Heterogeneous robot collaborative operation method and system

Abstract

The invention relates to the technical field of robots, and discloses a heterogeneous robot collaborative operation method and system. The method comprises the following steps: acquiring a first-level abnormal data packet uploaded by the wheeled mobile robot when executing a first-level detection task instruction, wherein the first-level abnormal data packet comprises an event abnormal position; based on an event abnormal position in the first-level abnormal data packet, determining a terrain category of an area where the abnormal event is located; if the terrain category is a non-flat terrain, generating and sending a secondary detection task instruction to the four-footed mobile robot to trigger the four-footed mobile robot to detect an area where the abnormal event is located to obtain a secondary abnormal data packet; and based on the abnormal equipment in the second-level abnormal data packet, generating and sending a maintenance task instruction to the wheeled mobile robot so as to trigger the wheeled mobile robot to execute maintenance operation on the abnormal equipment. By adopting the method, the whole-process unmanned closed-loop operation from abnormality discovery to final repair can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robots, and in particular to a heterogeneous robot cooperative operation method and system. BACKGROUND

[0002] With the development of industrial automation and intelligent operation and maintenance technology, mobile robots have been widely used in equipment inspection tasks to replace manual work to realize the automation of environment perception and data collection, and to improve operation safety and operational efficiency.

[0003] However, the existing inspection system generally has the problem of "heavy detection and light execution", that is, after discovering an exception, manual intervention is still needed for subsequent processing, and a closed-loop management from "discovering a problem" to "solving a problem" cannot be realized, resulting in a breakpoint in the operation and maintenance process. SUMMARY

[0004] Therefore, the embodiments of the present application provide a heterogeneous robot cooperative operation method and system, which can effectively solve the technical problem of "heavy detection and light execution" of the existing inspection system.

[0005] In a first aspect, the embodiments of the present application provide a heterogeneous robot cooperative operation method applied to a heterogeneous robot cooperative operation system, the system comprising a wheeled mobile robot, a quadruped mobile robot and a scheduling server, and the scheduling server performs the following method: obtaining a first-level exception data packet uploaded by the wheeled mobile robot executing a first-level detection task instruction, the first-level exception data packet comprising an event exception position; determining a terrain category of a region where an abnormal event is located based on the event exception position in the first-level exception data packet; if the terrain category is a non-flat terrain, generating and sending a second-level detection task instruction to the quadruped mobile robot to trigger the quadruped mobile robot to detect the region where the abnormal event is located, and obtaining a second-level exception data packet; based on an abnormal device in the second-level exception data packet, generating and sending a maintenance task instruction to the wheeled mobile robot to trigger the wheeled mobile robot to perform a maintenance operation on the abnormal device.

[0006] In a second aspect, the embodiments of the present application provide a heterogeneous robot cooperative operation system, comprising: a wheeled mobile robot configured to execute a first-level detection task instruction and perform a maintenance operation on an abnormal device according to a maintenance task instruction; The scheduling server is configured to acquire a first abnormal data packet uploaded by the wheeled mobile robot performing the first detection task instruction, the first abnormal data packet comprising an event abnormal position; determine a terrain category of a region where the abnormal event is located based on the event abnormal position in the first abnormal data packet; if the terrain category is a non-flat terrain, generate and send a second detection task instruction to a quadruped mobile robot to trigger the quadruped mobile robot to detect the region where the abnormal event is located to obtain a second abnormal data packet; and generate and send a maintenance task instruction to the wheeled mobile robot based on an abnormal device in the second abnormal data packet. The quadruped mobile robot is configured to receive the second detection task instruction sent by the scheduling server and execute the second detection task instruction.

[0007] The embodiments of the present application have the following beneficial effects: The scheduling server is configured to intelligently allocate tasks to the wheeled mobile robot and the quadruped mobile robot and manage the closed loop process. Specifically, a three-level task chain of "first detection -> second confirmation -> maintenance execution" is constructed to realize the full-process unmanned closed loop operation from abnormal discovery to final repair.

[0008] Moreover, the scheduling server in the present application does not simply allocate tasks according to distance or load, but dynamically judges based on terrain attributes (flat / non-flat) to form an intelligent decision mechanism with environmental awareness, which can cope with variable industrial site environments (such as temporary obstacles, water accumulation, and maintenance barriers). BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be considered as limiting the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0010] Figure 1 A framework diagram of the heterogeneous robot cooperative operation system in the embodiments of the present application is shown; Figure 2 A flowchart of the heterogeneous robot cooperative operation method in the embodiments of the present application is shown; Figure 3 Another flowchart of the heterogeneous robot cooperative operation method in the embodiments of the present application is shown. DETAILED DESCRIPTION

[0011] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application.

[0012] The components of the embodiments of the present application generally described and illustrated herein can be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of the present application.

[0013] Hereinafter, the terms "include", "have", and their conjugates used in various embodiments of the present application merely indicate that specific features, numbers, steps, operations, elements, components, or combinations thereof are present and do not exclude the presence or addition of one or more other features, numbers, steps, operations, elements, components, or combinations thereof. In addition, the terms "first", "second", "third", and the like are used only to distinguish the description, and cannot be understood as indicating or implying a relative importance.

[0014] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which various embodiments of the present application belong. The terms (such as terms defined in a generally used dictionary) will be interpreted as having the same meaning as the context in the relevant technical field and will not be interpreted as having an idealized or overly formal meaning unless clearly defined in various embodiments of the present application.

[0015] Some embodiments of the present application will be described in detail below with reference to the accompanying drawings. The following embodiments and features in the embodiments can be combined with each other without conflict.

[0016] The heterogeneous robot collaborative work system will be described below in combination with some specific embodiments. Figure 1 A framework diagram of the heterogeneous robot collaborative work system according to an embodiment of the present application is shown. Exemplarily, the heterogeneous robot collaborative work system includes a wheeled mobile robot, a scheduling server, and a quadruped mobile robot.

[0017] The wheeled mobile robot refers to an autonomous mobile platform based on a multi-wheel chassis, which has high-efficiency patrol and precise operation capabilities in flat or lightly unstructured terrain, and is used to perform primary abnormal detection tasks and perform maintenance operations on abnormal equipment according to scheduling instructions.

[0018] The scheduling server refers to a centralized computing unit for receiving the first-level abnormal data packet reported by the wheeled mobile robot, judging the terrain category based on the abnormal position; when it is a non-flat terrain, scheduling the quadruped mobile robot to perform fine detection, and generating a maintenance task instruction according to the detection result and issuing it to the wheeled mobile robot, to realize the cooperative decision and task arrangement of the heterogeneous robots.

