Robot inspection method and device, electronic device, and readable storage medium

Abstract

The present disclosure provides a robot inspection method and device, an electronic device, and a readable storage medium, which belongs to the field of robot technology. The method includes: in response to receiving a first inspection instruction, determining the robot's target inspection mode from multiple inspection modes based on the first inspection instruction, and determining the robot's target inspection scene from multiple inspection scenes based on the first inspection instruction; wherein the robot has different inspection parameters in multiple inspection modes, and the robot has different monitoring parameters in multiple inspection scenes; determining the inspection route of the target inspection scene, the inspection route includes multiple monitoring points; controlling the robot to monitor at each monitoring point of the inspection route according to the target inspection mode, so that the robot performs inspection in the target inspection scene. The robot inspection method and device, electronic device, and readable storage medium provided by the present disclosure can improve the stability and accuracy of robot inspection.

Description

Technical Field

[0001] The present disclosure belongs to the field of robot technology, and more specifically, to a robot inspection method and device, an electronic device, and a readable storage medium. Background Art

[0002] With the continuous development of science and technology, the demand for monitoring equipment and environment is increasing. Traditional manual inspection methods have gradually exposed many shortcomings. For example, manual inspection is inefficient and requires a lot of time to run between different inspection points. For large-scale areas, the inspection cycle is long and it is difficult to find problems in time. At the same time, the accuracy is greatly affected by human factors. Inspection personnel may miss or misdetect due to fatigue, lack of concentration, etc. Robots can be used to automatically complete inspection tasks for various environments and equipment, but their adaptability to complex environments needs to be improved, and there is a problem of low inspection efficiency. Summary of the invention

[0003] The purpose of the present disclosure is to provide a robot inspection method and device, an electronic device, and a readable storage medium to improve inspection efficiency.

[0004] A first aspect of the embodiments of the present disclosure provides a robot inspection method, comprising:

[0005] In response to receiving a first inspection instruction, determining a target inspection mode of the robot from a plurality of inspection modes based on the first inspection instruction, and determining a target inspection scene of the robot from a plurality of inspection scenes based on the first inspection instruction; wherein the inspection parameters of the robot in the plurality of inspection modes are different, and the monitoring parameters of the robot in the plurality of inspection scenes are different;

[0006] Determine an inspection route for the target inspection scene, wherein the inspection route includes a plurality of monitoring points;

[0007] The robot is controlled to monitor at each monitoring point of the inspection route according to the target inspection mode, so that the robot performs inspection in the target inspection scene.

[0008] A second aspect of the embodiments of the present disclosure provides a robot inspection device, comprising:

[0009] A patrol instruction module, configured to, in response to receiving a first patrol instruction, determine a target patrol mode of the robot from a plurality of patrol modes based on the first patrol instruction, and determine a target patrol scene of the robot from a plurality of patrol scenes based on the first patrol instruction; wherein the patrol parameters of the robot in the plurality of patrol modes are different, and the monitoring parameters of the robot in the plurality of patrol scenes are different;

[0010] An inspection route module, used to determine an inspection route of the target inspection scene, wherein the inspection route includes multiple monitoring points;

[0011] The inspection execution module is used to control the robot to monitor at each monitoring point of the inspection route according to the target inspection mode, so that the robot can perform inspection in the target inspection scene.

[0012] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of the above-mentioned robot inspection method when executing the computer program.

[0013] According to a fourth aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned robot inspection method are implemented.

[0014] The beneficial effects of the robot inspection method and device provided by the embodiments of the present disclosure are:

[0015] The disclosed embodiments improve the stability of robot inspection work and provide efficient and accurate solutions for various inspection needs. On the one hand, the inspection mode is dynamically selected according to specific needs, ensuring that the robot can perform tasks with the most appropriate parameters, thereby improving inspection efficiency and accuracy. On the other hand, monitoring parameters and inspection routes are customized for a variety of inspection scenarios, ensuring the comprehensiveness and pertinence of monitoring, and effectively avoiding waste of resources and omission of blind spots. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 A schematic diagram of a flow chart of a robot inspection method provided by an embodiment of the present disclosure;

[0018] Figure 2 A schematic diagram of a process flow of a robot heartbeat mechanism provided by an embodiment of the present disclosure;

[0019] Figure 3 A structural block diagram of a robot inspection device provided in one embodiment of the present disclosure;

[0020] Figure 4 A schematic block diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0021] In the following description, specific details such as specific system structures and technologies are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present disclosure. However, it should be clear to those skilled in the art that the present disclosure may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obstructing the description of the present disclosure with unnecessary details.

[0022] In order to make the purpose, technical solutions and advantages of the present disclosure more clear, specific embodiments will be described below in conjunction with the accompanying drawings.

[0023] Please refer to Figure 1 , Figure 1 A schematic diagram of a flow chart of a robot inspection method provided in an embodiment of the present disclosure, the method comprising:

[0024] S101: In response to receiving a first inspection instruction, determining a target inspection mode of the robot from multiple inspection modes based on the first inspection instruction, and determining a target inspection scene of the robot from multiple inspection scenes based on the first inspection instruction; wherein, the inspection parameters of the robot in the multiple inspection modes are different, and the monitoring parameters of the robot in the multiple inspection scenes are different.

[0025] In this embodiment, the first inspection instruction is a trigger instruction for starting the robot to perform inspection. The robot control platform sends the first inspection instruction, which may include task type, execution time and parameters.

[0026] Different inspection modes mean that the robot has different behaviors and parameter settings when executing the first inspection instruction. When the inspection mode is fast inspection, the moving speed is greater than the first speed and the monitoring accuracy is less than the first accuracy; when the inspection mode is fine inspection, the moving speed is less than the first speed and the monitoring accuracy is greater than the first accuracy. Among them, the first speed and the first accuracy are dynamically adjusted and set based on different inspection modes or different inspection scenarios.

[0027] For example, some inspection modes focus on rapid inspections, with faster movement speeds but relatively low monitoring accuracy; while other modes focus on detailed monitoring, with slower movement speeds but the ability to conduct more in-depth analysis of the target.

[0028] The robot can be equipped with temperature sensors, pressure sensors, humidity sensors, image sensors, infrared thermal imagers, sound sensors, communication modules, GPS modules, indoor positioning modules, microprocessors, memory, batteries, charging management modules, collision sensors, emergency brake modules, lighting modules and speakers. Different monitoring modules can be activated in different inspection scenarios, and the monitoring parameters of each monitoring module can be different. For example, the monitoring frequency and monitoring time can be adjusted according to actual needs.

