Inspection method, equipment and inspection system based on AI intelligent agent

Abstract

The invention relates to the technical field of inspection, in particular to an AI intelligent agent-based inspection method, equipment and system, and the method comprises the steps: obtaining a target region image of a to-be-inspected region, determining an abnormal region according to the target region image, generating a bidding request under the condition that the environment illumination condition of the abnormal region does not meet a preset condition, the bidding request comprises illumination demand information of an abnormal area and an area position of the abnormal area, sending the bidding request to a plurality of second devices, when bidding information generated by the second devices according to the bidding request is received, determining a target device from the plurality of second devices according to the bidding information, and sending a task execution instruction to the target device, and the target device provides illumination for the abnormal area according to the area position and the illumination demand information. By transferring the illumination energy consumption of the first equipment, the endurance time of the first equipment can be prolonged, the first equipment is prevented from being exposed, and the illumination requirement can be met to the maximum extent.

Description

Technical Field

[0001] The present invention relates to the field of inspection technology, and in particular to an inspection method, equipment and inspection system based on an AI intelligent agent. Background Art

[0002] Drone inspections involve the automated inspection and monitoring of specific areas or facilities using drone cameras. This technology has been widely used in recent years across a variety of sectors, particularly in the power, oil and gas, agriculture, environmental protection, and public safety industries, to improve efficiency, reduce costs, and reduce reliance on manual operations.

[0003] During nighttime inspections, the drone's camera may not produce ideal images due to insufficient light. If the drone is equipped with a light source module, it usually needs to be kept on to solve the problem of insufficient light, which will seriously affect the drone's flight time. Summary of the Invention

[0004] One purpose of the present invention is to provide an inspection method, equipment and inspection system based on AI intelligent agent to solve the technical problem of insufficient light during existing nighttime inspections.

[0005] In a first aspect, an embodiment of the present invention provides an AI agent-based inspection method, applied to a first device, comprising:

[0006] Acquire a target area image of the area to be inspected;

[0007] determining an abnormal area according to the target area image;

[0008] When the ambient lighting condition of the abnormal area does not meet the preset condition, generating a bidding request, the bidding request including the lighting demand information of the abnormal area and the area location of the abnormal area;

[0009] Sending the bidding request to multiple second devices, so that the multiple second devices generate bidding information according to the bidding request and send the bidding information to the first device;

[0010] Upon receiving the bidding information sent by the second device, determining a target device from the plurality of second devices according to the bidding information;

[0011] A task execution instruction is sent to the target device, so that when the target device receives the task execution instruction, it provides lighting for the abnormal area according to the area location and the lighting requirement information.

[0012] Optionally, the bidding information includes at least one of energy consumption, confidence value, light intensity, coverage and device location.

[0013] Optionally, the bidding information includes at least energy consumption;

[0014] Determining a target device from a plurality of second devices according to the bidding information includes:

[0015] A target device is selected from the plurality of second devices according to the energy consumption and the ambient light condition.

[0016] Optionally, the AI ​​agent-based inspection method further includes:

[0017] determining the abnormality type according to the target area image;

[0018] The bidding information at least includes a confidence value;

[0019] Determining a target device from a plurality of second devices according to the bidding information includes:

[0020] A target device is screened from the plurality of second devices according to the confidence value and the abnormality type.

[0021] Optionally, the AI ​​agent-based inspection method further includes:

[0022] determining an abnormality type according to the target area image, each abnormality type being configured with a weight coefficient corresponding to the bidding information;

[0023] Determining a target device from a plurality of second devices according to the bidding information includes:

[0024] Determining a normalized parameter value corresponding to the bidding information according to the bidding information;

[0025] Calculating a weighted sum based on the normalized parameter value corresponding to the bidding information and the weight coefficient corresponding to the bidding information;

[0026] A target device is screened from the plurality of second devices according to the weighted sum.

[0027] Optionally, the AI ​​agent-based inspection method further includes:

[0028] Acquiring a first regional image of the abnormal region;

[0029] determining first feature information of the abnormal area based on the first area image, where the first feature information includes at least one of shadow distribution information, geometric shape features, and reflective area information;

[0030] The lighting requirement information of the abnormal area is determined according to the characteristic information.

