstacles for obstacle removal operations, such as diamond tools, cemented carbide tools, and high-speed steel tools, etc.

[0024] Extract the tool hardness, tool remaining life, and tool expected life of the obstacle removal tools from the tool database, and combine the obstacle hardness to calculate the tool matching coefficient corresponding to each obstacle removal tool. Among them, the tool matching coefficient can be expressed as: In the formula, represents the tool matching coefficient between the i-th obstacle removal tool and the j-th obstacle type, S i represents the tool hardness of the i-th obstacle removal tool, Sj represents the hardness of the obstacle of the j-th obstacle type, α1 represents the hardness coefficient, M s,i represents the remaining tool life of the i-th obstacle removal tool, M i represents the expected tool life of the i-th obstacle removal tool, and α2 represents the life coefficient.

[0025] For different types of obstacles, different obstacle removal tools are required. To ensure the highest switching efficiency and the lowest usage cost when the obstacle removal robot switches tools, this embodiment uses a multi-objective optimization algorithm to calculate the optimal tool switching sequence. Specifically, with the minimization of tool cost and the minimization of tool switching time as the objective functions, and with the tool matching coefficient and the remaining tool life as the constraint conditions, a multi-objective optimization model is constructed. Among them, the objective functions can be expressed as: In the formula, n represents the total number of obstacle removal tools, C i represents the unit price of the i-th obstacle removal tool, N i represents the number of uses of the i-th obstacle removal tool, T s,i represents the switching time of the i-th obstacle removal tool, T o,i represents the operation time of the i-th obstacle removal tool, C total represents the tool cost, T total represents the tool switching time.

[0026] The constraint conditions can be expressed as: In the formula, represents the coefficient threshold, M t represents the life threshold.

[0027] For the above multi-objective optimization model, this embodiment preferably uses the NSGA-II algorithm for solution to screen out the Pareto optimal solution set. The specific solution steps can refer to the conventional NSGA-II algorithm solution steps and will not be elaborated here.

[0028] After obtaining the optimal tool switching sequence, according to the optimal tool switching sequence, determine the station position of the obstacle removal tool in the sequence in the obstacle removal robot. Assuming that the tool library structure of the obstacle removal robot is a multi-station circular arrangement structure, according to the station serial number of the obstacle removal tool in the sequence, calculate the current target station angle. For example, if the first obstacle removal tool in the sequence is on station 3 and the current target station angle with station 3 is 90 degrees, then the execution position of the obstacle removal robot needs to rotate 90 degrees, and a tool configuration adjustment instruction is generated according to the rotation angle and sent to the obstacle removal robot to achieve tool switching.

[0029] After the tool switching is completed, in order to improve the control accuracy of the obstacle clearing robot, in this embodiment, by analyzing and processing the reaction force and friction coefficient when the obstacle clearing robot contacts the obstacle, the control instructions of the obstacle clearing robot are dynamically adjusted. The specific steps include: Determine the target force according to the obstacle type, and calculate the acting force error and error change rate according to the reaction force when the obstacle clearing robot contacts the obstacle and the target force; Perform fuzzy PID control on the acting force error and the error change rate to obtain the output power; According to the friction coefficient when the obstacle clearing robot contacts the obstacle, perform friction compensation on the output power to obtain the final output power, and convert the final output power into motor speed parameters and motor torque parameters; Generate an initial path according to the obstacle position, and perform curvature smoothing on the initial path to obtain a smooth path; Adjust the initial movement speed according to the friction coefficient to obtain the movement speed; According to the smooth path and the movement speed, obtain the tool movement trajectory of the obstacle clearing robot.

[0030] In this embodiment, the reaction force and friction coefficient when the obstacle clearing robot contacts the obstacle are obtained through the force feedback sensor array and friction coefficient sensor installed on the end effector of the obstacle clearing robot. For the reaction force, sliding average filtering is used to filter the data, and for the friction coefficient, median filtering is performed to eliminate sudden noise. For the filtered reaction force, according to the target force, the real-time acting force error and error change rate are calculated. Among them, the target force is the acting force determined based on the obstacle material type, and the acting force error can be expressed as: In the formula, e(t) represents the acting force error at time t, F mb represents the target force, F ac represents the reaction force, that is, the actual reaction force after filtering.