[0029] Different inspection scenarios correspond to different monitoring requirements and parameters. For example, in the inspection scenario of a factory workshop, it is necessary to focus on monitoring parameters such as the operating temperature and noise of the equipment; while in the inspection scenario of a warehouse, more attention is paid to whether the goods are placed in a standardized manner and the environmental humidity.

[0030] The specific target inspection mode and target inspection scenario are determined according to the first inspection instruction to ensure that the robot can inspect in a manner that best suits the current task requirements.

[0031] For example, in factory inspections, robots can select different inspection modes and scenarios according to different production stages or equipment status. For example, when the equipment is just started, a more conservative inspection mode can be used to focus on monitoring the operating parameters of key equipment; during peak production periods, an efficient inspection mode can be switched to quickly inspect the production line to ensure smooth production.

[0032] The target inspection scene can be divided into a first area and a second area, wherein the inspection frequency of the first area is greater than the inspection frequency of the second area. The first area can be an area with a high degree of danger (such as flammable and explosive), and the second area can be an ordinary area.

[0033] For example, in warehouse inspections, the robot can determine the inspection mode and scenario based on different cargo types, storage areas, and other factors. For areas where flammable items are stored, more detailed inspections are required to monitor parameters such as temperature and humidity to prevent fires; while for areas with ordinary cargo, a more conventional inspection mode can be used.

[0034] The target inspection scenario can also be divided into the first weather type area and the second weather type area, wherein the monitoring accuracy of the first weather type area is greater than the monitoring accuracy of the second weather type area, the inspection speed of the first weather type area is less than the inspection speed of the second weather type area, the first weather type area may be rainy or snowy, and the second weather type area may be sunny.

[0035] For example, in the inspection of power facilities, the robot can select appropriate inspection modes and scenarios according to different weather conditions. In bad weather, it is necessary to strengthen the monitoring of key parts (such as cable lines, distribution boxes, etc.) and adopt more accurate inspection modes; in normal weather, large-scale rapid inspections can be carried out.

[0036] S102: Determine an inspection route of a target inspection scene, where the inspection route includes multiple monitoring points.

[0037] In this embodiment, the inspection route determines the movement trajectory of the robot in the target inspection scene. Different target inspection scenes have different characteristics and requirements. The multiple monitoring points on the inspection route are key locations for the robot to collect and monitor data.

[0038] Specifically, methods for determining inspection routes may include:

[0039] First, on-site investigation.

[0040] Conduct a detailed on-site survey of the target inspection scene to understand the layout of the site, equipment distribution, channel conditions, and existing obstacles, etc. This helps determine the robot's feasible walking path and potential inspection difficulties.

[0041] Second, equipment importance assessment.

[0042] Identify key equipment in the target inspection scenario, such as core power generation equipment, key substation equipment, etc. Plan inspection routes based on the importance of the equipment to ensure that key equipment can receive more frequent inspections.

[0043] Equipment that is prone to failure or has a greater impact on safety should be placed at a relatively early position in the inspection route so that problems can be discovered and measures can be taken in a timely manner.

[0044] Third, risk analysis.

[0045] Analyze the possible risk factors in the target inspection scene, such as fire hazard areas, parts prone to leakage, high temperature and high pressure areas, etc. Include these risk areas in the inspection route and increase the inspection frequency appropriately.

[0046] Fourth, adopt path planning algorithm.

[0047] You can select the A* algorithm to automatically generate the optimal inspection route based on the map information and constraints of the target inspection scene. The algorithm can take into account factors such as distance, time, and energy consumption to ensure that the inspection route is efficient and reasonable.

[0048] The generated inspection route is optimized and adjusted based on the robot's motion performance and sensor characteristics to ensure that the robot can successfully complete the inspection task.

[0049] The following aspects need to be considered when setting up monitoring points:

[0050] First, coverage of key locations: monitoring points should cover key locations and potential risk areas in the target inspection scenario;

[0051] Second, the need for data collection: determine the location and number of monitoring points based on different inspection tasks and monitoring parameters;

[0052] Third, operability and accessibility: The location of the monitoring point should be convenient for the robot to collect data and operate;

[0053] Fourth, dynamic adjustment: As the inspection task progresses and the environment changes, the location and number of monitoring points need to be dynamically adjusted.

[0054] Specifically, the monitoring point setting method and key location determination can be as follows:

[0055] First, set up monitoring points at key parts of key equipment, such as generator bearings, transformer windings, circuit breaker contacts, etc. Key parts are prone to overheating, abnormal vibration and other problems, and need to be monitored intensively.

[0056] For large equipment, multiple monitoring points can be set up at different heights and orientations to fully understand the operating status of the equipment.

[0057] Second, fault-prone areas. Identifying areas that have historically been prone to faults in target inspection scenarios, such as cable joints, insulator surfaces, and switch cabinet interiors, can strengthen monitoring of potential fault points.

[0058] Monitoring points should also be set up in areas that are often affected by external forces or disturbed by environmental factors, such as poles and towers close to roads and distribution rooms that are prone to flooding.

[0059] Third, safety-critical locations. Monitoring points are set up at locations where there may be safety hazards, such as storage areas for flammable and explosive items, around high-voltage equipment, near fire-fighting facilities, etc., so that environmental parameters can be monitored in real time to ensure safety.

[0060] For areas with frequent human activities, such as control rooms and distribution room entrances, setting up monitoring points can detect abnormal situations in a timely manner and ensure the safety of personnel.

[0061] S103: Control the robot to monitor at each monitoring point of the inspection route according to the target inspection mode, so that the robot performs inspection in the target inspection scene.

[0062] In this embodiment, different target inspection modes determine the robot's behavior and monitoring focus at the monitoring point. The inspection route includes both monitoring points and non-monitoring points, and the robot's operating mode is different for monitoring points and non-monitoring points. For example, the monitoring module at the monitoring point is larger than the monitoring module at the non-monitoring point. Therefore, the monitoring module started at the monitoring point is larger than the monitoring module started at the non-monitoring point, which improves the robot's endurance. In addition, when at the monitoring point, the robot's monitoring frequency is the first frequency; when at the non-monitoring point, the robot's monitoring frequency is the second frequency. Among them, the first frequency is greater than the second frequency, and the first frequency and the second frequency are dynamically adjusted.

[0063] Each monitoring point on the inspection route is a key location for the robot to monitor. The distribution and setting of each monitoring point directly affects the inspection effect. When controlling the robot to inspect, it is necessary to ensure that the robot can accurately reach each monitoring point and monitor in a predetermined order.

[0064] After arriving at the monitoring point, the robot needs to select appropriate sensors for monitoring according to the requirements of the target inspection mode, and process and analyze the collected data.