[0019] The quadruped mobile robot refers to a kind of mobile robot with four active drive legs, which can stably pass through complex non-flat terrains such as stairs, gullies and gravel, and is used to receive the second-level detection task instruction issued by the scheduling server, and perform high-precision local perception and data collection in the specified area.

[0020] Specifically, the wheeled mobile robot is responsible for executing the first-level detection task instruction and performing maintenance operation on the abnormal equipment according to the maintenance task instruction. The scheduling server is responsible for obtaining the first-level abnormal data packet uploaded by the wheeled mobile robot executing the first-level detection task instruction, the first-level abnormal data packet including the event abnormal position; determining the terrain category of the area where the abnormal event is located based on the event abnormal position in the first-level abnormal data packet; if the terrain category is a non-flat terrain, generating and sending a second-level detection task instruction to the quadruped mobile robot to trigger the quadruped mobile robot to detect the area where the abnormal event is located, and obtaining a second-level abnormal data packet; generating and sending a maintenance task instruction to the wheeled mobile robot based on the abnormal equipment in the second-level abnormal data packet. The quadruped mobile robot is responsible for receiving the second-level detection task instruction sent by the scheduling server and executing the second-level detection task instruction.

[0021] In an example, the wheeled mobile robot includes a multi-wheel moving structure, a navigation module, an on-board control module, a communication module and a maintenance module.

[0022] The multi-wheel moving structure refers to the moving chassis part of the wheeled mobile robot, which is composed of a plurality of independently driven active wheels, supports omnidirectional movement and high-precision positioning, and can autonomously navigate on flat ground and carry a mechanical arm and maintenance tools to complete the work task.

[0023] The navigation module refers to an environment perception unit integrated on the wheeled mobile robot, including a laser radar, a depth camera, an inertial measurement unit and an encoder, for constructing an environment map, realizing simultaneous localization and mapping, and generating a first-level abnormal data packet based on sensor data analysis.

[0024] The on-board control module refers to an embedded control system installed inside the wheeled mobile robot, which is used to execute the first-level detection task instruction, process local sensor data, control chassis movement and mechanical arm action, and respond to the maintenance task issued by the scheduling server.

[0025] The communication module refers to the data transmission interface between the wheeled mobile robot and the dispatch server, realizes bidirectional communication based on Wi-Fi6 or 5G private network, is used for uploading abnormal data packets, receiving detection or maintenance task instructions, and ensuring real-time and reliable information interaction between systems.

[0026] The maintenance module refers to a functional component on the wheeled mobile robot, including a multi-degree-of-freedom collaborative manipulator, a modular dexterous hand and a tool quick-change device, and is used for automatically replacing maintenance tools to complete fine maintenance operations such as bolt fastening and component replacement under force-position hybrid control.

[0027] Specifically, the multi-wheel moving structure is a chassis structure composed of multiple driving wheels, which is used for moving and steering. The navigation module is responsible for environment perception and generating a first-level abnormal data packet. The vehicle-mounted control module is responsible for executing a first-level detection task instruction. The communication module is responsible for sending the first-level abnormal data packet to the dispatch server and receiving the maintenance task instruction issued by the dispatch server. The maintenance module is responsible for performing maintenance operations on abnormal equipment.

[0028] In another example, the quadruped mobile robot comprises a quadruped moving structure, a sensor suite, a communication module and a vehicle-mounted control module.

[0029] The quadruped moving structure refers to the body movement mechanism of the quadruped mobile robot, which contains four support legs with more than three active degrees of freedom, realizes walking, climbing stairs, obstacle crossing and fall self-recovery through coordinated control, and guarantees the accessibility and stability of the quadruped mobile robot in complex terrain.

[0030] The sensor suite refers to a combination of high-precision detection equipment installed on the quadruped mobile robot, including a zoom camera, a macro thermal imager, an ultrasonic flaw detector, a microphone array and a gas sensor, which is used for performing multi-modal fine detection on the target and generating a second-level abnormal data packet.

[0031] The communication module refers to the data transmission interface between the quadruped mobile robot and the dispatch server, realizes bidirectional communication based on Wi-Fi6 or 5G private network, is used for uploading abnormal data packets, receiving detection or maintenance task instructions, and ensuring real-time and reliable information interaction between systems.

[0032] The vehicle-mounted control module refers to the main control system on the quadruped mobile robot, which is used for analyzing the second-level detection task instruction issued by the dispatch server, planning a complex terrain path, controlling the quadruped movement behavior, and coordinating the sensor suite to complete the fixed-point data acquisition and preprocessing.

[0033] Specifically, the quadruped moving structure comprises a plurality of support leg structures, which are used for moving and steering. The sensor suite is responsible for environment perception and generating a second-level abnormal data packet. The communication module is responsible for receiving the second-level detection task instruction sent by the dispatch server; and the vehicle-mounted control module is responsible for executing the second-level detection task instruction.

[0034] Optionally, the wheeled mobile robot performs flat area inspection, reports a first-level abnormal data packet after discovering an anomaly, receives a dispatch instruction, and goes to the scene to perform maintenance operations.

[0035] The quadruped mobile robot: according to the dispatch instruction (i.e., a second-level detection task instruction), enters the complex terrain area, approaches the abnormal point for high-definition photography, temperature measurement, and other detailed detection, and returns a second-level abnormal data packet.

[0036] The dispatch server: analyzes the terrain category according to the abnormal position, sends a quadruped mobile robot for re-inspection if it is a non-flat terrain, and generates a maintenance work order based on the result to assign a wheeled mobile robot for maintenance.

[0037] The operation and maintenance personnel terminal: monitors the system running state, participates in fault confirmation or approves high-risk operations when needed, and realizes man-machine collaborative control.

[0038] The heterogeneous robot collaborative work method will be described below in conjunction with some specific embodiments.

[0039] Figure 2 A flowchart of the heterogeneous robot collaborative work method of the embodiments of the present application is shown. Exemplarily, the heterogeneous robot collaborative work method includes the following steps: Step S202, obtaining a first-level abnormal data packet uploaded by a wheeled mobile robot executing a first-level detection task instruction, the first-level abnormal data packet including an event abnormal position.

[0040] The first-level detection task instruction is a periodic or on-demand execution wide-area inspection command issued by the dispatch server to the wheeled mobile robot, including the inspection area range, path planning parameters, sensor activation configuration, and data reporting frequency, used to drive the wheeled mobile robot to run autonomously in flat terrain and complete environmental preliminary screening.