[0031] The error change rate is expressed as: In the formula, △e(t) represents the error change rate at time t.

[0032] In view of the problem that during actual obstacle removal work, the reaction force and the fluctuation range of the friction coefficient are relatively large when the obstacle removal robot comes into contact with an obstacle, and the existing robot control instructions lack real-time dynamic adjustment, resulting in low control accuracy. In this embodiment, first, fuzzy PID control is performed on the force error and the error change rate to calculate the output power of the obstacle removal robot. Specifically, the force error and the error change rate are used as the inputs of the fuzzy controller, and rule reasoning is performed through a preset fuzzy rule table to output the dynamic adjustment amount of the PID parameters. Finally, based on the dynamic adjustment amount and the force error, the output power is obtained through PID calculation, where the output power can be expressed as: In the formula, P out represents the output power, K p , K i and K d are all dynamic adjustment amounts, and e(t) represents the force error at time t.

[0033] Since the output power calculated by PID does not take into account the influence factor of the friction coefficient, after obtaining the output power, it is also necessary to perform friction compensation on the output power according to the friction coefficient to obtain the final output power. In this embodiment, the output power is compensated based on the difference between the real-time friction coefficient and the reference friction coefficient. Specifically, when the friction coefficient is low, such as lower than the reference friction coefficient, the frictional force decreases and the tool is more likely to slip. At this time, the speed needs to be reduced, so the required driving force needs to be reduced. Similarly, when the friction coefficient is high, the frictional force increases. At this time, the speed can be maintained or increased to improve the obstacle removal efficiency. Therefore, the driving force can be increased. Based on this principle, the final output power after friction compensation can be expressed as: In the formula, P f represents the final output power, represents the compensation coefficient, represents the friction coefficient, represents the reference friction coefficient.

[0034] For the final output power, it is converted into the motor speed parameter and the motor torque parameter of the obstacle removal robot.

[0035] Regarding the tool movement trajectory of the obstacle removal robot, in this embodiment, first, an initial path is generated according to the position of the obstacle. With the maximum allowable curvature as the path curvature constraint condition, the initial path is smoothed in terms of curvature so that the smoothed path meets the constraint condition. Secondly, according to the relationship between the above-mentioned friction coefficient and the movement speed, the initial movement speed is adjusted, and its adjustment formula can be expressed as: where, v new represents the movement speed, and v base represents the initial movement speed, represents the friction coefficient, represents the reference friction coefficient.

[0036] According to the above-mentioned smooth path and movement speed, the tool movement trajectory of the obstacle clearing robot can be determined. At the same time, combined with the above-mentioned motor speed parameters and motor torque parameters, the control instructions of the obstacle clearing robot are finally generated to achieve precise control of the obstacle clearing robot. It should be noted that in this embodiment, the smooth path and movement speed refer to the movement speed and movement path of the obstacle clearing tool when the obstacle clearing robot uses the obstacle clearing tool for obstacle clearing work, rather than the movement path and movement speed of the obstacle clearing robot itself. The control instructions in this embodiment are for the control of the obstacle clearing tool when the obstacle clearing robot is performing obstacle clearing work. In the process of obstacle clearing in this embodiment, by analyzing the reaction force and friction coefficient between the obstacle clearing robot and the obstacle in real time, the control parameters of the obstacle clearing robot are adjusted dynamically in real time, which can improve the accuracy of the obstacle clearing robot's control of the obstacle clearing tool, thereby improving the obstacle clearing efficiency.

[0037] After the obstacle clearing robot completes the obstacle clearing work, since there will be particles of broken obstacles inside the cable duct, it is also necessary to perform particle cleaning work inside the cable duct. In this embodiment, by obtaining the particle distribution image inside the cable duct and performing image processing, a particle cleaning planning path is generated to improve the cleaning efficiency. The specific steps include: Obtain the particle distribution image inside the cable duct, generate a particle density map through the region growing algorithm, and rasterize the particle density map to obtain a rasterized network; Calculate the particle density weight and the dynamic weight factor according to the rasterized network; Construct a cost function based on the path planning algorithm according to the particle density weight and the dynamic weight factor, and solve it to obtain the particle cleaning planning path.