[0065] When the robot moves from one monitoring point to another, it needs to ensure safe and stable passage to avoid collisions and other unexpected situations. At the same time, during the movement, the robot can also perform some simple monitoring, such as environmental perception, obstacle monitoring, etc., so as to adjust its route in time.

[0066] From the above, it can be concluded that this embodiment improves the stability of robot inspection work and provides efficient and accurate solutions for various inspection needs. On the one hand, the inspection mode is dynamically selected according to specific needs, ensuring that the robot can perform tasks with the most appropriate parameters, thereby improving inspection efficiency and accuracy. On the other hand, monitoring parameters and inspection routes are customized for a variety of inspection scenarios, ensuring the comprehensiveness and pertinence of monitoring, and effectively avoiding waste of resources and omission of blind spots.

[0067] In one embodiment of the present disclosure, Figure 2 A schematic diagram of a robot heartbeat mechanism according to an embodiment of the present disclosure, see Figure 2 , the robot inspection method also includes:

[0068] Based on the database log table, record the data transmission status and generate an offline data table;

[0069] Backtrack or retransmit based on offline data table.

[0070] In this embodiment, by introducing a database log table, the data transmission status of each communication can be recorded, including successful and failed requests. In this way, even if there is a problem during the data transmission process, the robot can backtrack and retransmit through the database log table to ensure the integrity and consistency of the data.

[0071] During the task execution process, if the robot detects that the task cannot be completed, it will immediately report the exception code and reason. The background displays the task execution exception information on the front-end interface and updates the database log table record. The maintenance personnel will investigate and handle the exception information in the database log table to ensure the smooth completion of the inspection task.

[0072] In order to ensure that the connection status between the robot and the backend can be monitored in real time, a heartbeat mechanism based on the RPC protocol is designed and implemented. This mechanism allows the backend to quickly sense when the robot is offline. The specific steps are as follows:

[0073] First, define the heartbeat protocol.

[0074] Heartbeat request message format:

[0075] Robot unique identifier (String);

[0076] Timestamp (Long);

[0077] Request Code (String): A unique code used to identify a heartbeat request, for example: "HEARTBEAT_REQ_001".

[0078] Heartbeat response message format:

[0079] Response code (String): a unique code used to identify a heartbeat response, for example: "HEARTBEAT_RES_001";

[0080] Current server time (Long).

[0081] Second, implement heartbeat requests and responses.

[0082] The robot sends heartbeat requests to the backend at regular intervals, and the interval can be configured, for example, once every 30 seconds.

[0083] After receiving the heartbeat request, the background immediately sends a heartbeat response to confirm that the heartbeat request has been received.

[0084] If the background does not receive the robot's heartbeat request within the set time window (for example, not received within 1 minute), it is determined that the robot is in an offline state.

[0085] Third, deal with abnormal heartbeat.

[0086] If the background fails to receive the heartbeat request within the specified time, the exception handling logic is triggered.

[0087] Exception handling logic:

[0088] The current offline status of the robot is displayed on the front-end interface and logs are recorded.

[0089] The status of the heartbeat request is recorded in the database log table to facilitate subsequent analysis and processing.

[0090] Generate and send an exception code to the maintenance personnel. The exception code format is, for example, "ERROR_CODE_001".

[0091] Fourth, abnormal status is automatically reported.

[0092] When the robot detects an abnormality in itself (such as hardware failure or network interruption), it automatically reports the abnormality code to the background.

[0093] After the background receives the exception code, the robot failure information and specific reasons are displayed through the front-end interface.

[0094] The database log table records abnormal reporting information, which makes it easier for maintenance personnel to quickly locate and solve problems.

[0095] From the above, it can be concluded that the reliability and stability of robot inspections are significantly enhanced by introducing a database log table to record data transmission. The backtracking and retransmission mechanism effectively solves the problem of interruption or failure in data transmission, ensuring the integrity and accuracy of the data. At the same time, the robot reports the abnormal code and cause in a timely manner, providing maintenance personnel with a clear troubleshooting path, greatly shortening the troubleshooting time and ensuring the smooth execution and efficient completion of the inspection task.

[0096] In one embodiment of the present disclosure, the first inspection instruction includes an inspection level;

[0097] Determining a target inspection mode of the robot from a plurality of inspection modes based on the first inspection instruction includes:

[0098] If the inspection level is lower than the first level, the first inspection mode is used as the target inspection mode of the robot;

[0099] If the inspection level is greater than or equal to the first level and less than or equal to the second level, the second inspection mode is used as the target inspection mode of the robot;

[0100] If the inspection level is greater than the second level, the third inspection mode is used as the target inspection mode of the robot;

[0101] Among them, the first level is smaller than the second level, the accuracy of the inspection parameters in the first inspection mode is smaller than the accuracy of the inspection parameters in the second inspection mode, and the accuracy of the inspection parameters in the second inspection mode is smaller than the accuracy of the inspection parameters in the third inspection mode.

[0102] In this embodiment, the inspection level included in the first inspection instruction determines the target inspection mode adopted by the robot, and the most suitable inspection mode can be matched for the robot according to different inspection requirements to achieve efficient and accurate inspection operations.

[0103] Specifically, the method for determining the inspection level is as follows:

[0104] First, the scenarios are divided into different levels according to the degree of risk. For example, the probability of risk occurrence is less than the first percentage, which is the first level, and the probability of risk occurrence is greater than or equal to the first percentage, which is the second level. The first percentage can be adjusted dynamically and set according to the needs of the actual scenario; the storage point of safe items is the first level, and the storage point of flammable and explosive items is the second level.

[0105] Second, divide the fault data into different levels according to the history. For example, the fault frequency is less than the first threshold value, which is the first level, and the fault frequency is greater than or equal to the first threshold value, which is the second level. Among them, the first threshold can be adjusted dynamically and set according to the needs of the actual scenario. For example, analyze the historical fault data of the scenario, including fault frequency, fault severity, etc. For example, if a scenario has less than 5 faults in a month, the inspection level is the first level; if a scenario has more than or equal to 5 faults in a month, the inspection level is the second level. At the same time, a fault database can be established to determine the inspection level through data analysis.

[0106] Third, the inspection level is determined by using the similarity between the target scene and the standard scene.

[0107] Feature extraction: Detailed feature extraction of the target scene, including spatial layout, equipment type, environmental conditions, etc. At the same time, feature extraction is also performed on the built-in standard scene for comparison.

[0108] Similarity calculation: Use appropriate similarity calculation methods to compare the target scene with each standard scene. Algorithms such as cosine similarity and Euclidean distance can be used to measure the similarity between the target scene and the standard scene.

[0109] Inspection level determination: The inspection level is determined based on the similarity results. If the target scene is very similar to a standard scene, the inspection level of the standard scene can be used as a reference. As the similarity decreases, the inspection level can be adjusted appropriately.

