Full-coverage path planning method, device and equipment for wall-climbing robot and medium
By using a 2.5D enhanced hexagonal grid map and a bio-excited neural network model, combined with a stable adsorption and comprehensive energy consumption model, the problems of path planning accuracy and energy consumption control for wall-climbing robots in complex curved surface environments were solved, achieving safe, stable, full-coverage, and low-power operation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-23
- Publication Date
- 2026-04-24
Smart Images
Figure CN121916940A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of path planning technology, and in particular to a method, apparatus, equipment and medium for full-coverage path planning of a wall-climbing robot. Background Technology
[0002] In fields such as wind power and building curtain walls, wall-climbing robots are widely used in the operation and maintenance of large curved structures. Their operating environment is typically characterized by large curvature variations and unstable surface adhesion, while simultaneously requiring low power consumption and high safety for long-term operation. With the large-scale application of large curved structures, the full-coverage path planning of wall-climbing robots faces greater challenges. They must achieve complete area coverage, ensure adsorption stability in complex curved surface environments, and control energy consumption to extend operation time.
[0003] Currently, path planning schemes for wall-climbing robots are mostly based on traditional grid maps and conventional path algorithms. For example, they use rectangular grids to construct environmental models and combine A* algorithms or genetic algorithms for path search. Some schemes introduce energy consumption or stability constraints, but they mostly use a single indicator (such as path length or basic adhesion force) as the optimization objective and rely on simple geometric information (such as planar distance) for decision-making. At the same time, the application of existing bio-inspired neural network models in path planning focuses on the coverage efficiency of planar environments and does not adapt to the height, curvature, and other characteristics of curved surfaces.
[0004] However, existing technologies have low accuracy in path planning for wall-climbing robots. Summary of the Invention
[0005] This application provides a method, apparatus, device, and medium for full-coverage path planning of a wall-climbing robot, which can improve the accuracy of path planning for the wall-climbing robot.
[0006] To achieve the above objectives, this application adopts the following technical solution: In a first aspect, this application provides a full-coverage path planning method for a wall-climbing robot, including: Create a 2.5D enhanced hexagonal grid map that includes height, curvature, and normal vectors; Based on the aforementioned 2.5D enhanced hexagonal grid map, the biologically stimulated neural network model is initialized; Obtain the height, curvature, and normal vector information of the grid corresponding to the current position of the wall-climbing robot from the 2.5D enhanced hexagonal grid map; Based on the information of the corresponding grid, establish a stable adsorption model and a comprehensive energy consumption model; Based on the stable adsorption model and the comprehensive energy consumption model, the optimal grid is determined from the neighboring grids of the current grid. Based on the comprehensive evaluation value of the optimal grid and the activity values of the current grid and neighboring grids in the biologically stimulated neural network model, the updated activity value is obtained; based on the updated activity value and the comprehensive evaluation value of the optimal grid, the total activity value is obtained. If the wall-climbing robot does not enter the dead zone, the neighborhood grid with the highest total activity value is selected as the target grid for the wall-climbing robot's movement, and the target grid is marked as covered. Determine whether all grid cells in the 2.5D enhanced hexagonal grid map have been covered; If the coverage is not complete, obtain the height, curvature, and normal vector information of the grid corresponding to the next position of the wall-climbing robot; If the coverage is complete, then end the path planning.
[0007] Optionally, the stable adsorption model is obtained in the following way: Obtain the curvature deviation angle; Based on the curvature deviation angle, construct the gravity component function; Based on the gravity component function, and in conjunction with anti-slip constraints, anti-longitudinal overturning constraints, and anti-lateral overturning constraints, the minimum adsorption force required to satisfy each type of constraint is calculated respectively. The maximum value among the various minimum adsorption forces is taken as the minimum adsorption force required for stable adsorption; Based on the minimum adsorption force and reliability coefficient, a stable adsorption model is calculated.
[0008] Optionally, the comprehensive energy consumption model is obtained in the following way: Obtain the outputs of the motion energy consumption calculation model and the adsorption energy consumption calculation model; The output of the motion energy consumption calculation model is added to the output of the adsorption energy consumption calculation model to obtain the comprehensive energy consumption model.
[0009] Optionally, determining the optimal grid from the neighboring grids of the current grid based on the stable adsorption model and the comprehensive energy consumption model includes: Get the changes in movement distance and turning angle between the current grid and each neighboring grid; Based on the stable adsorption model, the change in adsorption reliability between the current grid and each neighboring grid is calculated. Based on the comprehensive energy consumption model, calculate the energy consumption change between the current grid and each neighboring grid; Based on the changes in moving distance, turning angle, adsorption reliability, and energy consumption, calculate the comprehensive evaluation value for each neighborhood grid. The neighborhood grid with the highest comprehensive evaluation value is determined as the optimal grid.
[0010] Optionally, obtaining the updated activity value based on the comprehensive evaluation value of the optimal grid and the activity values of the current grid and neighboring grids in the biologically activated neural network model includes: Based on the comprehensive evaluation value of the optimal grid, set the external excitation input for the current grid; Based on the external stimulus input, the activity values of the current grid and neighboring grids, the connection weights and decay rates between neurons, the updated activity values of the current grid and each neighboring grid are obtained.
[0011] Optionally, the step of calculating the comprehensive evaluation value of each neighboring grid based on the changes in moving distance, turning angle, adsorption reliability, and energy consumption includes:
[0012] in, Indicates the first The comprehensive evaluation value of each neighboring grid. Indicates the first weight. Indicates the first The change in the moving distance of each neighboring grid cell, This represents the maximum value of the change in distance traveled. Indicates the second weight. Indicates a change in steering direction. This indicates the maximum value of the change in steering direction. Indicates the third weight. This indicates a change in adsorption reliability. This represents the maximum value of the change in adsorption reliability. Indicates the fourth weight. Indicates changes in energy consumption. This represents the maximum value of the change in energy consumption.
[0013] Optionally, the outputs of the motion energy consumption calculation model and the adsorption energy consumption calculation model are obtained in the following ways:
[0014]
[0015] in, This represents the output of the motion energy consumption calculation model. This represents the output of the adsorption energy consumption calculation model. Indicates the energy conversion efficiency of the musculoskeletal system. This indicates the energy conversion efficiency of the adsorption system. Indicates the rated power of the motion system. Indicates the rated power of the adsorption system. This indicates the distance the robot moves within the corresponding grid area. This represents the robot's average moving speed within the corresponding area.