[0041] The first-level abnormal data packet is a data set generated by the wheeled mobile robot when it recognizes a potential anomaly through its navigation module during the execution of the first-level detection task, including at least: abnormal type, confidence, global coordinates (i.e., event abnormal position), timestamp, and one or more panoramic snapshot images, used to report preliminary abnormal information to the dispatch server.

[0042] The event abnormal position is the geographical spatial coordinates corresponding to the occurrence of the first-level anomaly, expressed in a three-dimensional Cartesian coordinate system (x, y, z), and aligned with the high-precision semantic map, serving as the reference positioning basis for subsequent terrain category judgment, task scheduling, and path planning.

[0043] Specifically, the heterogeneous robot collaborative work system is deployed in, for example, an industrial park. The dispatch server issues a first-level detection task instruction to a patrol maintenance robot (i.e., a wheeled mobile robot) to instruct it to regularly patrol the power distribution room, pipeline area, and equipment room along a predetermined route.

[0044] The wheeled mobile robot is equipped with a laser radar, an infrared thermal imager, and a high-definition panoramic camera to collect environmental data in real time during travel. The vehicle-mounted control module runs a lightweight AI model to locally analyze sensor data. When it detects that the surface temperature of a power distribution cabinet exceeds a preset threshold (e.g., 70°C) and the visual model identifies that the instrument pointer deviates from the normal interval, it determines that there is a potential anomaly.

[0045] Immediately, the wheeled mobile robot stops moving, calls the navigation module to lock the current position, and generates a first-level anomaly data packet. The data packet contains: anomaly type "overheating + instrument anomaly", confidence 85%, event anomaly location (X=12.4m, Y=5.6m, Z=0.0m), timestamp 2025-12-29T10:15:23, and a 360° panoramic image. The communication module uploads the first-level anomaly data packet to the dispatch server in real time through the 5G private network, completing the initial report of the anomaly.

[0046] Through the above embodiment, the wheeled mobile robot completes intelligent preliminary screening while performing regular patrols, avoiding the bandwidth pressure caused by the return of massive raw data. By generating a standardized first-level anomaly data packet, it ensures the completeness and analyzability of key information, improving the overall response efficiency of the system.

[0047] Step S204, based on the event anomaly location in the first-level anomaly data packet, determine the terrain category of the region where the abnormal event occurs.

[0048] In one embodiment, based on the event anomaly location in the first-level anomaly data packet, determine the region where the abnormal event occurs; divide the region where the abnormal event occurs into multiple sub-regions, and for each sub-region, calculate the ground slope and road width of each sub-region; based on the ground slope and road width, determine the terrain category of the sub-region; if the terrain category of each sub-region is flat terrain, determine the terrain category of the region where the abnormal event occurs as flat terrain; if there is at least one sub-region whose terrain category is non-flat terrain, determine the terrain category of the region where the abnormal event occurs as non-flat terrain.

[0049] Among them, the region where the abnormal event occurs refers to the geographical region formed by expanding a preset spatial range (such as a radius of 2 meters) centered on the "event anomaly location" contained in the first-level anomaly data packet, which is used as a basic analysis unit for subsequent terrain passability evaluation to ensure that no potential obstacles are missed.

[0050] The ground slope refers to an included angle between a ground plane normal vector obtained by fitting point cloud data in a certain sub-region and a direction of gravitational acceleration, is used to quantify the inclination degree of the ground in the region, and is an important parameter for judging whether a wheeled mobile robot has a stable passing ability.

[0051] The road width refers to a continuous unobstructed transverse distance available for the robot to pass in the current sub-region along a predetermined advancing direction, is extracted according to three-dimensional point cloud or semantic map, and is used to judge whether the path meets the physical size passing demand of the wheeled mobile robot.

[0052] The terrain category refers to a classification label used to represent whether a certain region is suitable for the wheeled mobile robot to pass, including two categories: flat terrain: the ground slope is less than a preset threshold (such as 10°) and the road width is greater than or equal to the width of the robot body; and non-flat terrain: any condition is not met, such as an area where there is a steep slope, a step, a ditch, a gravel area, etc., which causes the wheeled robot to be unable to safely reach.

[0053] In one of the embodiments, a preset slope threshold and a size of the wheeled mobile robot are obtained; the ground slope is compared with the preset slope threshold, and the road width is compared with the size of the wheeled mobile robot; if the ground slope is less than the preset slope threshold and the road width is less than the size of the wheeled mobile robot, it is determined that the terrain category of the sub-region is flat terrain; if the ground slope is not less than the preset slope threshold or the road width is not less than the size of the wheeled mobile robot, it is determined that the terrain category of the sub-region is non-flat terrain.

[0054] The preset slope threshold refers to a maximum allowable inclination angle used to judge whether the ground is suitable for the wheeled mobile robot to safely pass, and the unit is degree. The threshold is set according to the chassis structure, driving ability, center of gravity distribution and anti-skid performance of the wheeled mobile robot. When the ground slope of a certain region is less than the threshold, it is considered to have stable driving conditions; if it is equal to or exceeds the threshold, it is determined that there is a risk of slipping and overturning, which belongs to non-flat terrain and needs to be avoided or intervened by other types of robots.

[0055] The size of the wheeled mobile robot refers to the maximum physical outline size of the wheeled mobile robot body in the horizontal projection plane, including its length, width and ground clearance, wherein the key parameter for passing judgment is the maximum width (i.e. the transverse distance between the two sides of the wheel or the outermost point of the structure). The size is used to compare with the road width to judge whether the robot can pass through a narrow channel. If the path road width is less than this size, it is considered to be a physically unreachable region.

[0056] Optionally, the "preset slope threshold value" and the "size of the wheeled mobile robot" are both device capability parameters relied on by the scheduling server when performing terrain classification decisions, pre-stored in the system knowledge base, and support dynamic configuration by model. For example, different models of wheeled robots have different structural parameters and obstacle crossing capabilities. The system can automatically load the corresponding threshold value and size data according to the actual deployment model, improving compatibility and adaptability.

[0057] Specifically, the scheduling server receives a first abnormal data packet reported by the wheeled mobile robot, which records the event abnormal position, for example (X = 8.7m, Y = 14.2m, Z = 0.0m), located at the corner of the factory equipment passage. Then start the terrain judgment process.