[0110] For example, if the target scenario is highly similar to the standard scenario, but there are some special circumstances, such as older equipment, the inspection level can be slightly increased. For new scenarios that are not very similar to any standard scenario, the inspection level can be carefully determined by comprehensively considering factors such as the risk of the new scenario and the importance of the equipment, and continuously adjusted and optimized during the subsequent inspection process.

[0111] First, there are three different inspection modes, namely the first inspection mode, the second inspection mode and the third inspection mode, and the accuracy of their inspection parameters increases in sequence. The first inspection mode is suitable for situations where the inspection level is lower than the first level. At this time, the inspection accuracy requirement is lower than the first accuracy, and more attention is paid to the speed and coverage of the inspection. For example, in daily preliminary inspections, the first inspection mode can be used to quickly conduct a general inspection of each area to detect obvious abnormalities in a timely manner.

[0112] When the inspection level is between the first and second levels, the second inspection mode is selected. The inspection task at this time has certain requirements for accuracy, but it does not require extremely high accuracy like high-level inspections. For example, in regular and detailed inspections, the second inspection mode can more accurately monitor the operating status and potential problems of the equipment, providing a more reliable basis for maintenance and management.

[0113] When the inspection level is greater than the second level, the third inspection mode is used. In this mode, the inspection parameters are most accurate, and the target can be monitored and analyzed in great detail. Usually in scenarios such as in-depth inspections of key equipment and special inspections after major events, the third inspection mode can take advantage of its high accuracy to ensure that no minor fault hazards are missed.

[0114] From the above, we can conclude that by setting the inspection level and corresponding mode in a hierarchical manner, the robot's operating accuracy can be dynamically adjusted according to the actual inspection needs, which not only ensures the efficiency of daily inspections, but also enables detailed inspections in critical or high-risk scenarios. This not only improves inspection efficiency and safety, but also ensures the accuracy and timeliness of inspection results, providing strong technical support for the maintenance and management of various facilities.

[0115] In one embodiment of the present disclosure, the robot has multiple standard scenes built in, and the inspection method further includes:

[0116] The target scene is fused with multiple standard scenes to obtain multiple inspection scenes; among which the target scene is the actual scene collected by the robot.

[0117] In this embodiment, the robot has multiple built-in standard scenes, which provides an important reference framework for inspection. In the actual inspection process, it is first necessary to fuse the target scene (ie, the actual scene) collected by the robot with the built-in standard scene to obtain multiple inspection scenes.

[0118] Standard scenarios can be set up based on industry specifications and standards as well as common facilities for inspection scenarios.

[0119] Specifically, standard scenarios may include the following:

[0120] First, standard scenarios for the power industry.

[0121] Substation inspection scenarios: including transformer inspection, monitoring the transformer's oil temperature, oil level, winding temperature, partial discharge, etc.; switchgear inspection, checking the position status, contact temperature, operating mechanism status, etc. of circuit breakers, disconnectors, and earthing switches; busbar inspection, observing the busbar's appearance, temperature, insulation condition, etc.

[0122] Distribution room inspection scenario: Inspect distribution cabinets and distribution boxes to check the operating status of electrical components, instrument readings, wiring conditions, etc.; monitor indoor environmental parameters such as temperature, humidity, and smoke to ensure that the equipment operates in a good environment.

[0123] Second, standard scenarios for the petrochemical industry.

[0124] Refinery inspection scenario: inspect refinery equipment, pipelines, valves, etc., monitor equipment parameters such as temperature, pressure, flow, etc.; check pipeline leakage and valve opening and closing status; monitor gas concentration in flammable and explosive areas to ensure safe production.

[0125] Chemical plant inspection scenario: inspect chemical reactors, storage tanks, pipelines, etc., observe the appearance, corrosion, sealing performance, etc. of the equipment; monitor the temperature, pressure, liquid level and other parameters of key parts; monitor the emission of harmful gases to prevent environmental pollution.

[0126] The target scene is the scene captured by the robot in the actual working environment, which can contain multiple images actually captured from different angles. Through the image collection from these different angles, the situation of the target scene can be more comprehensively understood. For example, during the inspection of the factory workshop, the robot can shoot the equipment from different positions and angles to obtain image information from all sides of the equipment. The process of scene fusion is to match and integrate the features in the target scene with the built-in standard scene.

[0127] Specifically, the scene fusion method steps can be as follows:

[0128] First, data collection and analysis.

[0129] Comprehensive data collection of the target scene can include spatial layout, equipment distribution, and environmental parameters. Various sensors built into the robot can be used to obtain multi-dimensional information of the target scene.

[0130] Analyze the collected data and extract the characteristics of the target scene. For example, analyze the appearance characteristics and position relationship of the equipment through image recognition technology; analyze the environmental characteristics through temperature and humidity sensor data, etc.

[0131] Second, feature matching.

[0132] Compare and match the characteristics of the target scene with the built-in standard scenes. Pattern recognition algorithms, feature vector matching and other methods can be used to find the similarities and differences between the target scene and each standard scene.

[0133] For similar parts, the inspection strategies and parameter settings of standard scenarios can be directly referenced; for different parts, further analysis and processing are required.

[0134] Based on the results of feature matching, the standard scenes are integrated and adjusted to meet the needs of the target scenes. For example, if there are specific devices or areas in the target scene, it is necessary to add corresponding monitoring points to the inspection route; if the environmental conditions are different from the standard scene, the sensor parameter settings need to be adjusted.

[0135] Third, dynamic adjustment.

[0136] During the inspection process, the changes of the target scene are continuously monitored. If the target scene changes, such as equipment update, layout adjustment, environmental change, etc., dynamic adjustment of scene fusion is carried out in a timely manner.

[0137] By using machine learning algorithms, the robot can automatically adjust its inspection strategies based on historical data and real-time feedback, thereby improving the adaptability and intelligence of scene integration.

[0138] For example, similarities between the target scene and the standard scene are found through image feature point extraction and matching, and then the two are fused. The fused inspection scene contains both typical features and specification information in the standard scene and the specific conditions of the target scene actually collected.

[0139] From the above, we can conclude that, on the one hand, standard scenes provide a standardized reference for inspections, allowing the robot to quickly identify common inspection objects and problem areas. On the other hand, by integrating with the actual collected target scenes, it can better adapt to different actual working environments and improve the accuracy and pertinence of inspections. Therefore, integrating the target scene with multiple standard scenes can make full use of the robot's built-in standard information and the actual collected image data to provide richer and more accurate scene information for inspections, thereby improving the efficiency and quality of inspections.