[0031] Optionally, the AI ​​agent-based inspection method further includes:

[0032] When the target device provides illumination for the abnormal area, photographing the abnormal area to obtain a second area image of the abnormal area;

[0033] determining whether lighting requirements are met according to the second area image;

[0034] If the second region image determines that the lighting requirement is not met, determining second feature information of the abnormal region based on the second region image, where the second feature information includes at least one of shadow distribution information, geometric shape characteristics, and reflective region information;

[0035] determining fill light information according to the second characteristic information;

[0036] The fill light information is encapsulated in a bidding request and the bidding request is sent to multiple second devices.

[0037] Optionally, the AI ​​agent-based inspection method further includes:

[0038] Acquire first brightness information sent by the target device, where the first brightness information is brightness information of the abnormal area collected by the target device; and / or

[0039] Acquiring a regional illumination image of the abnormal area to determine second brightness information of the abnormal area;

[0040] Illumination adjustment information is generated according to the first brightness information and / or the second brightness information, and the illumination adjustment information is sent to the target device, so that the target device adjusts the illumination of the abnormal area according to the illumination adjustment information.

[0041] In a second aspect, an embodiment of the present invention provides an AI agent-based inspection method, applied to a second device, comprising:

[0042] Obtaining a bidding request sent by a first device, the bidding request being generated by the first device when it detects that ambient lighting conditions in an abnormal area do not meet a preset condition, the abnormal area being determined by the first device based on a target area image of the area to be inspected, the bidding request including lighting requirement information and a location of the abnormal area;

[0043] generating bidding information according to the bidding request;

[0044] sending the bidding information to the first device, so that the first device determines a target device from a plurality of second devices according to the bidding information;

[0045] When receiving the task execution instruction sent by the first device, lighting is provided for the abnormal area according to the area location and the lighting requirement information.

[0046] In a third aspect, an embodiment of the present invention provides a first device, comprising a memory and a processor, wherein the memory is connected to the processor, and the processor is used to execute one or more computer programs stored in the memory. When the processor executes the one or more computer programs, the first device implements the AI ​​agent-based inspection method as described above.

[0047] In a fourth aspect, an embodiment of the present invention provides a second device, wherein the first device includes a memory and a processor, the memory is connected to the processor, the processor is used to execute one or more computer programs stored in the memory, and when the processor executes the one or more computer programs, the second device implements the AI ​​agent-based inspection method as described above.

[0048] In a fifth aspect, an embodiment of the present invention provides an inspection system, including:

[0049] A first device as described above; and

[0050] A plurality of second devices as described above, wherein the plurality of second devices are respectively communicatively connected with the first device.

[0051] Optionally, the first device is an aerial flying device, and the second device is a ground walking device; or

[0052] The first device is a ground-moving device, and the second device is an aerial device; or

[0053] The first device and the second device are both aerial devices; or

[0054] Both the first device and the second device are ground walking devices.

[0055] In the sixth aspect, an embodiment of the present invention provides a storage medium, characterized in that the storage medium stores a computer program, the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes the above-mentioned AI agent-based inspection method.

[0056] Compared with the prior art, the embodiments of the present invention provide an inspection method, device and inspection system based on an AI intelligent agent, the inspection method including: obtaining a target area image of the area to be inspected, determining the abnormal area based on the target area image, generating a bidding request when the ambient lighting conditions in the abnormal area do not meet the preset conditions, the bidding request including the lighting demand information of the abnormal area and the area location of the abnormal area, sending the bidding request to multiple second devices so that the multiple second devices generate bidding information based on the bidding request and send it to the first device, and upon receiving the bidding information sent by the second device, determining the target device from the multiple second devices based on the bidding information, and sending a task execution instruction to the target device so that when the target device receives the task execution instruction, it provides lighting for the abnormal area based on the area location and lighting demand information. On the one hand, by controlling the second device through the first device to provide lighting for the abnormal area, the lighting energy consumption of the first device can be transferred to the second device, thereby saving the energy consumption of the first device and further increasing the battery life of the first device. Since the first device does not need to provide lighting for the abnormal area, it can also avoid exposure of the first device and ensure the safety of the first device. On the other hand, by selecting the second device that best matches the lighting needs to provide lighting for the abnormal area, the lighting needs can be met to the greatest extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0058] Figure 1 A schematic structural diagram of an inspection system provided by an embodiment of the present invention;