[0058] Based on the coordinate point, a circular area with a radius of 2 meters is drawn as the abnormal event area, and it is evenly divided into four fan-shaped sub-areas (A, B, C, D), corresponding to the north, east, south and west directions respectively. Then call the high-precision semantic map and the real-time updated laser point cloud data to analyze each sub-area: In sub-area A, the neighbor point set of multiple ground points to be evaluated is calculated, the plane fitting operation within a radius of 0.3 meters is performed, the ground normal vector is obtained, and the average ground slope is calculated as 6.5°; At the same time, the distance between the two equipment rooms is detected, and the road width is measured as 0.8 meters; Given that the wheeled mobile robot body width is 0.75 meters, and the preset slope threshold value is 10°; The judgment result is: 6.5°<10°, and 0.8m>0.75m, so it is judged that sub-area A belongs to flat terrain.

[0059] In sub-area B, the ground slope is calculated as 12.3°, which exceeds the preset threshold value; Although the road width is 0.9 meters, there is a risk of slipping due to the large slope; According to the judgment logic: "if the ground slope ≥ preset threshold value or the road width < robot size", it is classified as non-flat terrain; Therefore, sub-area B is marked as non-flat terrain.

[0060] Since at least one sub-area (B) is determined to be non-flat terrain, the terrain category of the entire abnormal event area is finally determined to be non-flat terrain. Accordingly, the scheduling server generates a second detection task instruction and prepares to dispatch a quadruped mobile robot to perform detailed detection.

[0061] The traditional method often makes a rough judgment on whether it can be reached. The present application subdivides the area into multiple sub-regions, respectively calculates the ground slope and road width, avoids misjudgment of a single sampling point, and improves the accuracy and robustness of terrain recognition. Moreover, not only the slope is considered, but also the robot's own physical size and road width are compared to comprehensively judge the feasibility of passing, fully embodying the "robot-centered" environmental adaptation concept, and preventing the engineering problem of "map display can pass but actually stuck".

[0062] In one of the embodiments, a plurality of to-be-evaluated ground points are determined in the current sub-region; for each to-be-evaluated ground point, a plane fitting operation is performed on the adjacent ground points within a preset radius range from the to-be-evaluated position point, to obtain a ground surface normal vector of the current sub-region; and based on an included angle between the normal vector and the direction of the gravitational acceleration, a ground slope of the current sub-region is calculated.

[0063] The sub-region refers to a local unit formed by dividing the "abnormal event region" according to the spatial position, which is used to realize fine terrain feature extraction. Each sub-region independently performs ground slope and road width analysis to improve the accuracy and robustness of terrain classification, and is usually divided in a grid or sector manner.

[0064] The to-be-evaluated ground point refers to a representative three-dimensional space point located in a sub-region, which is selected from the point cloud data obtained by a laser radar or a depth camera, and is used as a reference point for analyzing the surface characteristics of the sub-region. The Z-axis value represents the local ground height, which is the starting position for performing the plane fitting operation.

[0065] The preset radius range refers to a local neighborhood range drawn with the to-be-evaluated ground point as the center, and the typical value is 0.2 meters to 0.5 meters. It is used to screen the surrounding point set participating in the plane fitting, to ensure that the fitted plane reflects the continuous and stable surface trend, and to avoid interference from isolated noise points.

[0066] The adjacent ground point refers to a plurality of three-dimensional points located within the preset radius range and belonging to the same continuous surface structure, which are used to participate in the least squares plane fitting together with the to-be-evaluated ground point, to solve the geometric shape of the local surface.

[0067] The ground surface normal vector refers to the unit normal vector obtained by performing plane fitting on the to-be-evaluated ground point and its adjacent ground points, which represents the inclination direction of the local surface. Its direction in the three-dimensional space directly determines the size of the ground slope, and is the core parameter for slope calculation.

[0068] The direction of gravitational acceleration refers to the vertical downward direction under the action of the earth's gravity, which is usually represented as a unit vector (0, 0, -1) in the coordinate system. As a reference benchmark, it is used for angle calculation with the ground normal vector to obtain the actual inclination angle of the ground relative to the horizontal plane (i.e., the ground slope).

[0069] Specifically, it is judged whether there is a steep slope obstacle in the channel between certain devices. The region where the abnormal event occurs is divided into four sub-regions, and the ground slope analysis is performed for the east sub-region A.

[0070] First, three representative ground points P1, P2, P3 to be evaluated are selected in the sub-region, with coordinates (8.75m, 14.30m, 0.05m), (8.80m, 14.35m, 0.06m) and (8.85m, 14.40m, 0.07m) respectively, all from the high-precision point cloud data uploaded by the wheeled robot. For the evaluation point P1, a preset radius range of 0.3 meters is set, and all valid points within 0.3 meters from P1 are retrieved from the point cloud database, a total of 12 nearby ground points are selected, excluding suspended pipeline points and wall points, and only the continuously distributed point set on the ground is retained.

[0071] Then, the least squares method is used to perform plane fitting operation on the 13 points (including P1), to solve the optimal fitting plane equation ax + by + cz + d = 0, and to normalize to obtain the ground normal vector n = (a', b', c') of the local ground. Given that the direction of gravitational acceleration is the standard vertical downward vector g = (0, 0, -1), the angle θ between the two is the ground slope. After calculation, c' is equal to 0.978, and the corresponding angle θ is 12°, which is greater than the preset slope threshold of 10°. It is determined that there is a significant inclination in the sub-region, which belongs to non-flat terrain. It can be understood that the remaining two points P2 and P3 are also processed in the above manner, and finally it is determined that there is a significant inclination in the sub-region, which belongs to non-flat terrain.

[0072] The present application realizes the accurate quantitative measurement of the ground slope in complex industrial environments through the reasonable division of the "sub-region", the scientific selection of the "evaluation ground point", and the plane fitting based on the "nearby ground point" to solve the "ground normal vector". Not only does it solve the problem of strong subjectivity of traditional visual judgment, but also embodies the advancement of the invention in the aspects of environmental cognition intelligence and data-based decision-making process.

[0073] Step S206, if the terrain category is non-flat terrain, a secondary detection task instruction is generated and sent to the quadruped mobile robot to trigger the quadruped mobile robot to detect the region where the abnormal event occurs, and obtain a secondary abnormal data packet.

[0074] The secondary detection task instruction is a task command generated by the scheduling server and sent to the quadruped mobile robot, used to trigger it to go to the abnormal event area to perform high-precision, multi-modal fine detection operation, including target position, to-be-detected point list, stay time, detection orientation and path planning parameters and the like.