[0016] Secondly, this application provides a full-coverage path planning device for a wall-climbing robot, comprising: The acquisition module is used to establish a 2.5D enhanced hexagonal grid map containing height, curvature, and normal vector; initialize a biologically stimulated neural network model based on the 2.5D enhanced hexagonal grid map; and acquire the height, curvature, and normal vector information of the grid corresponding to the current position of the wall-climbing robot from the 2.5D enhanced hexagonal grid map. The processing module is used to establish a stable adsorption model and a comprehensive energy consumption model based on the information of the corresponding grid. Based on the stable adsorption model and the comprehensive energy consumption model, the optimal grid is determined from the neighboring grids of the current grid; the updated activity value is obtained based on the comprehensive evaluation value of the optimal grid and the activity values of the current grid and neighboring grids in the biologically stimulated neural network model; the total activity value is obtained based on the updated activity value and the comprehensive evaluation value of the optimal grid. The judgment module is used to select the neighborhood grid with the highest total activity value as the target grid for the wall-climbing robot if the wall-climbing robot has not entered the dead zone, and mark the target grid as covered; determine whether all grids in the 2.5D enhanced hexagonal grid map have been covered; if not completely covered, obtain the height, curvature and normal vector information of the grid corresponding to the next position of the wall-climbing robot; if completely covered, end the path planning.
[0017] Thirdly, this application provides a computing device, including a memory and a processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of the first aspects.
[0018] Fourthly, this application provides a computer-readable storage medium for storing a computer program for performing the method as described in any one of the first aspects.
[0019] As can be seen from the above technical solution, this application has at least the following beneficial effects: In this application, firstly, by constructing a 2.5D enhanced hexagonal grid map containing height, curvature, and normal vector, the shortcomings of traditional grids in representing curved surface features are made up for. Combined with a stable adsorption model based on curvature deviation and multiple constraints, the adsorption instability risk in curved surface environments can be accurately predicted, thereby improving the adsorption stability of the wall-climbing robot when operating on complex curved surfaces and effectively reducing safety hazards such as slippage and overturning.
[0020] Secondly, the integrated energy consumption model combines dynamic calculations of motion energy consumption and adsorption energy consumption with a multi-index weighted comprehensive evaluation algorithm, achieving global optimization of movement distance, turning angle, adsorption reliability, and energy consumption. This avoids path redundancy caused by single index optimization and significantly reduces power consumption while ensuring coverage efficiency, thus extending the robot's continuous operation time.
[0021] Finally, by combining a 2.5D grid map adapted to curved surface features with a biologically stimulated neural network model, and through dynamic updates of neuron activity values and dead zone judgment mechanisms, the autonomous adaptability of path planning is ensured, and full coverage operation without omissions is achieved, thereby improving the operational efficiency and practicality of the wall-climbing robot in the maintenance of large curved surface structures.
[0022] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description
[0023] Figure 1 A flowchart illustrating a full-coverage path planning method for a wall-climbing robot, provided in an embodiment of this application; Figure 2 A dynamic model analysis of a wall-climbing robot that integrates its posture and forward direction is provided for embodiments of this application; Figure 3 This application provides an embodiment of a wall-climbing robot that provides a stable three-dimensional waterfall with suction force under different postures and along different directions of movement. Figure 4 A schematic diagram of a full-coverage path planning device for a wall-climbing robot provided in an embodiment of this application; Figure 5 This is a schematic diagram of a computing device provided in an embodiment of this application. Detailed Implementation
[0024] The terms "first," "second," and "third," etc., used in this application specification and accompanying drawings are used to distinguish different objects, not to limit a specific order.
[0025] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is 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. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.
[0026] To ensure clarity and conciseness in the description of the following embodiments, a brief introduction to the related technologies is given first: Wall-climbing robots are intelligent devices capable of autonomously moving and performing specific tasks on vertical or steeply curved surfaces (such as wind turbine blades, building curtain walls, and bridge surfaces). They rely on adsorption mechanisms (such as ducted fans, vacuum suction cups, and magnetic adsorption devices) to achieve stable attachment. They integrate functions such as environmental perception, path planning, and motion control, and are widely used in scenarios such as cleaning and maintenance of large structures, crack detection, and operation and maintenance. They need to balance operational safety, energy efficiency, and coverage integrity in complex curved surface environments.
[0027] Existing full-coverage path planning methods for wall-climbing robots face technical challenges in the operation and maintenance of large curved structures in fields such as wind power and building curtain walls. These challenges include insufficient adsorption stability, low energy consumption control precision, and poor accuracy of full-coverage path decision-making. They are unable to simultaneously ensure the safety, long-term effectiveness, and high efficiency of operations in complex curved environments, and cannot meet the comprehensive requirements of actual operation and maintenance for complete coverage, low power consumption, and high safety assurance.
[0028] The main reason for the above problems is that the rectangular grid map used in traditional path planning lacks accurate representation of key features such as surface height, curvature, and normal vector, making it impossible for the robot to effectively perceive changes in environmental geometry to predict adsorption risks. At the same time, existing solutions mostly treat stability and energy consumption as single optimization objectives, without building a dynamic evaluation model coupled with surface features, and path decision-making relies on only a few indicators, lacking multi-dimensional comprehensive consideration. In addition, the biologically stimulated neural network model has not been adapted and adjusted for curved surface scenarios, ultimately resulting in path planning safety, energy consumption optimization effect, and coverage efficiency failing to meet the requirements of practical applications.
[0029] In view of this, embodiments of this application provide a full-coverage path planning method for a wall-climbing robot, which can be executed by a processing device. This processing device can be a terminal or a server. Terminals include, but are not limited to, smartphones, tablets, laptops, personal digital assistants, or smart wearable devices. The server can be a cloud server, such as a central server in a central cloud computing cluster or an edge server in an edge cloud computing cluster. Alternatively, the server can be a server in a local data center. A local data center refers to a data center directly controlled by the user.
[0030] To address the issues of insufficient adsorption stability, low energy consumption control precision, and poor path decision accuracy faced by wall-climbing robots in the operation and maintenance of large curved structures, this application establishes a 2.5D enhanced hexagonal grid map containing height, curvature, and normal vectors to compensate for the shortcomings of traditional modeling in representing the geometric features of curved surfaces. Simultaneously, a stable adsorption model coupled with the characteristics of the curved surface and a comprehensive energy consumption model are constructed to achieve a dynamic balance between safety constraints and energy consumption control. Furthermore, by combining a bio-excited neural network model adapted to curved surface scenarios with a comprehensive evaluation mechanism integrating multi-dimensional indicators, the robot is guided to autonomously select the optimal path, ultimately achieving the goal of safe, stable, low-consumption, high-efficiency, and comprehensive path planning in complex curved surface environments.