[0059] Figure 2 A schematic structural diagram of an inspection system provided in another embodiment of the present invention;

[0060] Figure 3 A flowchart of an AI-based inspection method provided by an embodiment of the present invention;

[0061] Figure 4 A schematic diagram of the process of step S35 in an AI agent-based inspection method provided in an embodiment of the present invention;

[0062] Figure 5 A flowchart of an AI-based inspection method according to another embodiment of the present invention is provided;

[0063] Figure 6A schematic structural diagram of an AI-based inspection device provided in an embodiment of the present invention;

[0064] Figure 7 A schematic structural diagram of an AI-based inspection device according to another embodiment of the present invention;

[0065] Figure 8 A schematic structural diagram of a second determination module in an AI-agent-based inspection device provided in an embodiment of the present invention;

[0066] Figure 9 A schematic structural diagram of an AI-based inspection device according to another embodiment of the present invention;

[0067] Figure 10 A schematic structural diagram of an AI-based inspection device according to another embodiment of the present invention;

[0068] Figure 11 A schematic structural diagram of an AI-based inspection device according to another embodiment of the present invention;

[0069] Figure 12 A schematic structural diagram of an AI-based inspection device according to another embodiment of the present invention;

[0070] Figure 13 A schematic structural diagram of an AI-based inspection device according to another embodiment of the present invention;

[0071] Figure 14 A schematic diagram of the hardware structure of a controller provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0072] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0073] It should be noted that, unless there is a conflict, the various features of the embodiments of the present invention may be combined with each other and are all within the scope of protection of the present invention. In addition, although the functional modules are divided in the device schematics and the logical order is shown in the flow charts, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flow charts. Furthermore, the terms "first," "second," "third," etc. used in the present invention do not limit the data or execution order, but only distinguish between identical or similar items with substantially the same functions and effects.

[0074] The embodiment of the present invention provides a patrol inspection system. Figure 1 The inspection system 100 includes a first device 101, multiple second devices 102, a server 103 and a terminal device 104.

[0075] The first device 101 has a mobile capability and can be moved to any location in the inspection area. Figure 2 The first device 101 includes a first controller 1011 and an image acquisition device 1012.

[0076] The first controller 1011 is electrically connected to the image acquisition device 1012, and is used to control the image acquisition device 1012 to acquire images of each key area in the area to be inspected. The area to be inspected is a physical or logical range that is pre-divided according to business needs and used for systematic inspection work. The key area is an area that requires key inspection.

[0077] The first controller 1011 serves as the control core of the first device 101 and is used to control the first device 101 to complete relevant logical operations and execute the inspection method applied to the first device 101 as described below.

[0078] The image acquisition device 1012 can be any hardware device for acquiring images, including but not limited to electronic devices such as cameras, infrared sensors, still cameras, video cameras, and depth cameras.

[0079] In some embodiments, as Figure 2 As shown, the first device 101 can also be installed with a first light source module 1013, the first light source module 1013 includes a first rotating base 10131 and a first light source 10132, the first light source 10132 is installed on the first rotating base 10131, the first rotating base 10131 and the first light source 10132 are electrically connected to the first controller 1011 respectively, the first controller 1011 can control the first rotating base 10131 to rotate to drive the first light source 10132 to rotate, so that the first light source 10132 can shine in any direction.

[0080] In some embodiments, first device 101 is an agent. In the field of artificial intelligence, an agent refers to a device that can perceive the environment through sensors and take actions through actuators to achieve its technical goals. In some embodiments, the agent can be any autonomous physical device, including but not limited to electronic devices such as drones, unmanned vehicles, humanoid robots, and insect robots.