[0075] The secondary abnormal data packet is a data set returned by the quadruped mobile robot after completing fine detection, at least including: high-definition image, macro thermal map, sound spectrum, ultrasonic flaw detection result, device state characteristic value (such as maximum temperature, abnormal sound frequency), global coordinates and timestamp, used to support subsequent fault diagnosis.

[0076] In one of the embodiments, the device distribution chart and the abnormal event area are matched to determine a plurality of to-be-detected points in the abnormal event area, each to-be-detected point being configured with a preset stay time and a preset detection orientation; a semantic road network is constructed based on a flat terrain sub-area, the semantic road network including a plurality of nodes and edges connecting the nodes, each edge being configured with a weight according to a passing condition, the weight including path length and preset passing time; a target detection path is constructed by fusing the semantic road network and the plurality of to-be-detected points, and a secondary detection task instruction is generated and sent to the quadruped mobile robot based on the target detection path.

[0077] The device distribution chart is a structured data table or a three-dimensional model layer stored in the scheduling server, recording the spatial layout, type identification, key detection part and priority information of each type of device in the work area, used to match the abnormal area and determine the detection points that need to be checked.

[0078] The to-be-detected point is one or more specific detection positions pre-set in the abnormal event area for the purpose of comprehensive diagnosis, usually corresponding to the key components of the device (such as instrument panel, terminal, valve handle), each point being configured with independent detection parameters.

[0079] The preset stay time is the length of time for the quadruped mobile robot to remain stationary at a to-be-detected point to complete data acquisition, a typical value being 3-10 seconds, ensuring that the sensor fully acquires stable signals (such as thermal imager temperature convergence, audio sampling completeness).

[0080] The preset detection orientation is the direction in which the quadruped robot should adjust the body posture after arriving at the to-be-detected point, so that the sensor main shaft is perpendicular to the specific surface of the point (such as perpendicular to the instrument panel surface), to obtain the best imaging quality and measurement accuracy.

[0081] The semantic road network is a topological network with attribute labels constructed based on a high-precision map, composed of “nodes” and “edges”, wherein the nodes represent positions where the robot can stop or turn, and the edges represent paths that can be traveled, and are labeled with passing conditions, distance, estimated passing time and the like metadata.

[0082] Edge refers to the path unit connecting two adjacent nodes in the semantic road network, representing a drivable route, with directionality or no directionality, and its weight can be used for path optimization calculation.

[0083] Node refers to the discrete position anchor point in the semantic road network, usually set at key geographical locations such as corridor intersections, equipment, stair starts, etc., as the basic unit of path planning.

[0084] Passing condition refers to the comprehensive basis for judging whether a certain "edge" is suitable for the robot to pass through, including factors such as road surface material, slope, obstacle density, and light intensity, affecting path selection and weight distribution.

[0085] Path length refers to the geometric distance of a "edge" in the semantic road network, measured in meters, and is one of the basic parameters for measuring travel cost.

[0086] Pre-set passing time refers to the time estimated by the four-legged mobile robot to pass through a certain "edge" based on historical data or motion model, measured in seconds, affected by factors such as terrain complexity and speed limit, and used for path timeliness evaluation.

[0087] Target detection path refers to the optimal inspection route generated by integrating the structure of the semantic road network and multiple detection points, meeting the requirements of visiting all necessary detection points in the shortest time while considering response flexibility and energy efficiency, and ultimately guiding the four-legged robot to perform tasks.

[0088] Specifically, it is determined that the area of the abnormal event in the distribution room is located near the stair platform, which belongs to a non-flat terrain, so the dispatch server starts the process of dispatching a detailed detection task to the four-legged mobile robot.

[0089] First, call the stored device distribution chart and spatially match the abnormal area with the device library to identify that there are three high-voltage switch cabinets (A1-A3) in the area. Further extract the key detection points on each cabinet: the current meter dial of A1 cabinet, the bus connection of A2 cabinet, and the heat dissipation air outlet of A3 cabinet, which are set as three detection points P1, P2, and P3. For each detection point: configure the preset dwell time to be 5 seconds to ensure that the infrared thermal imager completes temperature stable acquisition; set the preset detection direction to be straight ahead + 15° pitch angle to ensure that the camera lens is perpendicular to the target surface and avoid glare interference.

[0090] Then, a semantic road network is constructed based on the known flat ground parts in the area (such as the bottom passage of the stairs, the corner platform). Six nodes N1-N6 are set, respectively corresponding to the entrance, corner, front side of the equipment, and other safe passage points; and these nodes are connected by 8 edges e1-e8 to form a connected graph. Optionally, each edge is marked with a path length (such as e3=2.4m); a preset travel time (such as e3=8s, considering slow climbing); and a travel condition (such as "allowed to pass" and "need low speed mode").

[0091] Next, the above semantic road network and the three to-be-detected points P1-P3 are fused, and an initial detection path is constructed using a path search algorithm: N1→e2→N3(P1)→e4→N4(P2)→e6→N5(P3). Finally, a complete target detection path is generated and encapsulated as a secondary detection task instruction, which includes: target area coordinates; to-be-detected point list and respective parameters (position, stay time, orientation); optimal path sequence (nodes and edges); and emergency interruption strategy. The secondary detection task instruction is issued to the four-legged mobile robot with the highest current, closest distance, and idle through the communication module, starting the detection process.

[0092] Through the above embodiment, the target detection path ensures that all key detection points are covered, solving the problem of easy omission of details in traditional manual dispatching, and being particularly suitable for industrial scenes with dense equipment and complex structure. Moreover, by matching the equipment distribution chart with the abnormal area, the key to-be-detected points are automatically identified, and preset stay time and preset detection orientation are configured for them, ensuring consistency and repeatability of each detection process and improving diagnosis reliability.

[0093] In one of the embodiments, an initial detection path is constructed according to the semantic road network and the plurality of to-be-detected points, the initial detection path satisfying that each to-be-detected point is visited within a preset detection time length; a weighted function for adjusting the initial detection path is obtained, a first function item in the weighted function representing total travel time of the four-legged mobile robot along the initial detection path; a second function item in the weighted function representing expected response time of the four-legged mobile robot from any position on the initial detection path to the closest to-be-detected point; and based on the detection priority carried by the first-level abnormal data packet, weight coefficients of the first function item and the second function item are adjusted to obtain a target detection path.