[0140] In one embodiment of the present disclosure, the target scene is fused with multiple standard scenes to obtain multiple inspection scenes, including:

[0141] Determining a first feature set, the first feature set including scene features of a plurality of target scenes;

[0142] determining a second feature set, the second feature set including scene features of a plurality of standard scenes;

[0143] The first feature set and the second feature set are subjected to feature fusion to obtain a third feature set;

[0144] The third feature set is input into a scene generation model to obtain a plurality of inspection scenes, wherein the scene generation model is used to generate inspection scenes according to the plurality of scene features.

[0145] In this embodiment, first, a first feature set is determined, which includes scene features of multiple target scenes. The target scene is the scene actually collected by the robot, and these scene features may include but are not limited to the shape, color, texture, spatial layout, etc. of the objects in the target scene. At the same time, a second feature set is determined, which includes scene features of multiple standard scenes. The standard scene is a typical and normative scene built into the robot, and its scene features may also include but are not limited to various attributes and spatial relationships of objects in the standard scene.

[0146] Exemplarily, a specific method for extracting scene features may be as follows:

[0147] First, preprocess the target scene and standard scene, including image denoising, normalization and other operations.

[0148] Second, perform feature extraction. Use deep learning algorithms, such as convolutional neural networks, to extract features from the target scene and the standard scene, respectively, to obtain the first feature set and the second feature set. You can use a pre-trained convolutional neural network model for feature extraction to speed up training and improve feature quality. Analyze and screen the extracted features, remove redundant and irrelevant features, and retain features that are important to the inspection scene.

[0149] Then, the first feature set and the second feature set are subjected to feature fusion to obtain a third feature set. The feature fusion can be performed using a neural network model, and the specific steps are as follows:

[0150] First, model selection and construction.

[0151] Select a neural network model suitable for feature fusion tasks, such as a Transformer architecture model, which has strong representation and fusion capabilities and can effectively process multimodal data.

[0152] Adjust and optimize the selected neural network model according to the task requirements and data characteristics. You can adjust the model's parameters such as the number of layers, number of neurons, activation function, etc. to improve the model's performance and adaptability.

[0153] Design a suitable fusion layer in the model to fuse the first feature set with the second feature set. The fusion layer can be implemented in many ways, such as concatenation, attention mechanism, etc.

[0154] Second, the training process.

[0155] The prepared historical first feature set and historical second feature set are used as input data and input into the constructed neural network model for training, and the output is the historical third feature set.

[0156] Define a suitable loss function, such as mean square error, to measure the gap between the third feature set output by the neural network model and the true value.

[0157] Select a suitable optimization algorithm, such as stochastic gradient descent, to optimize and update the parameters of the neural network model. By continuously adjusting the parameters of the neural network model, the loss function is minimized, thereby improving the fusion effect of the neural network model.

[0158] Set reasonable training parameters, such as learning rate, batch size, number of training rounds, etc. The learning rate controls the update speed of the neural network model parameters, the batch size determines the amount of data used for each training, and the number of training rounds affects the degree of training of the neural network model.

[0159] Third, neural network model evaluation and optimization.

[0160] During the training process, the neural network model is evaluated regularly, and the validation set data is used to test the performance and fusion effect of the neural network model. The quality of the neural network model can be measured by calculating some evaluation indicators, such as accuracy, recall rate, F1 value, etc.

[0161] According to the evaluation results, the neural network model is optimized and adjusted. The structure, parameter settings, training strategies, etc. of the neural network model can be adjusted to improve the performance and fusion effect of the neural network model.

[0162] The training and evaluation process is repeated until the neural network model achieves satisfactory performance.

[0163] Fourth, application and deployment.

[0164] After the neural network model is trained, it is applied to the actual inspection scene. The feature sets of the new target scene and the standard scene are input into the neural network model to obtain the fused third feature set.

[0165] Based on the fused feature set, generate inspection scenarios or perform other related task processing.

[0166] Deploy the trained neural network model to the actual robot system to achieve real-time feature fusion and inspection scene generation. You can choose to deploy it in the cloud or locally according to your actual needs.

[0167] Finally, the third feature set is input into the scenario generation model to obtain multiple inspection scenarios. The scenario generation model is a tool that can generate specific inspection scenarios based on the input scene features. The AI ​​big model can be used to transform the fused feature set into a visual inspection scenario. The scenario generation model analyzes and processes the input third feature set, extracts key information, and generates inspection scenarios based on preset rules and algorithms.

[0168] Exemplarily, the scene generation model can be obtained by training an AI big model based on a training set, where the training set includes scene features and corresponding inspection scenes. The specific steps are as follows:

[0169] First, model selection and architecture design.

[0170] Choose a suitable AI large model and consider deep learning models with strong generation and representation capabilities, such as the generative adversarial network model, which has outstanding performance in image generation, scene understanding, etc.

[0171] According to the task requirements of the inspection scene generation, the selected AI big model is appropriately adjusted and optimized. The layer structure, number of neurons, activation function, etc. of the AI ​​big model can be increased or adjusted to better adapt to the tasks of the inspection scene generation. At the same time, appropriate input and output interfaces are designed to receive scene features and generate inspection scenes.

[0172] Second, AI large model training.

[0173] The sorted historical third feature set is divided into training set, validation set and test set. Usually it can be divided according to a certain ratio, for example, 70% for training, 20% for validation, and 10% for testing. Among them, the historical third feature set is used as the input of the AI ​​big model, and the historical inspection scene is used as the output of the AI ​​big model.

[0174] According to the goal of inspection scene generation, a suitable loss function is selected. For example, the structural similarity index can be used to measure the gap between the generated inspection scene and the real inspection scene.

[0175] Select an efficient optimization algorithm, such as RMSprop, to update the parameters of the AI ​​big model. The choice of optimization algorithm should be determined based on the characteristics of the AI ​​big model and the scale of the training data.

[0176] During the training process, the loss function is gradually reduced by continuously adjusting the parameters of the AI ​​big model, thereby improving the accuracy of the inspection scene generated by the AI ​​big model. The AI ​​big model is regularly evaluated using the validation set to monitor the performance and generalization ability of the AI ​​big model. The parameters and training strategy of the AI ​​big model are adjusted based on the evaluation results.

[0177] Third, evaluation and optimization of large AI models.

[0178] Choose appropriate evaluation indicators to measure the performance of the AI ​​model. In addition to the loss function value, other indicators such as peak signal-to-noise ratio (PSNR) and image quality index (IQI) can also be used to evaluate the quality of the generated inspection scene from different angles.

[0179] According to the evaluation results, the AI ​​big model can be further optimized. The structure of the AI ​​big model can be adjusted, data enhancement technology can be added, and hyperparameters can be adjusted. At the same time, different training strategies can be tried, such as increasing the number of training rounds and adjusting the learning rate decay strategy, to improve the performance and stability of the AI ​​big model.