[0081] The second device 102 is connected to the first device 101 for group collaboration with the first device 101. Group collaboration ensures the accuracy of inspection targets and the safety of task execution entities. The second device 102 also has mobility and can be moved to any location in the inspection area.

[0082] In some embodiments, the first device 101 is an aerial device or a ground-based device, and the second device 102 is an aerial device or a ground-based device. An aerial device is an electronic device that primarily operates in the air, such as a drone, and a ground-based device is an electronic device that primarily operates on the ground, such as a robot dog.

[0083] In some embodiments, the first device 101 is an aerial device and the second device 102 is a ground-based device. It is understood that in this inspection scenario, the first device 101 and the second device 102 implement ground-air collaborative inspection, with the first device 101 playing a leading role in the ground-air collaborative inspection.

[0084] In some embodiments, the first device 101 is a ground-based walking device, and the second device 102 is an aerial flying device. It is understood that in this inspection scenario, the first device 101 and the second device 102 implement ground-air collaborative inspection, and the second device 102 plays a leading role in the ground-air collaborative inspection.

[0085] In some embodiments, the first device 101 and the second device 102 are both aerial devices. It is understandable that in this inspection scenario, the first device 101 and the second device 102 implement air-to-air collaborative inspection.

[0086] In some embodiments, the first device 101 and the second device 102 are both ground-based walking devices. It is understood that in this inspection scenario, the first device 101 and the second device 102 implement ground-to-ground collaborative inspection.

[0087] In some embodiments, as Figure 2 As shown, the second device 102 includes a second controller 1021 , a second light source module 1022 , a light sensor 1023 , an inertial measurement unit 1024 and a laser radar 1025 .

[0088] The second controller 1021 serves as the control core of the second device 102 and is used to control the second device 102 to complete relevant logical operations and execute the inspection method applied to the second device 102 as described below.

[0089] The second light source module 1022 includes a second rotating base 10221 and a first light source 10222. The second light source 10222 is installed on the second rotating base 10221. The second rotating base 10221 and the second light source 10222 are electrically connected to the second controller 1021 respectively. The second controller 1021 can control the second rotating base 10221 to rotate to drive the second light source 10222 to rotate, so that the second light source 10222 can illuminate in any direction.

[0090] The light sensor 1023 is electrically connected to the second controller 1021 for sensing light energy ranging from ultraviolet light to infrared light, converting the light energy into electrical signals and transmitting them to the second controller 1021 so that the second controller 1021 can detect ambient light or other light according to the electrical signals.

[0091] The inertial measurement unit 1024 is electrically connected to the second controller 1021 and is used to detect in real time motion change information of the second device 102, such as angular rate, acceleration, orientation, etc., and transmit this motion change information to the second controller 1021 so that the second controller 1021 can calculate the position and attitude of the second device 102 based on this motion change information, thereby implementing positioning, obstacle avoidance, path planning, navigation, and other operations for the second device 102. In some embodiments, the inertial measurement unit 1024 may include a 3-axis accelerometer and a 3-axis gyroscope, which are used to measure the acceleration and angular velocity of the second device 102 in three-dimensional space, respectively, to calculate the position and attitude of the second device 102.

[0092] The laser radar 1025 is electrically connected to the second controller 1021, and is used to emit laser light into the surrounding environment and calculate the distance between the laser radar 1025 and the object based on the time difference of the signal reflected from the object, and transmit the distance information to the second controller 1021 so that the second controller 1021 can control the second device 102 to avoid the obstacle in time when it moves in front of the obstacle based on the distance information, thereby preventing the second device 102 from colliding with the obstacle. The laser radar 1025 can also collect environmental data of the environment around the second device 102 by continuously rotating and scanning, and transmit it to the second controller 1021 so that the second controller 1021 can perform environmental perception and positioning based on the environmental data, and generate a map by processing real-time environmental data through a map building algorithm, wherein the map building algorithm can be any type of algorithm such as a SLAM (Simultaneous Localization and Mapping) algorithm.

[0093] In some embodiments, the second device 102 is a proxy device, which includes but is not limited to electronic devices such as drones, unmanned vehicles, humanoid robots, and insect robots.