[0094] The initial detection path refers to a feasible inspection route preliminarily generated based on the semantic road network structure and position information of the plurality of to-be-detected points, ensuring that all to-be-detected points are visited within a preset time, serving as a basic solution for subsequent optimization.

[0095] The preset detection duration refers to the upper limit of the total allowable time from the departure of the quadruped mobile robot to the completion of the entire fine detection task, which is usually set to 5-30 minutes, and is used to constrain the rationality of the initial path to prevent the task from timing out and affecting the overall scheduling efficiency of the system.

[0096] The weighting function refers to an objective function used to evaluate and optimize the performance of the path, which is a linear combination of multiple factors, and is used to quantify the comprehensive cost of different paths and guide the generation of the optimal target detection path, including a first function term and a second function term.

[0097] The first function term refers to the total travel time (in seconds) required for the quadruped mobile robot to run along the initial detection path, which is the core component of the weighting function reflecting the efficiency of task execution, and the smaller the value, the more efficient the path.

[0098] The second function term refers to the average or maximum expected response time (in seconds) from any position on the path to the nearest detection point, reflecting the emergency flexibility and coverage balance of the system, and the smaller the value, the stronger the local response capability.

[0099] The expected response time refers to the estimated value of the time required for the quadruped robot to reach the nearest detection point from a certain point on the current travel path when a new abnormality occurs in a device, which is used to measure whether the path layout is conducive to rapid review or expanded detection.

[0100] The detection priority refers to a hierarchical label (such as high, medium, and low) carried by the first-level abnormality data packet, reflecting the urgency of the abnormal event, and directly affecting the subsequent resource allocation strategy; high-priority tasks will obtain more optimal path planning and higher scheduling authority.

[0101] The weight coefficient refers to the adjustment parameter before each function term in the weighting function, which is used to control the balance between different optimization objectives.

[0102] Specifically, a fine detection path for the quadruped mobile robot to access the three power distribution cabinets in the three-layer equipment room needs to be planned. The abnormal event is determined to be a "high temperature warning", and the detection priority marked in the first-level abnormality data packet is "high".

[0103] First, an initial detection path is constructed based on the semantic road network and three points to be detected, P1, P2, and P3. The Traveling Salesman Problem algorithm is used to find the shortest traversal order: starting point → N1 → N2(P1) → N4 → N5(P2) → N6 → N7(P3). The total travel time is calculated to be 280 seconds, which meets the preset detection duration requirement of 300 seconds. Then, a weighted function is introduced for path optimization: the first function term represents the overall energy consumption and task cycle; the second function term calculates the average response delay from each point on the path to the nearest detection point by simulating a random interruption scenario. Since this anomaly belongs to the high-priority detection category (from a first-level anomaly data packet), the weight coefficients are automatically adjusted to enhance the path's responsiveness to sudden events.

[0104] Then, based on this, several candidate paths were re-evaluated, and a new path, although slightly longer, with a more even distribution and faster response, was finally selected as the target detection path. This path was encapsulated in the secondary detection task instruction and successfully sent to the quadrupedal mobile robot with the best idle state to start the detection process.

[0105] Traditional path planning only pursues the "shortest path" or "fewest nodes," while this application introduces a weighted function to dynamically adjust the optimization objective based on task attributes (such as detection priority), enabling the system to ensure both efficiency and emergency response. Furthermore, the second function term, "expected response time," effectively measures the spatial coverage quality of the path. By optimizing this term, the problem of slow local responses caused by "excessively long path branches" is avoided, improving the overall task reliability.

[0106] Step S208: Based on the abnormal device in the secondary abnormal data packet, generate and send a maintenance task instruction to the wheeled mobile robot to trigger the wheeled mobile robot to perform maintenance operations on the abnormal device.

[0107] Among them, abnormal equipment refers to specific industrial devices (such as power distribution cabinets, pumps, valves, motors, etc.) that are identified as having faults or potential risks in the Level 2 abnormal data package. Their locations are uniquely identified by coordinates and equipment numbers, and they are the target objects for subsequent maintenance operations.

[0108] Maintenance task instructions are structured commands generated by the scheduling server and issued to wheeled mobile robots. They include information such as target device ID, operation type (e.g., tightening, replacement), tool selection, path planning parameters, and safety verification procedures, and are used to drive the robot to perform closed-loop maintenance operations.

[0109] In one embodiment, a maintenance task instruction is generated; the current status information of all quadruped mobile robots is obtained, including the current remaining battery power, current task idleness, and distance to the area where the abnormal event is located; for each quadruped mobile robot, a weighted scoring function is used to calculate the matching degree of the current remaining battery power, current task idleness, and distance to the area where the abnormal event is located for the corresponding quadruped mobile robot, and the matching degree of each quadruped mobile robot is obtained; the maintenance task instruction is sent to the wheeled mobile robot with the highest matching degree.

[0110] The current status information refers to a set of dynamic operating parameters reported by the quadrupedal mobile robot to the scheduling server, which is used to evaluate whether it is suitable as a candidate for task execution. These parameters include three core indicators: remaining battery power, task idle time, and distance from the target area.

[0111] Remaining battery power refers to the current state of charge of the quadruped mobile robot's battery, expressed as a percentage or remaining working time, reflecting its endurance; a threshold (such as ≥30%) is usually set as one of the conditions for dispatch.

[0112] The current task idleness refers to the status flag that indicates whether the quadrupedal mobile robot is currently performing other tasks. It is divided into three levels: "idle", "lightly loaded" and "heavy loaded". Only when it is idle or lightly loaded can it be capable of accepting new tasks.

[0113] The distance to the area where the abnormal event occurs refers to the ground path distance (in meters) between the current position of the quadruped mobile robot and the center point or the nearest reachable point of the area where the abnormal event occurs. It is used to measure the timeliness of the response, and the shorter the distance, the higher the priority.

[0114] The weighted scoring function is a multi-factor comprehensive evaluation model used to quantify the degree of adaptability of each quadrupedal mobile robot to the task.

[0115] Matching degree refers to a value calculated by a weighted scoring function, which represents the overall suitability of a quadrupedal mobile robot to perform a specific maintenance task; the higher the matching degree, the better the robot is in terms of power, load and position.

[0116] For example, the system has completed a detailed inspection of the high-voltage junction box in a stairwell and confirmed a "loose bolt" fault. The dispatch server then initiates the maintenance task scheduling process.