[0031] To make the technical solution of this application clearer and easier to understand, the following describes a full-coverage path planning method for a wall-climbing robot provided by an embodiment of this application, in conjunction with the accompanying drawings. Figure 1 As shown, this figure is a flowchart of a full-coverage path planning method for a wall-climbing robot provided in an embodiment of this application. The method includes: S201, The processing device creates a 2.5D enhanced hexagonal grid map containing height, curvature, and normal vectors.
[0032] 2.5D enhanced hexagonal grid maps are a form of environmental modeling that lies between two-dimensional (2D) and three-dimensional (3D). 2.5D means that while preserving planar coordinate information, key three-dimensional feature parameters are superimposed (a non-complete 3D model); enhanced means that the grid not only contains location information but also integrates core physical property data of the surface; the hexagonal grid uses regular hexagons as basic units to divide environmental areas, and compared to rectangular grids, it has advantages such as consistent distances between adjacent units, smooth boundary transitions, and no directional bias, making it more suitable for the geometric representation of curved environments.
[0033] Height refers to the average elevation value of each point on the surface within the grid area relative to a preset reference surface (such as the robot's starting working surface), used to reflect the undulations of the surface.
[0034] Curvature refers to the average quantitative value of the curvature of a surface within a grid area, used to characterize the geometric shape changes of a surface (e.g., a gentle surface has small curvature, while a steep, curved surface has large curvature).
[0035] The normal vector is a vector perpendicular to the tangent plane of the surface within the grid area. Its direction can reflect the tilt attitude of the surface and is a parameter for calculating the adsorption force and gravity component.
[0036] The processing equipment first acquires the geometric data of large curved structures such as wind turbine blades and building curtain walls (which can be obtained through engineering drawings, laser scanning, etc.). Then, it divides the curved environment into hexagonal units to form a basic grid framework. Next, the processing equipment calculates the average height, average curvature, and average normal vector within each hexagonal grid, and stores these characteristic parameters in association with the corresponding grid's planar coordinate information. This ultimately constructs a 2.5D enhanced hexagonal grid map that reflects both planar positional relationships and the key three-dimensional characteristics of the curved surface, providing comprehensive and accurate environmental data support for subsequent stable adsorption model construction, integrated energy consumption calculation, and path decision-making.
[0037] S202, the processing device initializes the biologically stimulated neural network model based on a 2.5D enhanced hexagonal grid map.
[0038] The biologically activated neural network model is an intelligent decision-making model that simulates the information transmission and interaction mechanisms of the biological nervous system. It guides decision-making through the excitation-inhibition interaction of neurons. In this application, each neuron in the model corresponds one-to-one with a grid in a 2.5D enhanced hexagonal grid map, and changes in the neuron's activity value serve as the basis for robot path selection.
[0039] The wall-climbing robot's processing equipment uses a pre-constructed 2.5D enhanced hexagonal grid map as data to initiate the initialization process of the bio-stimulated neural network model. Specifically, the processing equipment maps each hexagonal grid in the map to a neuron in the model, and sets initial activity values for all neurons. For example, neurons corresponding to uncovered grids are set to basic excitation values, and neurons corresponding to identified obstacle areas are set to inhibition values. Basic parameters such as connection weights and signal attenuation rates between neurons are also configured, ensuring an accurate correspondence between the neural network model and the grid division of the curved environment. This establishes the initial framework for environment-model linkage, preparing for subsequent dynamic updates of neuron activity values and guidance of the robot's path decisions by combining a stable adsorption model and a comprehensive energy consumption model.
[0040] S203, The processing device obtains the height, curvature, and normal vector information of the grid corresponding to the current position of the wall-climbing robot from the 2.5D enhanced hexagonal grid map.
[0041] The current location grid refers to the real-time physical location of the wall-climbing robot, which is determined by the positioning module (such as GPS or visual positioning) and mapped to the corresponding hexagonal cell in the 2.5D enhanced hexagonal grid map.
[0042] The wall-climbing robot's processing equipment first determines the robot's real-time physical position in the actual curved environment through its self-localization module. Then, it maps this physical position onto a pre-constructed 2.5D enhanced hexagonal grid map, locking onto the corresponding hexagonal grid cell. Subsequently, the processing equipment accurately extracts three key surface feature parameters—average height, average curvature, and average normal vector—from the pre-stored data of this grid cell. This provides accurate environmental data input for the current position, ensuring that the model calculations closely match actual working conditions, for the subsequent construction of stable adsorption and comprehensive energy consumption models.
[0043] S204. The treatment equipment establishes a stable adsorption model and a comprehensive energy consumption model based on the information of the corresponding grid.
[0044] The stable adsorption model is a mathematical model designed for the safe operation of wall-climbing robots on curved surfaces. By quantifying the minimum adsorption force required for stable adsorption under different working conditions, it assesses the adsorption reliability of the path and avoids risks such as robot slippage and overturning.
[0045] The integrated energy consumption model is an energy consumption assessment model that integrates the movement and adsorption processes of the wall-climbing robot. It can accurately calculate the total energy consumption during the operation and provide a quantitative basis for the selection of low-power paths.
[0046] After extracting the height, curvature, and normal vector information of the grid corresponding to the current position, the processing device builds a stable adsorption model and a comprehensive energy consumption model based on these core surface feature data. The stable adsorption model is obtained as follows: First, the processing equipment obtains the curvature deviation angle; based on the curvature deviation angle, a gravity component function is constructed.
[0047] Curvature deviation angle refers to the angle between the actual curvature direction of the current grid surface and the preset reference curvature direction (such as the reference direction of a smooth surface). It is a parameter that quantifies the degree of surface curvature deviation and directly affects the decomposition results of the robot's gravity components.
[0048] The gravity component function is a mathematical expression that integrates the curvature deviation angle and the direction of motion, describing the decomposition of gravity in different directions such as the tangential, normal, and transverse directions of a curved surface. It is the foundation for calculating adsorption forces and constraint conditions. The expression for the gravity component function is: (1) in, This indicates the component of the center of gravity of the wall-climbing robot. This indicates the weight of the wall-climbing robot. This represents the component of gravity in the transverse direction of the curved surface (parallel to the tangent plane and perpendicular to the tangent). This represents the component of gravity along the tangent of the surface (a direction parallel to the plane of the tangent). This represents the component of gravity in the normal direction of the surface (perpendicular to the plane tangent to the surface). Indicates the curvature deviation angle. This indicates the direction of the wall-climbing robot's next movement. This represents auxiliary angle parameters used to accurately describe the orientation of the surface.