[0038] In this embodiment, for the conventional path planning algorithm, such as the traditional A algorithm, when performing path planning, aiming at the problem of insufficient response efficiency to the dynamic change of particle distribution, the A algorithm is improved by introducing a dynamic weight mechanism, so as to be able to respond to the change of particle density in real time and optimize the path priority. Specifically, first, the inside of the pipeline is imaged through a high-definition camera, and after median filtering denoising, a particle density map is generated through the region growing algorithm, and the particle density map is rasterized to obtain a rasterized network and the density value of each grid.

[0039] According to the ratio of the network density to the average density, calculate the particle density weight, and its formula is expressed as: In the formula, d(n) represents the particle density weight of node n, ρ(n) represents the grid density of node n, and ρ t represents the average grid density. The meaning of this particle density weight formula is that the higher the density of node n, the greater its particle density weight.

[0040] Then, according to the ratio of the real-time particle density to the maximum density, the dynamic weight factor is calculated, and its formula is expressed as: In the formula, w(t) represents the dynamic weight factor at time t, represents the particle density at time t, represents the maximum particle density.

[0041] In the dynamic weight factor formula, t represents the real-time timestamp, which is used to reflect the dynamic change of the particle distribution. The particle density at time t refers to the particle density of the entire particle density map at time t, and the maximum particle density is a preset value. The dynamic weight factor is used to characterize the high-density area, and the path priority of the high-density area should be improved.

[0042] Introduce the above particle density weight and dynamic weight factor into the path cost function, so as to obtain the cost function of the improved A algorithm: In the formula, f(n) represents the cost function of node n, g(n) represents the actual cost from the starting point to node n, h(n) represents the estimated cost from node n to the end point, w(t) represents the dynamic weight factor at time t, d(n) represents the particle density weight of node n, and β represents the density coefficient.

[0043] For the above cost function, first initialize the positions of the starting point, end point and obstacles, and then traverse the adjacent nodes according to the improved A algorithm, and preferentially select the path with the smallest cost function. The specific calculation steps can refer to the steps of the conventional A algorithm for path planning, and will not be elaborated here one by one.

[0044] In this embodiment, the real-time particle density is associated through the timestamp t, so that the algorithm can adapt to the diffusion or accumulation changes of particles during the cleaning process, and the weight is dynamically adjusted according to the density. It can preferentially cover the high-density area while taking into account the path length efficiency. Through the optimization of the particle cleaning path, the cleaning efficiency of the obstacle clearing robot is effectively improved.

[0045] In a preferred embodiment, in order to reduce the damage caused by the obstacle clearing work to the cable pipeline, the present invention will also real-time judge the damage state inside the current pipeline during the obstacle clearing work, so as to adjust the obstacle clearing instruction. The specific steps include: In response to the obstacle removal robot executing the control instruction, obtain the laser scanning data of the cable pipeline, filter and denoise the laser scanning data, and generate a pipe wall contour image through three-dimensional reconstruction; Input the pipe wall contour image into a preset damage classification model to obtain a damage classification result, where the damage classification model is constructed based on a convolutional neural network model; According to the damage classification result, extract the damage features of the corresponding area from the laser scanning data, where the damage features include scratch depth and damage area; Determine the damage level according to the damage area, and adjust the tool movement trajectory according to the damage level; Adjust the output power of the obstacle removal robot according to the scratch depth.

[0046] In this embodiment, when the obstacle removal robot is performing obstacle removal work, the laser scanning data of the cable pipeline is obtained through laser scanning. These laser scanning data are filtered by low-pass filtering to eliminate vibration noise and then three-dimensionally reconstructed to generate a pipe wall contour image. Then, the pipe wall contour image is input into a pre-trained damage classification model constructed based on a convolutional neural network model for damage identification to obtain a damage classification result, including different types of damage such as scratches, corrosion, and wear.

[0047] Then, according to the damage classification result, extract the damage features of the corresponding area from the filtered laser scanning data. Among them, for the scratch type, the scratch depth is extracted by calculating the height difference of the point cloud, such as the height difference between the scratch area and the normal area. For the corrosion type and wear type, the projected area of the damaged area is statistically analyzed through connected component analysis to calculate the damage area.