[0094] The server 103 is in communication with the first device 101 and is configured to receive data such as images uploaded by the first device 101 and remotely monitor and manage the first device 101 based on the data. In some embodiments, the server 103 may be a physical server or a logical server virtualized from multiple physical servers. In some embodiments, the server 103 may also be a server cluster consisting of multiple interconnected and communicating servers, with each functional module distributed across each server in the server cluster.

[0095] In some embodiments, the server 103 is configured with an alarm function. The server 103 can identify the image uploaded by the first device 101 and determine whether there is any abnormality. If there is an abnormality, an alarm operation is performed through the alarm interface configured by the alarm function.

[0096] The terminal device 104 is communicatively connected to the server 103 and is used to receive alarm information through an alarm interface corresponding to the alarm function of the server 103. The alarm information can be information presented through email, text message, etc., so that when the user receives the alarm information through the terminal device 104, he can take corresponding actions quickly.

[0097] The terminal device 104 includes but is not limited to user equipment (UE) such as smart phones, desktop computers, laptop computers, digital broadcast receivers, personal digital assistants (PDAs), tablet computers (Portable Android Devices, PADs), handheld devices, vehicle-mounted devices, wearable devices, computing devices or other processing devices connected to wireless modems, mobile stations (MSs), mobile terminals, etc.

[0098] See also Figure 3 The embodiment of the present invention provides an inspection method based on AI agent, wherein AI refers to artificial intelligence (Artificial Intelligence), and agent is a concept in the field of AI, which refers to an agent that can perceive the environment and take actions to achieve specific goals. The agent can be software, hardware or a system with autonomy, adaptability and interaction capabilities. The inspection method is applied to a first device, such as Figure 3 As shown, the inspection method includes:

[0099] S31 , obtaining a target area image of the area to be inspected.

[0100] In this step, as previously described, the area to be inspected is a physical or logical area pre-demarcated based on business needs and used for systematic inspections. The area to be inspected may include multiple key areas, such as areas where critical equipment or facilities are deployed, areas with higher risk levels, or areas where failures or anomalies frequently occur. The target area image is an image encompassing the area to be inspected. When the first device arrives at the area to be inspected, the first controller may trigger the image acquisition device to capture an image of the area to be inspected, thereby obtaining the target area image.

[0101] S32. Determine the abnormal area according to the target area image.

[0102] In this step, the abnormal area is a key area where abnormal conditions exist. The first controller can input the target area image into the image recognition model to determine whether there is an abnormal area. The first controller can obtain multiple sample images of each key area, and then associate each sample image with a preset label to obtain a sample data set, and finally train the sample data set according to the preset machine learning algorithm to obtain an image recognition model. The type of preset machine learning algorithm can be set according to actual needs, including but not limited to decision tree algorithm, random forest algorithm, logistic regression algorithm, neural network algorithm, support vector machine algorithm, etc. The above algorithms can be combined to form a machine learning algorithm library. The first controller can call the machine learning algorithm library, adjust the optimal algorithm parameters according to the data characteristics of the sample data set, and then input the sample data set for training. After the training is completed, the machine learning algorithm library will give an optimal recognition model, and the first device can use the recognition model as an image recognition model.

[0103] In some embodiments, the first controller may determine the abnormality type of the abnormal area based on the target area image. In some embodiments, the first controller may extract image features of the abnormal area from the target area image and determine the abnormality type of the abnormal area based on the image features. The abnormality type may cover various abnormal situations, including security abnormalities, equipment abnormalities, environmental abnormalities, or other abnormalities. Among them, security abnormalities may include human intrusion, suspicious vehicle intrusion, etc., equipment abnormalities may include equipment failure, oil and gas pipeline leakage, environmental abnormalities may include fire, etc., and other abnormalities may include animal intrusion, etc.

[0104] Ambient lighting conditions refer to the lighting conditions in the abnormal area, such as light intensity. The first controller can classify the ambient lighting conditions according to the intensity of the ambient light, for example, dividing the ambient lighting conditions into complete darkness, low light, medium light and high light. Complete darkness means there is no natural light source (such as late at night), low light means there is a weak light source (such as moonlight or distant lights), medium light means there is a certain amount of ambient light (such as nearby lighting), and high light means there is a strong ambient light (such as special night lighting).