[0049] Then, based on the gravity component function, and combined with anti-slip constraints, anti-longitudinal overturning constraints, and anti-lateral overturning constraints, the minimum adsorption force required to satisfy each type of constraint is calculated.
[0050] Anti-slip constraints are mechanical equilibrium conditions set to prevent robots from slipping when moving on curved surfaces because the tangential component of gravity is greater than the frictional force. The core is to ensure that the frictional force generated by the adsorption force is sufficient to resist the tangential effect of gravity.
[0051] like Figure 2 (b) and Figure 2 As shown in (c), the distribution of gravity components shifts with the wall-climbing robot in different postures and directions of movement. The wall-climbing robot conforms to the mechanical equilibrium equations in three directions: (2) in, These are the lateral friction forces (y-direction) of the left and right drive wheels, respectively. This represents the longitudinal frictional force (x-direction) between the left and right wheels of the drive wheel. This represents the longitudinal frictional force (x direction) between the front and rear wheels of the driven wheel. This represents the normal support force experienced by the front and rear wheels of the driven wheel. This represents the normal support force experienced by the left and right drive wheels. This indicates the thrust of the ducted fan.
[0052] The environment of each wheel is the same. and The minimum friction force required by the wheel, evenly distributed across its lateral and longitudinal friction forces, is: (3) The minimum suction force of a wall-climbing robot is determined by the minimum required support force, and the formula for the limit of friction force provided by the minimum required support force of the wheels satisfies: (4) in, This represents the coefficient of friction between the robot wheel and the curved surface.
[0053] Combining formulas (2), (3), and (4), the minimum suction force required for the robot to avoid slippage in different postures and directions of movement is obtained: (5) in, This represents the minimum adsorption force required for anti-slip constraint.
[0054] Longitudinal tipping restraint is a torque balance condition set to prevent the robot from flipping around the front and rear edges of the contact face. It needs to achieve longitudinal attitude stability by canceling out the torque of the adsorption force and gravity.
[0055] like Figure 2 As shown in (c), when the robot is in different postures and directions of movement, the robot is at the critical point of longitudinal overturning. Determine the torque at point A and establish the torque balance equation: (6) in, Indicates the wheel diameter of the robot. This indicates the height of the robot's center of gravity.
[0056] Combining formula (2), the ultimate longitudinal overturning instability state is calculated. : (7) Under the extreme longitudinal overturning instability state, combining formula (7) and The minimum friction required for the wheel is: (8) Under the extreme longitudinal overturning instability state, combined with formula (8), the formula for the limit value of friction force that the minimum support force required by the wheel can provide is satisfied: (9) Combining formulas (2), (8), and (9), the minimum adsorption force of the robot is calculated, yielding the minimum adsorption force required to prevent longitudinal overturning under different postures and directions of movement: (10) in, This represents the minimum adsorption force required to prevent longitudinal overturning. This represents the first intermediate calculation variable, which integrates the effects of gravity, curvature deviation angle, and direction of motion. The calculation expression is:
[0057] in, This represents the second intermediate calculation variable, which incorporates the effects of wheel diameter, center of gravity height, and friction coefficient. The calculation expression is:
[0058] The anti-lateral tipping constraint is a torque balance condition set to prevent the robot from tipping over to the left or right around the contact surface. It needs to offset the lateral gravitational torque through the reasonable distribution of the adsorption force.
[0059] like Figure 2 As shown in (c), when the robot is in different postures and directions of movement, the robot is at the critical point of lateral overturning. Determine the torque at point B and establish the torque balance equation, the formula of which is as follows: (11) Combining formula (2), the ultimate lateral overturning instability state is calculated. : (12) Under the extreme lateral overturning instability state, combining formula (12) and The minimum friction required for the wheel is: (13) Under the extreme lateral overturning instability state, combined with formula (13), the formula for the limit value of friction force that the minimum support force required by the wheel can provide is satisfied: (14) Combining formulas (2), (13), and (14), the minimum adsorption force of the robot is calculated, yielding the minimum adsorption force required to prevent lateral tipping under different postures and directions of movement: (15) in, This represents the minimum adsorption force required to prevent lateral overturning. This represents the third intermediate calculation variable, which integrates the effects of motion direction, center of gravity height, and friction coefficient. The calculation expression is:
[0060] The expression for the lateral overturning angle condition is:
[0061] This is the range of angles that trigger the risk of lateral tipping. When the direction of motion meets this condition, the robot needs greater suction force to avoid tipping over.
[0062] The maximum value among various minimum adsorption forces is taken as the minimum adsorption force required for stable adsorption, and its expression is:
[0063] in, This represents the minimum adsorption force required for stable adsorption.
[0064] like Figure 3 As shown, based on the stable adsorption model, Matlab simulation revealed that when the robot's movement direction is consistent with the tangential direction of the gravitational component, the minimum required adsorption force is minimized and the stability is maximized.
[0065] To ensure the safety and reliability of the wall-climbing robot's adsorption process, a stable adsorption model is calculated based on the minimum adsorption force and reliability coefficient. The calculation expression for the stable adsorption model is as follows:
[0066] in, This represents the output of the stable adsorption model. This represents the reliability coefficient.
[0067]
[0068] in, and Indicates the weighting coefficient. Indicates the magnitude of local curvature. This represents the robot's optimal direction of motion angle. This indicates the maximum permissible range of the direction angle of motion.
[0069] The integrated energy consumption model is obtained in the following way: The processing equipment acquires the outputs of the motion energy consumption calculation model and the adsorption energy consumption calculation model; the outputs of the motion energy consumption calculation model and the adsorption energy consumption calculation model are added together to obtain the comprehensive energy consumption model.
[0070] The motion energy consumption calculation model is a mathematical model used to quantify the energy consumed by a robot to overcome friction, gravity, and surface resistance during movement on a curved surface. The inputs typically include the movement distance, speed, and surface curvature, and the output is the motion energy consumption value per unit time or unit distance.
[0071] The adsorption energy consumption calculation model is a mathematical model used to quantify the energy consumed by a robot to maintain its adsorption state. The inputs are the adsorption force requirement and the efficiency of the adsorption mechanism (such as a ducted fan or vacuum pump), and the output is the adsorption energy consumption value per unit time.
[0072] The integrated energy consumption model is a total energy consumption assessment model that integrates motion energy consumption and adsorption energy consumption. The output is the total energy consumption of the robot during operation. It is the core basis for evaluating the energy economy of the path and realizing low-power path planning.