[0048] In this embodiment, the damage area is used to measure the spatial range of the damaged area of the pipe wall. The larger the damage area of the pipeline, the greater the indication of a large range of structural weakening in this area. Therefore, the damage degree can be defined according to the damage area. If the damage area is greater than a preset area threshold, it is marked as a high damage area. At this time, the tool movement trajectory needs to be adjusted to avoid this area as much as possible to prevent further damage to the pipe wall. For the scratch depth, if the scratch depth is greater than the depth threshold, it indicates that the pipe wall damage at this location is relatively serious. For the obstacle removal work at this part, the operation force of the obstacle removal tool can be reduced by reducing the output power, thereby avoiding further damage to the pipe wall.

[0049] This embodiment dynamically adjusts the control instruction of the obstacle removal tool of the obstacle removal robot based on the pipe wall damage. While ensuring the obstacle removal efficiency, it further reduces the damage of the obstacle removal work to the pipe wall of the cable pipeline, thereby effectively extending the maintenance cycle of the cable pipeline.

[0050] In another preferred embodiment, the present invention calculates the safety factor based on the reaction force when the obstacle removal tool contacts the obstacle, and adjusts the output power of the obstacle removal robot according to the safety factor. The specific steps include: Calculate the actual stress according to the reaction force and the tool contact area; Calculate the safety factor according to the allowable stress of the cable duct and the actual stress; Adjust the output power of the obstacle removal robot according to the safety factor.

[0051] In this embodiment, first, the actual stress is calculated based on the reaction force when the obstacle removal tool contacts the obstacle and the tool contact area. The actual stress can be expressed as: In the formula, represents the actual stress, F ac represents the reaction force, A represents the tool contact area, and K represents the stress coefficient.

[0052] Then, the safety factor is calculated according to the actual stress and the allowable stress of the cable duct. The material strength is evaluated through the safety factor. Among them, the calculation formula of the safety factor can be expressed as: In the formula, G represents the safety factor, represents the allowable stress, that is, the allowable stress of the pipe wall material of the cable duct, which is a theoretical value based on static ideal conditions. The static ideal conditions refer to no damage and uniform material.

[0053] In this embodiment, it is judged whether the output of the current obstacle removal robot needs to be adjusted according to the safety factor. If S = 1.0, it means that the pipe wall is in a critical state at this time. However, in actual working conditions, it is also necessary to consider the problem of excessive instantaneous stress caused by dynamic loads and excessive local stress in the damaged area caused by the stress concentration effect. The safety factor should be appropriately increased to ensure stress safety under actual working conditions. Therefore, in this embodiment, preferably, when S≥1.5, it is considered that the stress on the pipe wall is within the safe range at this time, and there is no need to adjust the output power of the obstacle removal robot. If 1.2 < S < 1.5, it means that there is a certain risk of material stress exceeding the limit, and it is necessary to appropriately reduce the output power of the obstacle removal robot, such as reducing it by 20%. If S≤1.2, after considering preventive measures for actual risks such as dynamic loads, detection errors, and material fatigue, emergency shutdown should be carried out at this time for pipeline maintenance to ensure pipeline safety to the greatest extent.

[0054] This embodiment comprehensively evaluates the pipe material and the output of the obstacle removal robot through the safety factor, realizes the dynamic regulation of the output power of the obstacle removal robot, and effectively ensures the safety of the cable duct during the obstacle removal process.

[0055] A cable pipeline detection and obstacle removal method based on a robot provided in this embodiment. Through multi-sensor data fusion modeling, the present invention improves the accuracy of obstacle recognition inside the cable pipeline; through a multi-objective optimization algorithm to optimize the dynamic scheduling of tools, the tool switching efficiency of the obstacle removal robot is improved, and the obstacle removal cost is reduced; through fuzzy PID control to dynamically regulate the output power of the obstacle removal robot, precise control of the obstacle removal robot is achieved, and the obstacle removal efficiency is improved; through intelligent particle cleaning path planning, the cleaning efficiency is effectively improved; and through a closed-loop protection mechanism for pipe wall damage, the safety of the cable pipeline during the obstacle removal process is effectively ensured. Through multi-modal perception fusion, dynamic optimization decision-making, and output force motion collaborative control, the present invention effectively improves the efficiency of cable pipeline obstacle removal work, reduces the obstacle removal cost, ensures the safety of the cable pipeline during the obstacle removal process, and provides an effective solution for the intelligent operation and maintenance of urban pipe networks.