[0105] The lighting requirements of the abnormal area can be the same or different depending on the type of abnormality. For example, if the abnormality type is a security abnormality, high-brightness lighting is required to clearly identify the target. If the abnormality type is an equipment abnormality, medium-brightness lighting is required to check the equipment status in detail. If the abnormality type is an environmental abnormality, low-brightness lighting is required to avoid interfering with the natural light at the fire scene. If the abnormality type is other abnormalities, the light source brightness can be adjusted to the lowest or the lighting can be dynamically adjusted according to the specific situation.

[0106] Since the ambient lighting conditions vary at different time points or in different areas, for example, for security-type anomalies that require high-brightness lighting, it may be possible to clearly identify the target without providing high-brightness lighting, or for equipment-type anomalies that require medium-brightness lighting, it may be possible to provide high-brightness lighting to check the equipment status in detail. Therefore, in order to avoid wasting light sources while meeting lighting needs, the first controller can determine the lighting needs based on the specific anomalies of each type of anomaly and the real-time ambient lighting conditions.

[0107] For example, when the specific anomaly is human intrusion and the ambient lighting condition is completely dark, in order to ensure clear identification of the target, the first controller may determine the lighting requirement to be the highest light source brightness (eg, 100%).

[0108] For another example, when the specific abnormality is a device failure and the ambient light condition is low light, the first controller may determine that the lighting requirement is a medium to high light source brightness (such as 70%).

[0109] For another example, when the specific anomaly is a fire and the ambient light condition is medium light, the first controller may determine the lighting requirement to be the minimum light source brightness (eg, 20%) to avoid interfering with the natural light at the fire scene.

[0110] For another example, when the specific anomaly is an animal intrusion and the ambient light condition is high light, the first controller may determine that the lighting requirement is a medium light source brightness (eg, 50%).

[0111] S33. When the ambient lighting conditions in the abnormal area do not meet the preset conditions, generate a bidding request, where the bidding request includes lighting demand information of the abnormal area and the area location of the abnormal area.

[0112] In this step, the first controller can detect the ambient light intensity of the abnormal area and determine whether the ambient light conditions in the abnormal area meet the preset conditions based on the ambient light intensity. For example, when the ambient light intensity is greater than the preset ambient light intensity, the first controller can determine that the ambient light conditions in the abnormal area meet the preset conditions. When the ambient light intensity is less than the preset ambient light intensity, the first controller can determine that the ambient light conditions in the abnormal area do not meet the preset conditions.

[0113] The bid request is used to request the second device to participate in the bidding for the lighting task of performing the abnormal area. The lighting requirement information may include target light intensity, target lighting duration, or target coverage. The area location is used to indicate the specific location of the abnormal area within the inspection area. In some embodiments, the first controller may generate the bid request based on the abnormality type and ambient lighting conditions.

[0114] In some embodiments, the first controller may determine lighting demand information based on the abnormality type and ambient lighting conditions, and encapsulate the lighting demand information and the location of the abnormal area into a bidding request.

[0115] S34: Send a bidding request to the plurality of second devices, so that the plurality of second devices generate bidding information according to the bidding request and send the bidding information to the first device.

[0116] In this step, the bidding information is information submitted to the first device by each second device after evaluating the cost and benefit of performing the lighting task itself, in order to compete with other second devices for the lighting task. In some embodiments, the bidding information may include at least one of energy consumption, confidence value, light intensity, coverage, or device location. For example, the bidding information includes energy consumption, another example includes light intensity and coverage, and another example includes energy consumption, confidence value, light intensity, coverage, and device location.

[0117] Energy consumption refers to the total amount of energy required for the second device to perform the lighting task. The total amount of energy can be electricity. The confidence value refers to the confidence index of the second device in its successful completion of the lighting task. The second device can determine the confidence value based on its own historical task execution or familiarity with the abnormal area. The light intensity refers to the illumination energy that the second device can provide when performing the lighting task. The coverage range refers to the area that the light can cover when the second device performs the lighting task. The device location refers to the current location of the second device.