[0073] The outputs of the motion energy consumption calculation model and the adsorption energy consumption calculation model are obtained in the following ways:
[0074]
[0075] in, This represents the output of the motion energy consumption calculation model. This represents the output of the adsorption energy consumption calculation model. Indicates the energy conversion efficiency of the musculoskeletal system. This indicates the energy conversion efficiency of the adsorption system. Indicates the rated power of the motion system. Indicates the rated power of the adsorption system. This indicates the distance the robot moves within the corresponding grid area. This represents the robot's average moving speed within the corresponding area.
[0076] The calculation expression is:
[0077]
[0078]
[0079] in, This represents the driving torque required for the robot to move. This indicates the radius of the robot's wheels.
[0080] The calculation expression is:
[0081]
[0082]
[0083] in, This indicates the rotational speed of the ducted fan motor. This represents the output of the stable adsorption model. This indicates the thrust coefficient of the ducted fan. Indicates air density, Indicates the diameter of the ducted fan. The KV value represents the fan's rotational speed corresponding to 1V of voltage applied under no-load conditions. and These are the rated voltage and rated power of the ducted fan, respectively. This indicates the operating voltage of the ducted fan.
[0084] The calculation expression for the integrated energy consumption model is as follows:
[0085] in, This represents the output of the integrated energy consumption model.
[0086] S205. The processing equipment determines the optimal grid from the neighboring grids of the current grid based on the stable adsorption model and the comprehensive energy consumption model.
[0087] Specifically, the processing device acquires the changes in the moving distance and turning angle between the current grid and each neighboring grid; calculates the changes in adsorption reliability between the current grid and each neighboring grid based on a stable adsorption model; calculates the changes in energy consumption between the current grid and each neighboring grid based on a comprehensive energy consumption model; calculates the comprehensive evaluation value of each neighboring grid based on the changes in moving distance, turning angle, adsorption reliability, and energy consumption; and determines the neighboring grid with the highest comprehensive evaluation value as the optimal grid.
[0088] The current grid is the hexagonal cell in which the robot is located in real time within the 2.5D enhanced hexagonal grid map.
[0089] The neighborhood grid consists of all hexagonal cells that are directly adjacent to the current grid, and represents the candidate locations where the robot may move next.
[0090] The change in movement distance is the straight-line distance the robot travels from the current grid cell to a neighboring grid cell, and it is a direct reflection of the path length.
[0091] The change in steering angle is the angle by which a robot changes its direction of motion when moving from the current grid cell to a neighboring grid cell, reflecting the smoothness of the path. A larger steering angle usually means higher energy consumption and control complexity.
[0092] The change in adsorption reliability is calculated based on a stable adsorption model, representing the change in the safety redundancy of adsorption force as the robot moves between the current grid and neighboring grids. It reflects the impact of different neighboring grids on the robot's adsorption stability.
[0093] The energy consumption change is calculated based on the comprehensive energy consumption model, which measures the total change in motion energy consumption and adsorption energy consumption when the robot moves from the current grid to a neighboring grid.
[0094] The comprehensive evaluation value is a weighted score obtained by assigning different weights to four indicators—movement distance, turning angle, adsorption reliability, and energy consumption—for each neighboring grid, and it serves as the basis for path selection.
[0095] The optimal grid is the grid with the highest overall evaluation value among all neighboring grids, which is the optimal next movement target planned by the robot.
[0096] The formula for calculating the comprehensive evaluation value is:
[0097] in, Indicates the first The comprehensive evaluation value of each neighboring grid. Indicates the first weight. Indicates the first The change in the moving distance of each neighboring grid cell, This represents the maximum value of the change in distance traveled. Indicates the second weight. Indicates a change in steering direction. This indicates the maximum value of the change in steering direction. Indicates the third weight. This indicates a change in adsorption reliability. This represents the maximum value of the change in adsorption reliability. Indicates the fourth weight. Indicates changes in energy consumption. This represents the maximum value of the change in energy consumption.
[0098] The calculation expression is:
[0099] in, Indicates the first The reliability coefficient corresponding to each neighborhood grid is calculated using the stable adsorption model. ) represents the reliability coefficient corresponding to the optimal motion direction.
[0100] The calculation expression is:
[0101] in, Indicates the first The total energy consumption of a neighborhood grid This indicates the total energy consumption of the current grid.
[0102] After calculating the comprehensive evaluation value of all neighboring grids, the processing device sorts all candidate neighboring grids by their comprehensive evaluation values and determines the grid with the highest comprehensive evaluation value as the robot's optimal grid. This selection means that the grid has the best overall performance in four dimensions: travel distance, turning angle, adsorption reliability, and energy consumption. It can ensure the stability and safety of the robot's movement while also taking into account the efficiency of the path and the economy of energy consumption, making it the optimal next movement target within the current decision cycle.
[0103] S206. The processing device obtains the updated activity value based on the comprehensive evaluation value of the optimal grid and the activity values of the current grid and neighboring grids in the biologically stimulated neural network model; and obtains the total activity value based on the updated activity value and the comprehensive evaluation value of the optimal grid.
[0104] Specifically, the processing device sets the external stimulus input for the current grid based on the comprehensive evaluation value of the optimal grid; based on the external stimulus input, the activity values of the current grid and neighboring grids, the connection weights and decay rates between neurons, it obtains the updated activity values of the current grid and each neighboring grid. Then, based on the updated activity values and the comprehensive evaluation value of the optimal grid, it obtains the total activity value.
[0105] This process essentially simulates a dynamic path optimization mechanism similar to a neural network. The processing device first transmits the comprehensive evaluation value of the previously selected optimal grid as an external stimulus input to the current grid. Then, starting from this external stimulus, it calculates the new activity value for each neighboring grid by combining the preset connection weights (representing the correlation strength between grids) and the signal attenuation rate (simulating the characteristic of signal weakening with transmission) between the current grid and its surrounding neighboring grids. Finally, these newly calculated activity values are saved as the latest state of each grid, thereby dynamically updating the priority distribution of the entire grid map. This allows the robot's path planning to respond to environmental changes in real time and continuously adjust towards a better direction.
[0106] The processing device first determines the area attribute of the best grid cell based on its comprehensive evaluation value. Since the best grid cell is the neighboring grid cell with the highest comprehensive evaluation value, it must belong to an uncovered area (if it is an obstacle or a covered area, the comprehensive evaluation value will be very low and it will not be selected as a target).
[0107] Next, the system will substitute the determination result of this uncovered area into the value rules of the external excitation input to set the external excitation input for the current grid. In other words, the role of the comprehensive evaluation value of the optimal grid is to filter out the target grid with the highest priority and determine the attributes of its uncovered area, while the external excitation input is a signal directly generated based on this attribute.