[0056] Please refer to Figure 2 , based on the same inventive concept, a cable pipeline detection and obstacle removal system proposed in the second embodiment of the present invention includes: An obstacle classification module 10, configured to obtain multi-source data inside the cable pipeline, perform data processing and feature analysis on the multi-source data to obtain obstacle information, where the obstacle information includes obstacle type and obstacle position; A tool instruction generation module 20, configured to calculate an optimal tool switching sequence using a multi-objective optimization algorithm according to the tool parameters of a preset obstacle removal tool and the obstacle information, generate a tool configuration adjustment instruction according to the optimal tool switching sequence, and send it to the obstacle removal robot; An obstacle removal instruction generation module 30, configured to, in response to the obstacle removal robot executing the tool configuration adjustment instruction to replace the obstacle removal tool, generate an operation force control parameter and a tool motion trajectory according to the reaction force and friction coefficient when the obstacle removal robot contacts the obstacle, generate an obstacle removal control instruction according to the operation force control parameter and the tool motion trajectory, and send it to the obstacle removal robot; A cleaning instruction generation module 40, configured to, in response to the obstacle removal robot executing the control instruction, obtain a particle distribution image inside the cable pipeline, perform image processing to obtain a particle cleaning planning path, generate a particle cleaning instruction according to the particle cleaning planning path, and send it to the obstacle removal robot to perform a cleaning operation.

[0057] The technical features and technical effects of the cable duct inspection and obstacle clearance system based on a robot proposed in the embodiments of the present invention are the same as those of the method proposed in the embodiments of the present invention, and will not be elaborated here. Each module in the above-mentioned cable duct inspection and obstacle clearance system based on a robot can be implemented in whole or in part by software, hardware, and their combinations. The above-mentioned modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so as to facilitate the processor to call and execute the operations corresponding to the above-mentioned modules.

[0058] In summary, an embodiment of the present invention proposes a method and system for cable duct inspection and obstacle clearance based on a robot. The method obtains multi-source data inside the cable duct, processes the multi-source data and analyzes the features to obtain obstacle information, where the obstacle information includes the obstacle type and the obstacle position; according to the tool parameters of the preset obstacle clearance tool and the obstacle information, uses a multi-objective optimization algorithm to calculate the optimal tool switching sequence, generates a tool configuration adjustment instruction according to the optimal tool switching sequence, and sends it to the obstacle clearance robot; in response to the obstacle clearance robot executing the tool configuration adjustment instruction to replace the obstacle clearance tool, generates an operation force control parameter and a tool movement trajectory according to the reaction force and the friction coefficient when the obstacle clearance robot contacts the obstacle, generates a control instruction according to the operation force control parameter and the tool movement trajectory, and sends it to the obstacle clearance robot; in response to the obstacle clearance robot completing the execution of the control instruction, obtains the particle distribution image inside the cable duct, performs image processing to obtain the particle cleaning planning path, generates a particle cleaning instruction according to the particle cleaning planning path, and sends it to the obstacle clearance robot to perform the cleaning operation. The present invention effectively improves the efficiency of cable duct obstacle clearance work, reduces the obstacle clearance cost, ensures the safety of the cable duct during the obstacle clearance process, and provides an effective solution for the intelligent operation and maintenance of urban pipe networks through multi-modal perception fusion, dynamic optimization decision-making, and force output and motion collaborative control.

[0059] Each embodiment in this specification is described in a progressive manner. For parts that are the same or similar in each embodiment, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and reference can be made to the partial description of the method embodiment for related parts. It should be noted that the above-mentioned technical features of the embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the above-mentioned technical features in the embodiments are described. However, as long as the combination of these technical features does not conflict, it should be considered as the scope recorded in this specification.

[0060] The above-described embodiments merely represent several preferred embodiments of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the technical principles of the present invention, several improvements and substitutions can be made, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the protection scope of the claims described above.