[0180] Fourth, generate inspection scenarios.

[0181] When the AI ​​big model is trained and fully evaluated, the third feature set is input into the scene generation AI big model. The third feature set is obtained by fusion of the target scene and the standard scene, and contains rich scene information.

[0182] The AI ​​big model generates multiple corresponding inspection scenarios based on the input third feature set. These inspection scenarios can provide more comprehensive and accurate references for robot inspections, helping robots better understand and adapt to different inspection environments.

[0183] Fifth, practical application and continuous improvement.

[0184] The generated inspection scenarios are applied to robot inspections to observe their effects in actual applications. Based on the feedback from actual applications, the AI ​​big model is further adjusted and optimized to improve its performance and practicality in actual scenarios.

[0185] As new data is continuously collected and technology continues to develop, the scene generation AI big model is continuously improved and updated. The AI ​​big model can be retrained regularly to add new data and features to improve the adaptability and accuracy of the AI ​​big model.

[0186] From the above, it can be concluded that through the fine matching and fusion of the first feature set and the second feature set, it is ensured that the inspection scene can accurately reflect the actual environment and potential risks, providing a more realistic and comprehensive simulation environment for subsequent inspection work.

[0187] In one embodiment of the present disclosure, the first feature set and the second feature set are subjected to feature fusion to obtain a third feature set, including:

[0188] Determining a first weight of a first feature, where the first feature is a scene feature in the first feature set;

[0189] determining a second weight of a second feature, where the second feature is a scene feature in the second feature set;

[0190] weighting the first weight based on the first weight to obtain a first weighted feature;

[0191] weighting the second weight based on the second weight to obtain a second weighted feature;

[0192] generating a third feature based on the first weighted feature and the second weighted feature;

[0193] Among them, the third feature is a scene feature in the third feature set, and the first weight is greater than the second weight.

[0194] In this embodiment, the first weight may be determined based on the proportion of the first feature in all scene features, where the proportion refers to the importance to the scene. The second weight may be determined based on the proportion of the second feature in all scene features.

[0195] The calculation formula of the first weight is as follows:

[0196]

[0197] in, represents the first weight, Indicates the importance ratio of the first feature in all scene features. Represents the impact factor.

[0198] The calculation formula of the second weight is as follows:

[0199]

[0200] in, represents the second weight, Indicates the importance ratio of the second feature in all scene features. Represents the impact factor.

[0201] The calculation formula of impact factor is as follows:

[0202]

[0203] The parameter values ​​can be temperature, humidity, pressure, etc., which will change dynamically according to the needs of the target scene. For example, in a warehouse, the stored materials are not easy to get wet, and humidity is an influencing factor.

[0204] The first feature represents a specific attribute of the target scene. The first weight is greater than the second weight, which means that the first feature is more important than the second feature in the fusion process.

[0205] The second feature comes from multiple standard scenarios and represents the normative and typical features under normal circumstances. The second weight is less than the first weight, indicating that the second feature is less important than the first feature in the fusion process. This is because although the standard scenario has a certain reference value, in actual inspections, the uniqueness of the target scenario may be more critical. However, the features of the standard scenario still have a certain auxiliary role in inspections.

[0206] The first feature is weighted based on the first weight to obtain the first weighted feature; the second feature is weighted based on the second weight to obtain the second weighted feature. The weighting operation makes the important features more prominent in the fused feature set, while the less important features are relatively weakened.

[0207] By combining the first weighted feature and the second weighted feature, the scene feature in the third feature set can be generated. The generated third feature set contains both the actual features of the target scene and the reference features of the standard scene, providing a more comprehensive information basis for subsequent inspection scene generation.

[0208] Simply put, the third feature is obtained by weighting all the first features in the first feature set and all the second features in the second feature set, and all the third features form the third feature set.

[0209] First, the first feature set and the second feature set are prepared.

[0210] The first feature set usually contains the actual features of the target scene, which are obtained by direct observation and data collection of the target scene. The second feature set contains the reference features of the standard scene, which are extracted based on the built-in standard scene. The standard scene is usually a representative scene determined through a lot of practice and optimization, and its features can provide a reference for the new target scene.

[0211] Then, the third feature is generated by weighted calculation.

[0212] Determine weights: In order to weight the features in the first feature set and the second feature set, it is necessary to first determine the weight of each feature. The weight can be determined based on a variety of factors, such as the importance, reliability, and relevance of the feature to the inspection target. For important features that are highly reliable and closely related to the inspection target, a higher weight can be assigned; while for relatively unimportant or less reliable features, a lower weight can be assigned.

[0213] Weighted calculation: Multiply each first feature in the first feature set and each second feature in the second feature set by their corresponding weights. Then add the weighted first and second features to obtain the third feature.

[0214] For example, assuming that the temperature feature of the target scene has a value of T1 in the first feature set, and the corresponding weight is W1; the reference temperature feature of the standard scene has a value of T2 in the second feature set, and the corresponding weight is W2. Then the weighted temperature feature is W1×T1+W2×T2, and this result is the temperature feature value in the generated third feature.

[0215] Finally, the third feature set is formed.

[0216] All third features obtained through weighted calculation are combined to form a third feature set.

[0217] From the above, it can be concluded that by assigning different weights to the scene features in the first feature set and the second feature set, and performing weighted fusion accordingly, the generated third feature set can more accurately reflect the fusion effect of the target scene and the standard scene. In particular, by setting the first weight greater than the second weight, the importance of the target scene features is highlighted, making the generated inspection scene closer to actual needs. Therefore, not only the accuracy and pertinence of scene fusion are improved, but also the realism and practicality of the inspection scene are enhanced.

[0218] In one embodiment of the present disclosure, the robot inspection method further includes:

[0219] The robot is controlled to collect images in the target inspection scene to obtain the target image.

[0220] In this embodiment, the current brightness value is determined based on the robot;

[0221] Determine a target image enhancement strategy based on the current brightness value;

[0222] The robot enhances the first image based on the target image enhancement strategy;

[0223] Among them, the robot has built-in environmental sensors.

[0224] During the robot inspection process, controlling the robot to collect images in the target inspection scene is an important step in obtaining key information. The target inspection scene can be a variety of different actual environments, such as factory workshops, warehouses, public facilities, etc. Through image acquisition, the robot can obtain visual information of the target inspection scene, providing a basis for subsequent analysis and judgment.

[0225] The robot's built-in environmental sensors can monitor various parameters of the current environment, including brightness values, and can perceive the lighting conditions of the surrounding environment in real time, providing basic data for determining target image enhancement strategies.