[0108] Then, based on this external stimulus, and combining the connection weights (determined by distance) between the current grid and neighboring grids and the decay rate of neuron activity, a dynamic equation is used to calculate the change in activity value for each neuron (grid). This equation simulates signal transmission and feedback between neurons through the dynamic balance of excitation and inhibition terms, ultimately yielding the updated activity value for each grid.
[0109] The expression for the external excitation input value is:
[0110] in, Indicates the first External stimulus input to each neuron (grid), This represents a preset, very large constant used to ensure that the excitation intensity in the uncovered area is much greater than the inhibition intensity in the barrier area. This indicates that the robot has not yet completed the task of covering the grid. A grid indicates an obstacle that the robot cannot pass through. This indicates that the robot has completed the task of covering the grid, and the excitation is 0 to prevent the robot from entering the grid again.
[0111] The expression for connection weights is:
[0112] in, Indicates the first The first neuron and the second Connection weights between neurons This represents a preset constant used to uniformly scale the size of connection weights. Indicates the first The first neuron and the second The Euclidean distance between neurons indicates that the closer the distance, the greater the connection weight. Indicates the first A region centered on the nth neuron and with a radius of R1, only neurons within the receptive field will interact with the nth neuron. A neuron creates a connection.
[0113] Dynamic update equation for neuron activity values:
[0114] in, Indicates the first The current activity value of each neuron (grid). Indicates the first The rate of change of the current activity value of each neuron over time. Indicates the decay rate of neuronal activity. This represents the upper limit of neuron activity. This represents the lower limit of neuronal activity. Indicates the first The number of neighboring neurons within the perceptual domain of a single neuron. Indicates the first The current activity value of each neuron.
[0115] The definitions of the excitation and inhibition terms are as follows:
[0116] in, This represents the excitation operator, retaining only the positive portion of the signal. Positive signals enhance neuronal activity. This represents a suppression operator that retains only the negative portion of the signal, which inhibits neuronal activity.
[0117] Finally, the processing device obtains the total activity value based on the updated activity value and the comprehensive evaluation value of the optimal grid.
[0118] The processing device will fuse the grid activity value updated by the bio-stimulated neural network model with the comprehensive evaluation value of the best grid to obtain the total activity value of each neighboring grid. This value is a composite priority score that takes into account both the dynamic optimization results of the neural network and the comprehensive evaluation of multiple dimensions.
[0119] S207. If the wall-climbing robot does not enter the dead zone, then select the neighborhood grid with the highest total activity value as the moving target grid of the wall-climbing robot, and mark the moving target grid as covered.
[0120] A dead zone refers to a state where all neighboring grids around a robot are either obstacle areas or covered areas, and there are no movable, valid target grids. In this state, the robot cannot continue to perform its task.
[0121] First, determine if the robot has entered a dead zone, which means that all surrounding grid cells are blocked by obstacles or have already completed their tasks, leaving no effective moving target. If it is confirmed that the robot has not entered a dead zone, select the grid cell with the highest total activity value from all surrounding grid cells as the robot's next moving target grid. At the same time, mark the selected moving target grid cell as covered to prevent the robot from repeatedly planning to this area, ensuring the efficiency of global coverage operations.
[0122]
[0123] in, Represents a grid The total activity value, i.e., the grid The highest total activity value, Indicates the first The current activity value of each neuron, where c is a preset normal value used to adjust proportionally. Weight in total activity value Indicates the first The overall evaluation value of each neighborhood grid.
[0124] If the wall-climbing robot enters a dead zone—that is, when the system detects that all neighboring grids of the current grid are obstacles or covered areas, and there are no movable valid targets—it will immediately trigger a backtracking and replanning mechanism: First, the system will backtrack from the robot's movement history to the most recent branch point grid that still has uncovered neighboring grids. This grid is a node in the previous path that retained multiple possible directions. Next, the system will reset the external stimulus input of this branch point grid to E (the positive stimulus value for uncovered areas) to reactivate the grid's neural activity, making it a new path planning target. The system first marks the starting point; then, based on the dynamic update equation of the biologically activated neural network, it recalculates the activity values of all neighboring grids at that branch point, focusing on evaluating directions that were not previously selected; next, it selects the neighboring grid with the highest recalculated activity value as the new target grid for movement and marks it as covered, allowing the robot to continue its work from that branch point; finally, the system updates the movement history, removing the current dead zone grid from the history and replacing it with a new target grid to ensure the accuracy of subsequent backtracking logic, thus enabling the robot to jump out of the dead zone and continue to efficiently complete the global coverage task.
[0125] S208. Determine whether all grids in the 2.5D enhanced hexagonal grid map have been covered.
[0126] If the coverage is not complete, execute S209 to obtain the height, curvature, and normal vector information of the grid corresponding to the next position of the wall-climbing robot; if the coverage is complete, execute S210 to end the path planning.
[0127] S209. Obtain the height, curvature, and normal vector information of the grid corresponding to the next position of the wall-climbing robot.
[0128] If the detection result is incomplete coverage, it means that there are still unoperated areas. At this time, the system will first obtain key wall geometry information such as the height, curvature and normal vector of the grid corresponding to the next position planned by the robot. This provides environmental data support for the robot to adjust parameters such as adsorption reliability, motion energy consumption and turning angle, ensuring the rationality and safety of subsequent path planning and actual movement.
[0129] S210, End of path planning.
[0130] If the detection result is that the area is completely covered, it means that the work on all the grids in the entire work area has been completed. No further path calculation and movement planning are required. The entire path planning process can be terminated directly, and the climbing task can be completed.
[0131] Based on the above description, this application has the following beneficial effects: In this application, firstly, by constructing a 2.5D enhanced hexagonal grid map containing height, curvature, and normal vector, the shortcomings of traditional grids in representing curved surface features are made up for. Combined with a stable adsorption model based on curvature deviation and multiple constraints, the adsorption instability risk in curved surface environments can be accurately predicted, thereby improving the adsorption stability of the wall-climbing robot when operating on complex curved surfaces and effectively reducing safety hazards such as slippage and overturning.
[0132] Secondly, the integrated energy consumption model combines dynamic calculations of motion energy consumption and adsorption energy consumption with a multi-index weighted comprehensive evaluation algorithm, achieving global optimization of movement distance, turning angle, adsorption reliability, and energy consumption. This avoids path redundancy caused by single index optimization and significantly reduces power consumption while ensuring coverage efficiency, thus extending the robot's continuous operation time.