Claims

1. A robot-based cable duct inspection and obstacle removal method, characterized in that, Including: Obtain multi-source data inside the cable duct, perform data processing and feature analysis on the multi-source data to obtain obstacle information, where the obstacle information includes obstacle type and obstacle location; According to the tool parameters of a preset obstacle removal tool and the obstacle information, use a multi-objective optimization algorithm to calculate the optimal tool switching sequence. According to the optimal tool switching sequence, generate a tool configuration adjustment instruction and send it to the obstacle removal robot; In response to the obstacle removal robot executing the tool configuration adjustment instruction to replace the obstacle removal tool, generate an operation force control parameter and a tool movement trajectory according to the reaction force and friction coefficient when the obstacle removal robot contacts the obstacle. According to the operation force control parameter and the tool movement trajectory, generate a control instruction and send it to the obstacle removal robot; In response to the obstacle removal robot completing the execution of the control instruction, obtain the particle distribution image inside the cable duct, perform image processing to obtain the particle cleaning planning path, and generate a particle cleaning instruction according to the particle cleaning planning path and send it to the obstacle removal robot to perform the cleaning operation.

2. The cable duct detection and obstacle removal method based on a robot according to claim 1, characterized in that The step of obtaining multi-source data inside the cable duct, performing data processing and feature analysis on the multi-source data to obtain obstacle information, where the obstacle information includes obstacle type and obstacle location includes: Collect laser point cloud data, acoustic wave data, and infrared data based on obstacles inside the cable duct, and perform data processing and particle filtering to obtain a three-dimensional obstacle feature map; According to a preset material matching rule and the three-dimensional obstacle feature map, perform obstacle recognition to obtain the obstacle type and obstacle location of each obstacle.

3. The cable duct detection and obstacle removal method based on a robot according to claim 1, characterized in that The step of using a multi-objective optimization algorithm to calculate the optimal tool switching sequence according to the tool parameters of a preset obstacle removal tool and the obstacle information includes: According to the obstacle type, determine the obstacle hardness, and extract the tool hardness, remaining tool life, and expected tool life of each obstacle removal tool from a preset tool database; Calculate the tool matching coefficient according to the obstacle hardness, the tool hardness, the remaining tool life, and the expected tool life; Establish a multi-objective optimization model with the minimum tool cost and the minimum tool switching time as the objective functions and the tool matching coefficient and the remaining tool life as the constraint conditions; Solve the multi-objective optimization model to obtain the optimal tool switching sequence.

4. The method for cable duct detection and obstacle clearance based on a robot according to claim 3, characterized in that, The tool matching coefficient is expressed by the following formula: In the formula, represents the tool matching coefficient between the i-th obstacle removal tool and the j-th obstacle type, S i represents the tool hardness of the i-th obstacle removal tool, S j represents the obstacle hardness of the j-th obstacle type, α1 represents the hardness coefficient, M s,i represents the remaining tool life of the i-th obstacle removal tool, M i represents the expected tool life of the i-th obstacle removal tool, α2 represents the life coefficient; The objective function is expressed by the following formula: Where n represents the total number of obstacle clearing tools, C i represents the unit price of the i-th obstacle clearing tool, N i represents the usage times of the i-th obstacle clearing tool, T s,i represents the switching time of the i-th obstacle clearing tool, T o,i represents the operation time of the i-th obstacle clearing tool, C total represents the tool cost, T total represents the tool switching time; The constraint condition is expressed by the following formula: In the formula, represents the coefficient threshold, and M t represents the life threshold.

5. The cable duct detection and obstacle clearance method based on a robot according to claim 1, characterized in that, The step of generating an operation force control parameter and a tool movement trajectory according to the reaction force and friction coefficient when the obstacle removal robot contacts the obstacle includes: According to the obstacle type, determine the target force, and calculate the force error and the error change rate according to the reaction force when the obstacle removal robot contacts the obstacle and the target force; Perform fuzzy PID control on the force error and the error change rate to obtain the output power; Perform friction compensation on the output power according to the friction coefficient when the obstacle removal robot contacts the obstacle, obtain the final output power, and convert the final output power into motor speed parameters and motor torque parameters; Generate an initial path according to the obstacle position, and perform curvature smoothing on the initial path to obtain a smooth path; Adjust the initial motion speed according to the friction coefficient to obtain the motion speed; Obtain the tool motion trajectory of the obstacle removal robot according to the smooth path and the motion speed.