[0226] The current brightness value reflects the light intensity of the target inspection scene. Different brightness values ​​will affect the quality of the collected image. For example, in a low-brightness environment, the image will appear darker and the details will not be clear; while in a high-brightness environment, the image will be overexposed. Therefore, determining the current brightness value is crucial to choosing an appropriate image enhancement strategy.

[0227] Based on the current brightness value, the robot can determine the corresponding target image enhancement strategy. For example, if the current brightness value is lower than the standard brightness value, strategies such as brightness enhancement and contrast enhancement are needed to highlight the details in the image. If the current brightness value is higher than the standard brightness value, exposure adjustment and color balance are needed to avoid overexposure and color distortion.

[0228] In this embodiment, the target image enhancement strategy is determined based on the current brightness value, including:

[0229] If the current brightness value is less than the first brightness value, the first image enhancement strategy is used as the target image enhancement strategy;

[0230] If the current brightness value is greater than or equal to the first brightness value, the second image enhancement strategy is used as the target image enhancement strategy;

[0231] Among them, the accuracy of the first image enhancement strategy is greater than that of the second image enhancement strategy.

[0232] In the process of determining the target image enhancement strategy, the current brightness value plays a key role in the judgment basis. By comparing the current brightness value with the specific first brightness value, it can be determined which image enhancement strategy to adopt.

[0233] When the current brightness value is less than the first brightness value, the first image enhancement strategy is selected. At this time, the current environment brightness is less than the normal brightness, which will cause the captured image to be darker and the details to be unclear. At this time, the first image enhancement strategy with higher precision can better improve the image quality. The first image enhancement strategy can include a variety of processing methods, such as a significant brightness increase, contrast enhancement, noise reduction processing, etc. Through these operations, the dark details in the image can be made more obvious, improving the readability and analyzability of the image.

[0234] When the current brightness value is greater than or equal to the first brightness value, the second image enhancement strategy is selected. At this time, the current environment brightness is greater than the normal brightness, and there is a risk of overexposure or color distortion. The second image enhancement strategy aims to make appropriate adjustments to images in high-brightness environments to maintain the naturalness and authenticity of the image. For example, it may include moderate exposure adjustment, color balance, highlight suppression and other processing. In a high-brightness environment, the image quality issues are relatively few, and overly complex and high-precision processing is not required. Therefore, the accuracy of the second image enhancement strategy is relatively low to improve processing efficiency while avoiding image distortion caused by excessive processing.

[0235] From the above, it can be concluded that by intelligently identifying the brightness value of the current inspection scene and dynamically adjusting the image enhancement strategy accordingly, a significant improvement in image quality is achieved.

[0236] Corresponding to the robot inspection method of the above embodiment, Figure 3 This is a structural block diagram of a robot inspection device provided by an embodiment of the present disclosure. For ease of explanation, only the parts related to the embodiment of the present disclosure are shown. Figure 3 The robot inspection device 20 includes: an inspection instruction module 21, an inspection route module 22 and an inspection execution module 23.

[0237] The inspection instruction module 21 is used to determine the target inspection mode of the robot from multiple inspection modes based on the first inspection instruction in response to receiving the first inspection instruction, and determine the target inspection scene of the robot from multiple inspection scenes based on the first inspection instruction; wherein the inspection parameters of the robot in the multiple inspection modes are different, and the monitoring parameters of the robot in the multiple inspection scenes are different;

[0238] Inspection route module 22, used to determine the inspection route of the target inspection scene, the inspection route includes multiple monitoring points;

[0239] The inspection execution module 23 is used to control the robot to monitor at each monitoring point of the inspection route according to the target inspection mode, so that the robot can perform inspection in the target inspection scene.

[0240] In one embodiment of the present disclosure, the robot inspection device 20 further includes: a data recording module;

[0241] A data recording module is used to record data transmission status based on a database log table and generate an offline data table;

[0242] Backtrack or retransmit based on offline data table.

[0243] In one embodiment of the present disclosure, the inspection instruction module 21 is specifically used to:

[0244] The first inspection instruction includes the inspection level;

[0245] Determining a target inspection mode of the robot from a plurality of inspection modes based on the first inspection instruction includes:

[0246] If the inspection level is lower than the first level, the first inspection mode is used as the target inspection mode of the robot;

[0247] If the inspection level is greater than or equal to the first level and less than or equal to the second level, the second inspection mode is used as the target inspection mode of the robot;

[0248] If the inspection level is greater than the second level, the third inspection mode is used as the target inspection mode of the robot;

[0249] Among them, the first level is smaller than the second level, the accuracy of the inspection parameters in the first inspection mode is smaller than the accuracy of the inspection parameters in the second inspection mode, and the accuracy of the inspection parameters in the second inspection mode is smaller than the accuracy of the inspection parameters in the third inspection mode.

[0250] In one embodiment of the present disclosure, the robot inspection device 20 further includes: an inspection scene module;

[0251] Inspection scene module, which is used in robots with multiple built-in standard scenes, and the inspection methods also include:

[0252] The target scene is fused with multiple standard scenes to obtain multiple inspection scenes; among which the target scene is the actual scene collected by the robot.

[0253] In one embodiment of the present disclosure, the inspection scene module is specifically used to:

[0254] Determining a first feature set, the first feature set including scene features of a plurality of target scenes;

[0255] determining a second feature set, the second feature set including scene features of a plurality of standard scenes;

[0256] The first feature set and the second feature set are subjected to feature fusion to obtain a third feature set;

[0257] The third feature set is input into a scene generation model to obtain a plurality of inspection scenes, wherein the scene generation model is used to generate inspection scenes according to the plurality of scene features.

[0258] In one embodiment of the present disclosure, the inspection scene module is also used to:

[0259] Determining a first weight of a first feature, where the first feature is a scene feature in the first feature set;

[0260] determining a second weight of a second feature, where the second feature is a scene feature in the second feature set;

[0261] weighting the first weight based on the first weight to obtain a first weighted feature;

[0262] weighting the second weight based on the second weight to obtain a second weighted feature;

[0263] generating a third feature based on the first weighted feature and the second weighted feature;

[0264] Among them, the third feature is a scene feature in the third feature set, and the first weight is greater than the second weight.

[0265] In one embodiment of the present disclosure, the robot inspection device 20 further includes: an image acquisition module;

[0266] The image acquisition module is used to control the robot to acquire images in the target inspection scene to obtain the target image.