[0117] First, the system generates a maintenance task instruction, which includes: Abnormal equipment ID: PDB-203; Fault type: loose mechanical connection; Operation steps (move to the ready position, automatically switch to an electric wrench, tighten the M8 bolt, switch to a camera tool, and take a verification image); Verification mechanism) and uploads the result image for system review.

[0118] Then, the system retrieves the current status information of all available wheeled mobile robots from the network, including the quadruped mobile robot's ID, remaining battery power, current task idle time, and distance to the abnormal area. For each quadruped mobile robot, a preset weighted scoring function is called to calculate the matching degree, ultimately obtaining the quadruped mobile robot with the highest matching degree. The quadruped mobile robot corresponding to the highest matching degree is selected as the execution subject, and the complete maintenance task instruction is sent out in real time through the communication module. After receiving the instruction, the quadruped mobile robot corresponding to the highest matching degree immediately plans a path to the support point near the abnormal device, completes the maintenance operation, and sends back verification images, achieving closed-loop processing.

[0119] Understandably, this application adopts a collaborative process of "wheeled inspection → quadruped precision inspection → wheeled maintenance," rather than direct maintenance by quadruped mobile robots. The main reason is the mismatch between the physical characteristics and task requirements of the two types of platforms. Wheeled mobile robots have high mobility, stable chassis, and strong load capacity on flat terrain, making them suitable as a reliable base for maintenance operations and capable of supporting multi-degree-of-freedom robotic arms to achieve high-precision force and position control. While quadruped mobile robots have the ability to traverse complex terrain and can be used for precise inspection in hard-to-reach areas, their bodies exhibit dynamic fluctuations, limited load capacity, and short endurance, making it difficult to meet the stability and precise control requirements of maintenance. Furthermore, quadruped mobile robots are expensive, and equipping them with maintenance functions would reduce the system's cost-effectiveness. Therefore, a heterogeneous division of labor—with quadruped mobile robots focusing on "arrival and perception" and wheeled mobile robots responsible for "stable execution"—can leverage the advantages of each while achieving an efficient, economical, and scalable "inspection-maintenance" closed loop, representing a more suitable optimized design for industrial realities.

[0120] By introducing a weighted scoring function that comprehensively considers three dimensions—power consumption, load, and distance—the energy consumption imbalance caused by relying on a single criterion (such as distance alone) is avoided. This ensures that the system always selects the robot with the optimal overall condition to perform tasks, extending the overall service life of the cluster. This process embodies a shift from "passive assignment" to "proactive optimization," and is a key manifestation of the core concept of intelligent collaboration in this application, significantly improving the level of operation and maintenance automation.

[0121] Figure 3 Another flowchart of a heterogeneous robot collaborative operation method according to an embodiment of this application is shown. Exemplarily, the operation process mainly includes: Patrol findings: The wheeled mobile robot autonomously cruises in flat areas according to the inspection plan issued by the scheduling server and collects environmental data in real time through its onboard wide-area inspection sensor suite; the onboard control module runs a lightweight AI model to perform edge analysis. When it detects potential faults such as abnormal temperature, visual deviation or abnormal sound, it generates a first-level abnormal data packet containing the abnormality type, confidence level and abnormal location of the event, and reports it to the scheduling server.

[0122] Scheduling decision: After receiving the first-level abnormal data packet, the scheduling server determines whether the area where the abnormal event is located belongs to non-flat terrain based on the spatial information of the abnormal location in the semantic map. If it is flat terrain, a self-inspection or maintenance process is triggered. If it is non-flat terrain, it is determined that the wheeled robot cannot reliably reach it. Then, the coordination mechanism is activated to generate a second-level detection task instruction and assign the quadrupedal mobile robot in the best state to perform fine detection.

[0123] Detailed detection: After receiving scheduling instructions, the quadrupedal mobile robot plans a movement path that adapts to complex terrain, stably traversing stairs, ditches, or gravel areas to approach abnormal equipment; it uses sensors such as high-resolution zoom cameras, macro thermal imagers, and microphone arrays to complete multimodal data acquisition, obtain high-definition images, accurate temperature rise, and local acoustic characteristics, and form a secondary abnormal data packet to be sent back to the server, providing a high-confidence basis for fault diagnosis.

[0124] Patrol robot self-inspection: For abnormal points located on flat terrain, the system can choose to have the original wheeled mobile robot perform a self-inspection: when it travels to the vicinity of the target, it calls higher precision sensors to perform close-range imaging or temperature measurement to verify the authenticity of the initial alarm, avoid false alarms that cause redundant scheduling, and improve the system's operating efficiency.

[0125] Maintenance execution: The server combines the secondary abnormal data packets with the maintenance knowledge base to automatically match the maintenance process corresponding to the fault type and generate a structured maintenance work order; it comprehensively evaluates the remaining battery power, task load and distance factors of each wheeled mobile robot, calculates the matching degree through a weighted scoring function, and issues the maintenance task instruction to the optimal robot; the robot moves to the operation support point, automatically changes tools, and completes precise maintenance operations such as bolt tightening and component replacement through visual servo and force-position hybrid control.

[0126] Closed-loop verification: After maintenance is completed, the system initiates a verification mechanism: the original wheeled robot can take images to make a preliminary confirmation, or the quadruped robot can be dispatched again to perform a retest (such as secondary temperature measurement or comparative imaging); after the verification is passed, the server updates the equipment status to "normal", closes the work order and records a complete operation and maintenance log, realizing the fully automated closed-loop management from "discovering the problem" to "solving the problem".

[0127] It is understood that the device in this embodiment corresponds to the heterogeneous robot collaborative operation method in the above embodiments, and the options in the above embodiments are also applicable to this embodiment, so they will not be described again here.

[0128] This application also provides a terminal device, exemplary of which includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to enable the terminal device to perform the functions of the various modules in the above-described heterogeneous robot collaborative operation method or the above-described heterogeneous robot collaborative operation system.

[0129] The processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, including at least one of a Central Processing Unit (CPU), Graphics Processing Unit (GPU), Network Processor (NP), Digital Signal Processor (DSP), Application-Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application.

[0130] Memory can be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), and Electrically Erasable Programmable Read-Only Memory (EEPROM). Memory is used to store computer programs, and the processor can execute these programs upon receiving execution instructions.

[0131] This application also provides a computer-readable storage medium for storing computer programs used in the aforementioned terminal devices. For example, the computer-readable storage medium may include, but is not limited to, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0132] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that, as an alternative implementation, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0133] In addition, the functional modules or units in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0134] If a function is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a smartphone, personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application.