[0118] S35: Upon receiving the bidding information sent by the second device, determine a target device from the plurality of second devices according to the bidding information.

[0119] In this step, the first controller can set a threshold for the designated target in the bidding information, and exclude the second device that is obviously unsuitable for performing the lighting task according to the set threshold, thereby quickly screening out the second device that is truly suitable for performing the lighting task and improving inspection efficiency.

[0120] S36. Sending a task execution instruction to the target device, so that when the target device receives the task execution instruction, it provides lighting for the abnormal area according to the area location and lighting requirement information.

[0121] In this step, the task execution instruction is used to trigger the target device to execute the lighting task. When the target device receives the task execution instruction, it navigates to the area and controls the second light source module to illuminate the abnormal area based on the parameters such as light intensity and coverage in the bidding information submitted by itself.

[0122] Although the first device, which is the main inspection force, is equipped with a first light source module and can illuminate autonomously during nighttime inspections, the first device will consume a large amount of electricity if the first light source module is continuously turned on, and it is easy to expose its own position, thereby increasing the risk of the first device being attacked. In this embodiment, the first device controls the second device to provide lighting for the abnormal area through the first device, which not only transfers the lighting energy consumption of the first device to the second device, thereby saving the energy consumption of the first device and thus increasing the battery life of the first device, but also the first device does not need to provide lighting for the abnormal area, which can avoid the exposure of the first device and ensure the concealment and safety of the first device. Since the first device can collect the bids submitted by each second device for the lighting task, the first device can select the best bid from each bid and instruct the second device that gave the best bid to perform the lighting task, which is conducive to selecting the second device that best matches the lighting needs and is most suitable for performing the lighting task to provide lighting for the abnormal area, thereby meeting the lighting needs to the greatest extent.

[0123] In some embodiments, before sending a bidding request to multiple second devices, the first controller may also determine whether to perform an autonomous lighting operation or an assisted lighting operation based on the type of anomaly or the terrain conditions near the abnormal area. Autonomous lighting operation means that the first controller controls the first light source module to illuminate the abnormal area, thereby autonomously providing lighting for the abnormal area. For example, when an equipment abnormality such as an oil leak occurs, or when it is inconvenient for the second device to approach the abnormal area due to terrain reasons, the first controller may perform an autonomous lighting operation. In the event of a more dangerous abnormality, such as when someone steals oil, the first controller may perform an assisted lighting operation, on the one hand to capture and transmit high-quality images to the server, and on the other hand to avoid exposing the location of the first device. When the first controller determines to perform an assisted lighting operation, it enters the step of generating a bidding request and sending the bidding request to multiple second devices.

[0124] In some embodiments, the bidding information includes at least energy consumption, and S35 includes: screening a target device from a plurality of second devices according to energy consumption and ambient light conditions.

[0125] In this embodiment, different ambient lighting conditions may correspond to different energy consumption thresholds. Generally speaking, the weaker the ambient lighting conditions, the higher the energy consumption required, and the corresponding energy consumption threshold is also higher. The first controller may select a target device from a plurality of second devices based on energy consumption and a first preset configuration table, wherein the first preset configuration table is configured with a correspondence between ambient lighting conditions and energy consumption thresholds. For example, the first preset configuration table is shown in Table 1 below:

[0126] Table 1

[0127] Ambient lighting conditions Energy consumption threshold (watt-hour / Wh) Complete darkness ETH1 Low light ETH2 Medium light ETH3 High light ETH4

[0128] In some embodiments, as shown in Table 1 above, ETH1>ETH2>ETH3>ETH4.

[0129] The first controller can determine the energy consumption threshold corresponding to the ambient lighting condition according to the first preset configuration table and judge whether the energy consumption is greater than or equal to the energy consumption threshold. If it is greater than or equal to, the second device corresponding to the energy consumption is determined to be the target device; if it is less than, the second device corresponding to the energy consumption is screened out.

[0130] In some embodiments, the bidding information includes at least energy consumption, and S35 includes: screening the target device from the plurality of second devices according to the confidence value and the abnormality type.