[0267] See also Figure 4 , Figure 4 A schematic block diagram of an electronic device provided by an embodiment of the present disclosure. Figure 4 The electronic device 300 in the embodiment shown may include: one or more processors 301, one or more input devices 302, one or more output devices 303 and one or more memories 304. The processors 301, input devices 302, output devices 303 and memories 304 communicate with each other via a communication bus 305. The memory 304 is used to store computer programs, which include program instructions. The processor 301 is used to execute the program instructions stored in the memory 304. The processor 301 is configured to call the program instructions to execute the functions of each module / unit in the above-mentioned device embodiments, such as Figure 3 The functions of modules 21 to 23 are shown.

[0268] It should be understood that in the embodiment of the present disclosure, the processor 301 may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0269] The input device 302 may include a touch panel, a fingerprint collection sensor (for collecting the user's fingerprint information and fingerprint direction information), a microphone, etc., and the output device 303 may include a display (LCD, etc.), a speaker, etc.

[0270] The memory 304 may include a read-only memory and a random access memory, and provide instructions and data to the processor 301. A portion of the memory 304 may also include a non-volatile random access memory. For example, the memory 304 may also store information about the device type.

[0271] In a specific implementation, the processor 301, input device 302, and output device 303 described in the embodiments of the present disclosure can execute the implementation methods described in the first and second embodiments of the robot inspection method provided in the embodiments of the present disclosure, and can also execute the implementation methods of the electronic device described in the embodiments of the present disclosure, which will not be repeated here.

[0272] In another embodiment of the present disclosure, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by the processor, all or part of the processes in the above-mentioned embodiment method are implemented, and the computer program can also be completed by instructing the relevant hardware through the computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, the steps of each of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0273] The computer-readable storage medium may be an internal storage unit of the electronic device of any of the aforementioned embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Furthermore, the computer-readable storage medium may also include both an internal storage unit of the electronic device and an external storage device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0274] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this disclosure.

[0275] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the electronic devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0276] In the several embodiments provided in the present application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces or units, or it can be an electrical, mechanical or other form of connection.

[0277] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present disclosure.

[0278] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0279] The above are only specific embodiments of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present disclosure, and these modifications or replacements should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A robot inspection method, characterized in that: include: In response to receiving a first inspection instruction, determining a target inspection mode of the robot from a plurality of inspection modes based on the first inspection instruction, and determining a target inspection scene of the robot from a plurality of inspection scenes based on the first inspection instruction; wherein the inspection parameters of the robot in the plurality of inspection modes are different, and the monitoring parameters of the robot in the plurality of inspection scenes are different; Determine an inspection route for the target inspection scene, wherein the inspection route includes a plurality of monitoring points; Controlling the robot to monitor at each monitoring point of the inspection route according to the target inspection mode, so that the robot performs inspection in the target inspection scene; When the inspection mode is fast inspection, the moving speed is greater than the first speed, and the monitoring accuracy is less than the first accuracy; when the inspection mode is fine inspection, the moving speed is less than the first speed, and the monitoring accuracy is greater than the first accuracy; The target inspection scene is divided into a first area and a second area, wherein the inspection number of the first area is greater than the inspection number of the second area; The target inspection scene is further divided into a first weather type area and a second weather type area, wherein the monitoring accuracy of the first weather type area is greater than the monitoring accuracy of the second weather type area, and the inspection speed of the first weather type area is less than the inspection speed of the second weather type area; The monitoring module activated at the monitoring point is larger than the monitoring module activated at the non-monitoring point; When at the monitoring point, the monitoring frequency of the robot is a first frequency; When at the non-monitoring point, the monitoring frequency of the robot is a second frequency; wherein the first frequency is greater than the second frequency; The robot has multiple standard scenes built in, and the inspection method also includes: The target scene is respectively integrated with the multiple standard scenes to obtain the multiple inspection scenes; wherein the target scene is the actual scene collected by the robot; The step of fusing the target scene with the multiple standard scenes to obtain the multiple inspection scenes includes: Determining a first feature set, the first feature set comprising scene features of a plurality of target scenes; determining a second feature set, wherein the second feature set includes scene features of a plurality of standard scenes; Fusing the first feature set and the second feature set to obtain a third feature set; Inputting the third feature set into a scene generation model to obtain the multiple inspection scenes, wherein the scene generation model is used to generate the inspection scenes according to the multiple scene features; The step of fusing the first feature set and the second feature set to obtain a third feature set includes: determining a first weight of a first feature, where the first feature is a scene feature in the first feature set; determining a second weight of a second feature, where the second feature is a scene feature in the second feature set; weighting the first weight based on the first weight to obtain a first weighted feature; weighting the second weight based on the second weight to obtain a second weighted feature; generating a third feature based on the first weighted feature and the second weighted feature; The third feature is a scene feature in the third feature set, and the first weight is greater than the second weight; The calculation formula of the first weight is as follows: in, represents the first weight, Indicates the importance ratio of the first feature in all scene features. represents the impact factor; The calculation formula of the second weight is as follows: in, represents the second weight, Indicates the importance ratio of the second feature in all scene features. Represents the impact factor.

2. The robot inspection method according to claim 1, characterized in that: Also includes: Based on the database log table, record the data transmission status and generate an offline data table; Backtracking or retransmission is performed based on the offline data table.

3. The robot inspection method according to claim 1, characterized in that: The first inspection instruction includes an inspection level; The determining a target inspection mode of the robot from a plurality of inspection modes based on the first inspection instruction comprises: If the inspection level is lower than the first level, the first inspection mode is used as the target inspection mode of the robot; If the inspection level is greater than or equal to the first level and less than or equal to the second level, the second inspection mode is used as the target inspection mode of the robot; If the inspection level is greater than the second level, the third inspection mode is used as the target inspection mode of the robot; Among them, the first level is lower than the second level, the accuracy of the inspection parameters in the first inspection mode is lower than the accuracy of the inspection parameters in the second inspection mode, and the accuracy of the inspection parameters in the second inspection mode is lower than the accuracy of the inspection parameters in the third inspection mode.

4. The robot inspection method according to claim 2, characterized in that: Also includes: The robot is controlled to collect images in the target inspection scene to obtain a target image.

5. A robot inspection device, characterized in that: A method for implementing the robot inspection method according to any one of claims 1 to 4, comprising: A patrol instruction module, configured to, in response to receiving a first patrol instruction, determine a target patrol mode of the robot from a plurality of patrol modes based on the first patrol instruction, and determine a target patrol scene of the robot from a plurality of patrol scenes based on the first patrol instruction; wherein the patrol parameters of the robot in the plurality of patrol modes are different, and the monitoring parameters of the robot in the plurality of patrol scenes are different; An inspection route module, used to determine an inspection route of the target inspection scene, wherein the inspection route includes multiple monitoring points; The inspection execution module is used to control the robot to monitor at each monitoring point of the inspection route according to the target inspection mode, so that the robot can perform inspection in the target inspection scene.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images