[0135] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for collaborative operation of heterogeneous robots, characterized in that, An application is made in a heterogeneous robot collaborative operation system, the system including a wheeled mobile robot, a quadrupedal mobile robot, and a scheduling server, wherein the scheduling server performs the following method: Obtain the first-level abnormal data packet uploaded by the wheeled mobile robot when executing the first-level detection task instruction, wherein the first-level abnormal data packet includes the location of the abnormal event; Based on the abnormal location of the event in the first-level abnormal data packet, the terrain category of the area where the abnormal event is located is determined; If the terrain category is non-flat terrain, a secondary detection task instruction is generated and sent to the quadrupedal mobile robot to trigger the quadrupedal mobile robot to detect the area where the abnormal event is located and obtain a secondary abnormal data packet; Based on the abnormal device in the secondary abnormal data packet, a maintenance task instruction is generated and sent to the wheeled mobile robot to trigger the wheeled mobile robot to perform maintenance operations on the abnormal device.

2. The method according to claim 1, characterized in that, The terrain categories include flat terrain and non-flat terrain; Determining the terrain category of the area where the abnormal event occurs based on the abnormal location of the event in the first-level abnormal data packet includes: Based on the abnormal location of the event in the first-level abnormal data packet, the region where the abnormal event is located is determined; The area where the abnormal event occurred is divided into multiple sub-regions, and for each sub-region, the ground slope and road width are calculated. Based on the ground slope and the road width, determine the terrain category of the target sub-region; If the terrain category of each of the sub-regions is flat terrain, then the terrain category of the region where the abnormal event is located is determined to be flat terrain. If at least one of the sub-regions has a terrain category of non-flat terrain, then the terrain category of the region where the abnormal event occurs is determined to be non-flat terrain.

3. The method according to claim 2, characterized in that, The determination of the terrain category for a specific sub-region based on the ground slope and the road width includes: Obtain the preset slope threshold and the dimensions of the wheeled mobile robot; The ground slope is compared with the preset slope threshold, and the road width is compared with the size of the wheeled mobile robot; If the ground slope is less than the preset slope threshold and the road width is less than the size of the wheeled mobile robot, then the terrain category of the targeted sub-region is determined to be the flat terrain. If the ground slope is not less than the preset slope threshold, or the road width is not less than the size of the wheeled mobile robot, then the terrain category of the targeted sub-region is determined to be the non-flat terrain.

4. The method according to claim 2, characterized in that, The calculation of the ground slope for each sub-region includes: Identify multiple ground points to be evaluated within the current sub-region; For each ground point to be evaluated, a plane fitting operation is performed on the neighboring ground points within a preset radius of the target location point to obtain the surface normal vector of the current sub-region; The ground slope of the current sub-region is calculated based on the angle between the normal vector and the direction of gravitational acceleration.

5. The method according to claim 2, characterized in that, The process of generating and sending secondary detection task instructions to the quadrupedal mobile robot includes: The device distribution map is matched with the area where the abnormal event is located to determine multiple detection points in the area where the abnormal event is located. Each detection point is configured with a preset dwell time and a preset detection orientation. A semantic road network is constructed based on sub-regions with flat terrain. The semantic road network includes multiple nodes and edges connecting the nodes. Each edge is configured with a weight according to the travel conditions. The weight includes path length and preset travel time. The semantic road network and multiple points to be detected are integrated to construct a target detection path, and a secondary detection task instruction is generated and sent to the quadrupedal mobile robot based on the target detection path.

6. The method according to claim 5, characterized in that, The process of constructing a target detection path by fusing the semantic road network and multiple points to be detected includes: Based on the semantic road network and the multiple points to be detected, an initial detection path is constructed, wherein the initial detection path satisfies that each of the points to be detected is visited within a preset detection time. Obtain a weighting function for adjusting the initial detection path, wherein the first function term in the weighting function represents the total travel time of the quadrupedal mobile robot along the initial detection path; and the second function term in the weighting function represents the expected response time of the quadrupedal mobile robot from any position on the initial detection path to the nearest detection point. Based on the detection priority carried by the first-level abnormal data packet, the weight coefficients of the first function term and the second function term are adjusted to obtain the target detection path.

7. The method according to claim 1, characterized in that, The process of generating and sending maintenance task instructions to the wheeled mobile robot includes: Generate maintenance task instructions; Obtain all current status information of the quadrupedal mobile robot, including current remaining battery power, current task idle time, and distance from the area where the abnormal event is located; For each of the quadruped mobile robots, a weighted scoring function is used to calculate the matching degree of each quadruped mobile robot by weighting the current remaining battery power, current task idle time and distance to the area where the abnormal event is located. The maintenance task is assigned to the wheeled mobile robot with the highest matching degree.

8. A heterogeneous robot collaborative operation system, characterized in that, include: Wheeled mobile robots are used to execute first-level inspection tasks and maintenance tasks to perform maintenance operations on malfunctioning equipment. A scheduling server is used to obtain the first-level anomaly data packet uploaded by the wheeled mobile robot when executing the first-level detection task instruction, the first-level anomaly data packet including the location of the event anomaly; and to determine the terrain category of the area where the anomaly event is located based on the location of the event anomaly in the first-level anomaly data packet. If the terrain category is non-flat terrain, a secondary detection task instruction is generated and sent to the quadrupedal mobile robot to trigger the quadrupedal mobile robot to detect the area where the abnormal event is located and obtain a secondary abnormal data packet; Based on the abnormal device in the secondary abnormal data packet, a maintenance task instruction is generated and sent to the wheeled mobile robot; A quadrupedal mobile robot is used to receive and execute secondary detection task instructions sent by the scheduling server.

9. The system according to claim 8, characterized in that, The wheeled mobile robot includes: A multi-wheeled mobility structure is a chassis structure consisting of multiple drive wheels, used for movement and steering; The navigation module is used for environmental awareness and generates first-level exception data packets; The vehicle control module is used to execute the instructions for the first-level detection task; The communication module is used to send first-level abnormal data packets to the scheduling server and to receive maintenance task instructions issued by the scheduling server. The maintenance module is used to perform maintenance operations on the malfunctioning equipment.

10. The system according to claim 8, characterized in that, The quadrupedal mobile robot includes: A quadrupedal mobility structure, comprising multiple supporting leg structures, for movement and steering; A sensor suite for environmental perception and generation of secondary anomaly data packets; The communication module is used to receive the secondary detection task instruction sent by the scheduling server; The vehicle control module executes the instructions for the secondary detection task.

Images