[0133] Finally, by combining a 2.5D grid map adapted to curved surface features with a biologically stimulated neural network model, and through dynamic updates of neuron activity values and dead zone judgment mechanisms, the autonomous adaptability of path planning is ensured, and full coverage operation without omissions is achieved, thereby improving the operational efficiency and practicality of the wall-climbing robot in the maintenance of large curved surface structures.
[0134] The above text combined Figure 1 The full-coverage path planning method for wall-climbing robots provided in this application embodiment has been described in detail. The device and equipment provided in this application embodiment will be described below with reference to the accompanying drawings.
[0135] like Figure 4 As shown in the figure, this is a schematic diagram of a full-coverage path planning device for a wall-climbing robot provided in an embodiment of this application. The device includes: The acquisition module 301 is used to establish a 2.5D enhanced hexagonal grid map containing height, curvature, and normal vector; initialize a biologically stimulated neural network model based on the 2.5D enhanced hexagonal grid map; and acquire the height, curvature, and normal vector information of the grid corresponding to the current position of the wall-climbing robot from the 2.5D enhanced hexagonal grid map. The processing module 302 is used to establish a stable adsorption model and a comprehensive energy consumption model based on the information of the corresponding grid. Based on the stable adsorption model and the comprehensive energy consumption model, the optimal grid is determined from the neighboring grids of the current grid; the updated activity value is obtained based on the comprehensive evaluation value of the optimal grid and the activity values of the current grid and neighboring grids in the biologically stimulated neural network model; the total activity value is obtained based on the updated activity value and the comprehensive evaluation value of the optimal grid. The judgment module 303 is used to select the neighborhood grid with the highest total activity value as the target grid for the wall-climbing robot if the wall-climbing robot has not entered the dead zone, and mark the target grid as covered; to determine whether all grids in the 2.5D enhanced hexagonal grid map have been covered; if not completely covered, to obtain the height, curvature and normal vector information of the grid corresponding to the next position of the wall-climbing robot; if completely covered, to end the path planning.
[0136] Optionally, the processing module 302 is specifically used to obtain the curvature deviation angle; Based on the curvature deviation angle, construct the gravity component function; Based on the gravity component function, and in conjunction with anti-slip constraints, anti-longitudinal overturning constraints, and anti-lateral overturning constraints, the minimum adsorption force required to satisfy each type of constraint is calculated respectively. The maximum value among the various minimum adsorption forces is taken as the minimum adsorption force required for stable adsorption; Based on the minimum adsorption force and reliability coefficient, a stable adsorption model is calculated.
[0137] Optionally, the processing module 302 is specifically used to acquire the output of the motion energy consumption calculation model and the output of the adsorption energy consumption calculation model; The output of the motion energy consumption calculation model is added to the output of the adsorption energy consumption calculation model to obtain the comprehensive energy consumption model.
[0138] Optionally, the processing module 302 is specifically used to obtain the changes in the moving distance and turning angle between the current grid and each neighboring grid; Based on the stable adsorption model, the change in adsorption reliability between the current grid and each neighboring grid is calculated. Based on the comprehensive energy consumption model, calculate the energy consumption change between the current grid and each neighboring grid; Based on the changes in moving distance, turning angle, adsorption reliability, and energy consumption, calculate the comprehensive evaluation value for each neighborhood grid. The neighborhood grid with the highest comprehensive evaluation value is determined as the optimal grid.
[0139] Optionally, the processing module 302 is specifically used to set the external excitation input of the current grid based on the comprehensive evaluation value of the optimal grid; Based on the external stimulus input, the activity values of the current grid and neighboring grids, the connection weights and decay rates between neurons, the updated activity values of the current grid and each neighboring grid are obtained.
[0140] Optionally, the processing module 302 is specifically used to calculate a comprehensive evaluation value for each neighboring grid based on the changes in moving distance, turning angle, adsorption reliability, and energy consumption, including:
[0141] in, Indicates the first The comprehensive evaluation value of each neighboring grid. Indicates the first weight. Indicates the first The change in the moving distance of each neighboring grid cell, This represents the maximum value of the change in distance traveled. Indicates the second weight. Indicates a change in steering direction. This indicates the maximum value of the change in steering direction. Indicates the third weight. This indicates a change in adsorption reliability. This represents the maximum value of the change in adsorption reliability. Indicates the fourth weight. Indicates changes in energy consumption. This represents the maximum value of the change in energy consumption.
[0142] Optionally, the processing module 302 specifically obtains the outputs of the motion energy consumption calculation model and the adsorption energy consumption calculation model in the following ways:
[0143]
[0144] in, This represents the output of the motion energy consumption calculation model. This represents the output of the adsorption energy consumption calculation model. Indicates the energy conversion efficiency of the musculoskeletal system. This indicates the energy conversion efficiency of the adsorption system. Indicates the rated power of the motion system. Indicates the rated power of the adsorption system. This indicates the distance the robot moves within the corresponding grid area. This represents the robot's average moving speed within the corresponding area.
[0145] The full-coverage path planning device for a wall-climbing robot according to the embodiments of this application can correspond to the execution of the method described in the embodiments of this application, and the other operations and / or functions of each module / unit of the full-coverage path planning device for the wall-climbing robot are respectively for implementing Figure 1 For the sake of brevity, the corresponding processes of each method in the illustrated embodiments will not be described in detail here.
[0146] This application also provides a computing device. For example... Figure 5 As shown in the figure, this is a schematic diagram of a computing device provided in an embodiment of this application. The computing device 700 includes a bus 701, a processor 702, a communication interface 703, and a memory 704. The processor 702, the memory 704, and the communication interface 703 communicate with each other via the bus 701.
[0147] The 701 bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 5 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.
[0148] The processor 702 can be any one or more of the following processors: central processing unit (CPU), graphics processing unit (GPU), microprocessor (MP), or digital signal processor (DSP).
[0149] The communication interface 703 is used for communication with external devices.
[0150] Memory 704 may include volatile memory, such as random access memory (RAM). Memory 704 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).
[0151] The memory 704 stores executable code, and the processor 702 executes the executable code to perform the aforementioned full-coverage path planning method for the wall-climbing robot.
[0152] Specifically, in achieving Figure 4 In the case of the illustrated embodiment, and Figure 4 When the modules or units of the full-coverage path planning device for the wall-climbing robot described in the embodiment are implemented through software, the execution... Figure 4 The software or program code required for the functions of each module / unit can be partially or entirely stored in the memory 704. The processor 702 executes the program code corresponding to each unit stored in the memory 704, and executes the aforementioned full-coverage path planning method for the wall-climbing robot.
[0153] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that a computing device can store, or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct the computing device to execute the aforementioned full-coverage path planning method for the wall-climbing robot.