6. The cable duct detection and obstacle removal method based on a robot according to claim 1, wherein, The steps of acquiring the particle distribution image inside the cable duct and performing image processing to obtain the particle cleaning planning path include: Acquire the particle distribution image inside the cable duct, generate a particle density map through the region growing algorithm, and rasterize the particle density map to obtain a rasterized network; Calculate the particle density weight and the dynamic weight factor according to the rasterized network; Construct a cost function based on the path planning algorithm according to the particle density weight and the dynamic weight factor, and solve it to obtain the particle cleaning planning path.

7. The method for cable duct detection and obstacle clearance based on a robot according to claim 6, characterized in that, The particle density weight is expressed by the following formula: where d(n) represents the particle density weight of node n, ρ(n) represents the grid density of node n, and ρ t represents the average grid density; The dynamic weight factor is expressed by the following formula: where \(w(t)\) represents the dynamic weight factor at time \(t\), represents the particle density at time \(t\), represents the maximum particle density; The cost function is expressed by the following formula: In the formula, f(n) represents the cost function of node n, g(n) represents the actual cost from the starting point to node n, h(n) represents the estimated cost from node n to the end point, w(t) represents the dynamic weight factor at time t, d(n) represents the particle density weight of node n, and β represents the density coefficient.

8. The method for detecting and removing obstacles in a cable duct based on a robot according to claim 1, characterized in that, After the step of generating a control instruction according to the operation force control parameter and the tool motion trajectory and sending it to the obstacle removal robot, it further includes: In response to the obstacle removal robot executing the control instruction, acquire the laser scan data of the cable duct, filter and denoise the laser scan data, and generate a pipe wall contour image through three-dimensional reconstruction; Input the pipe wall contour image into a preset damage classification model to obtain a damage classification result, and the damage classification model is constructed based on a convolutional neural network model; Extract the damage features of the corresponding area from the laser scan data according to the damage classification result, and the damage features include scratch depth and damage area; Determine the damage level according to the damage area, and adjust the tool motion trajectory according to the damage level; Adjust the output power of the obstacle removal robot according to the scratch depth.

9. The method for cable duct detection and obstacle removal based on a robot according to claim 8, characterized in that, After the step of generating a control instruction according to the operation force control parameter and the tool motion trajectory and sending it to the obstacle removal robot, it further includes: Calculate the actual stress according to the reaction force and the tool contact area; Calculate the safety factor according to the allowable stress of the cable duct and the actual stress; Adjust the output power of the obstacle removal robot according to the safety factor; Among them, the actual stress is expressed by the following formula: In the formula, represents the actual stress, F ac represents the reaction force, A represents the tool contact area, and K represents the stress coefficient; The safety factor is expressed by the following formula: In the formula, G represents the safety factor, represents the allowable stress.

10. A robot-based cable duct detection and obstacle removal system, characterized in that, It includes: An obstacle classification module, configured to acquire multi-source data inside the cable duct, perform data processing and feature analysis on the multi-source data to obtain obstacle information, and the obstacle information includes obstacle type and obstacle position; The tool instruction generation module is used to calculate the optimal tool switching sequence by using a multi-objective optimization algorithm according to the tool parameters of the preset obstacle removal tool and the obstacle information, generate a tool configuration adjustment instruction according to the optimal tool switching sequence, and send it to the obstacle removal robot; The obstacle removal instruction generation module is used to respond to the obstacle removal robot to execute the tool configuration adjustment instruction to replace the obstacle removal tool, generate an operation force control parameter and a tool motion trajectory according to the reaction force and friction coefficient when the obstacle removal robot contacts the obstacle, generate an obstacle removal control instruction according to the operation force control parameter and the tool motion trajectory, and send it to the obstacle removal robot; The cleaning instruction generation module is used to respond to the obstacle removal robot to execute the control instruction, obtain the particle distribution image inside the cable duct, perform image processing to obtain the particle cleaning planning path, generate a particle cleaning instruction according to the particle cleaning planning path, and send it to the obstacle removal robot to perform the cleaning operation.

Images