[0154] This application also provides a computer program product comprising one or more computer instructions. When the computer instructions are loaded and executed on a computing device, all or part of the processes or functions described in this application are generated.
[0155] The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, or data center to another website, computer, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means.
[0156] When the computer program product is executed by a computer, the computer executes any of the aforementioned full-coverage path planning methods for the wall-climbing robot. The computer program product can be a software installation package; when any of the aforementioned full-coverage path planning methods for the wall-climbing robot needs to be used, the computer program product can be downloaded and executed on the computer.
[0157] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.
[0158] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.
Claims
1. A method for full-coverage path planning for a wall-climbing robot, characterized in that, The method includes: Create a 2.5D enhanced hexagonal grid map that includes height, curvature, and normal vectors; Based on the aforementioned 2.5D enhanced hexagonal grid map, the biologically stimulated neural network model is initialized; Obtain the height, curvature, and normal vector information of the grid corresponding to the current position of the wall-climbing robot from the 2.5D enhanced hexagonal grid map; Based on the information of the corresponding grid, establish a stable adsorption model and a comprehensive energy consumption model; Based on the stable adsorption model and the comprehensive energy consumption model, the optimal grid is determined from the neighboring grids of the current grid. Based on the comprehensive evaluation value of the optimal grid and the activity values of the current grid and neighboring grids in the biologically stimulated neural network model, the updated activity value is obtained; based on the updated activity value and the comprehensive evaluation value of the optimal grid, the total activity value is obtained. If the wall-climbing robot does not enter the dead zone, the neighborhood grid with the highest total activity value is selected as the target grid for the wall-climbing robot's movement, and the target grid is marked as covered. Determine whether all grid cells in the 2.5D enhanced hexagonal grid map have been covered; If the coverage is not complete, obtain the height, curvature, and normal vector information of the grid corresponding to the next position of the wall-climbing robot; If the coverage is complete, then end the path planning.
2. The method according to claim 1, characterized in that, The stable adsorption model was obtained in the following way: Obtain the curvature deviation angle; Based on the curvature deviation angle, construct the gravity component function; Based on the gravity component function, and in conjunction with anti-slip constraints, anti-longitudinal overturning constraints, and anti-lateral overturning constraints, the minimum adsorption force required to satisfy each type of constraint is calculated respectively. The maximum value among the various minimum adsorption forces is taken as the minimum adsorption force required for stable adsorption; Based on the minimum adsorption force and reliability coefficient, a stable adsorption model is calculated.
3. The method according to claim 1, characterized in that, The comprehensive energy consumption model is obtained in the following way: Obtain the outputs of the motion energy consumption calculation model and the adsorption energy consumption calculation model; The output of the motion energy consumption calculation model is added to the output of the adsorption energy consumption calculation model to obtain the comprehensive energy consumption model.
4. The method according to claim 1, characterized in that, The step of determining the optimal grid from the neighboring grids of the current grid based on the stable adsorption model and the comprehensive energy consumption model includes: Get the changes in movement distance and turning angle between the current grid and each neighboring grid; Based on the stable adsorption model, the change in adsorption reliability between the current grid and each neighboring grid is calculated. Based on the comprehensive energy consumption model, calculate the energy consumption change between the current grid and each neighboring grid; Based on the changes in moving distance, turning angle, adsorption reliability, and energy consumption, calculate the comprehensive evaluation value for each neighborhood grid. The neighborhood grid with the highest comprehensive evaluation value is determined as the optimal grid.
5. The method according to claim 1, characterized in that, The updated activity value is obtained based on the comprehensive evaluation value of the optimal grid and the activity values of the current grid and neighboring grids in the bio-stimulated neural network model, including: Based on the comprehensive evaluation value of the optimal grid, set the external excitation input for the current grid; Based on the external stimulus input, the activity values of the current grid and neighboring grids, the connection weights and decay rates between neurons, the updated activity values of the current grid and each neighboring grid are obtained.
6. The method according to claim 4, characterized in that, The calculation of a comprehensive evaluation value for each neighboring grid cell based on changes in moving distance, turning angle, adsorption reliability, and energy consumption includes: in, Indicates the first The comprehensive evaluation value of each neighboring grid. Indicates the first weight. Indicates the first The change in the moving distance of each neighboring grid cell, This represents the maximum value of the change in distance traveled. Indicates the second weight. Indicates a change in steering direction. This indicates the maximum value of the change in steering direction. Indicates the third weight. This indicates a change in adsorption reliability. This represents the maximum value of the change in adsorption reliability. Indicates the fourth weight. Indicates changes in energy consumption. This represents the maximum value of the change in energy consumption.
7. The method according to claim 3, characterized in that, The outputs of the motion energy consumption calculation model and the adsorption energy consumption calculation model are obtained in the following ways: in, This represents the output of the motion energy consumption calculation model. This represents the output of the adsorption energy consumption calculation model. Indicates the energy conversion efficiency of the musculoskeletal system. This indicates the energy conversion efficiency of the adsorption system. Indicates the rated power of the motion system. Indicates the rated power of the adsorption system. This indicates the distance the robot moves within the corresponding grid area. This represents the robot's average moving speed within the corresponding area.
8. A full-coverage path planning device for a wall-climbing robot, characterized in that, The device includes: The acquisition module is used to establish a 2.5D enhanced hexagonal grid map containing height, curvature, and normal vector; based on the 2.5D enhanced hexagonal grid map, initialize a biologically stimulated neural network model; and acquire the height, curvature, and normal vector information of the grid corresponding to the current position of the wall-climbing robot from the 2.5D enhanced hexagonal grid map. The processing module is used to establish a stable adsorption model and a comprehensive energy consumption model based on the information of the corresponding grid. Based on the stable adsorption model and the comprehensive energy consumption model, the optimal grid is determined from the neighboring grids of the current grid; based on the comprehensive evaluation value of the optimal grid and the activity values of the current grid and neighboring grids in the biologically stimulated neural network model, the updated activity value is obtained; based on the updated activity value and the comprehensive evaluation value of the optimal grid, the total activity value is obtained. The judgment module is used to select the neighborhood grid with the highest total activity value as the target grid for the wall-climbing robot if the wall-climbing robot has not entered the dead zone, and mark the target grid as covered; determine whether all grids in the 2.5D enhanced hexagonal grid map have been covered; if not completely covered, obtain the height, curvature and normal vector information of the grid corresponding to the next position of the wall-climbing robot; if completely covered, end the path planning.
9. A computing device, characterized in that, Including memory and processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program for performing the method as described in any one of claims 1 to 7.