[0131] In this embodiment, different anomaly types may correspond to different confidence thresholds. Generally speaking, a higher confidence value is required to deal with anomalies with higher security risks, and the corresponding confidence threshold is also higher. The first controller may filter the target device from multiple second devices based on the confidence value and a second preset configuration table, wherein the second preset configuration table is configured with a correspondence between anomaly types and confidence thresholds. For example, the second preset configuration table is shown in Table 2 below:

[0132] Table 2

[0133] Exception Type Confidence threshold Security exceptions 0.9 Environmental anomalies 0.8 Device exception 0.7 Other abnormalities 0.6

[0134] The first controller can determine the confidence threshold corresponding to the abnormal type according to the second preset configuration table and judge whether the confidence value is greater than or equal to the confidence threshold. If it is greater than or equal to, the second device corresponding to the confidence value is determined to be the target device; if it is less than, the second device corresponding to the confidence value is screened out.

[0135] In some embodiments, each anomaly type is configured with a weight coefficient corresponding to the bidding information. For example, as shown in Table 3 below, the weight coefficients corresponding to the bidding information include a first weight coefficient, a second weight coefficient, a third weight coefficient, a fourth weight coefficient, and a fifth weight coefficient. In some embodiments, the first weight coefficient is a weight coefficient corresponding to the device location, the second weight coefficient is a weight coefficient corresponding to the light intensity, the third weight coefficient is a weight coefficient corresponding to the coverage range, the fourth weight coefficient is a weight coefficient corresponding to the energy consumption, and the fifth weight coefficient is a weight coefficient corresponding to the confidence value.

[0136] Table 3

[0137]

[0138] In some embodiments, for each anomaly type, the sum of the first weight coefficient, the second weight coefficient, the third weight coefficient, the fourth weight coefficient, and the fifth weight coefficient is 1.

[0139] For security-related exceptions, α1+α2+α3+α4+α5=1.

[0140] For device-related abnormalities, α6+α7+α8+α9+α10=1.

[0141] For environmental anomalies, α11+α12+α13+α14+α15=1.

[0142] For other anomalies, α16+α17+α18+α19+α20=1.

[0143] In some embodiments, see Figure 4 , S35 includes:

[0144] S351. Determine a normalized parameter value corresponding to the bidding information based on the bidding information;

[0145] S352: Calculate a weighted sum based on the normalized parameter value corresponding to the bidding information and the weight coefficient corresponding to the bidding information;

[0146] S353: Filter a target device from the plurality of second devices according to the weighted sum.

[0147] In S351, because the dimensions and value ranges of different bidding parameters may differ, the first controller may determine a relative distance based on the device location and the regional location. The relative distance is the distance from the second device to the abnormal area. The relative distance may be a straight-line distance or a non-straight-line distance. For example, when there are no obstacles between the abnormal area and the second device, the relative distance may be a straight-line distance. When there are obstacles between the abnormal area and the second device, the relative distance may not be a straight-line distance. In this case, the relative distance can be determined by combining the device location and the regional location after considering the location of the obstacle. Generally speaking, the shorter the relative distance between the second device and the abnormal area, the greater the advantage the second device has in the competition to become the target device.

[0148] The first controller can normalize the relative distance, light intensity, coverage, and energy consumption separately, standardize each bidding parameter, and convert them to the same scale for comparison and comprehensive calculation. After the first controller normalizes the relative distance, it determines that the normalized parameter value corresponding to the device position is a first normalized value. After the first controller normalizes the light intensity, it determines that the normalized parameter value corresponding to the light intensity is a second normalized value. After the first controller normalizes the coverage, it determines that the normalized parameter value corresponding to the coverage is a third normalized value. After the first controller normalizes the energy consumption, it determines that the normalized parameter value corresponding to the energy consumption is a fourth normalized value. Since the confidence value is a value between 0 and 1, the first controller does not need to normalize the confidence value.

[0149] In some embodiments, the relative distance is negatively correlated with the first normalized value, the light intensity is positively correlated with the second normalized value, the coverage is positively correlated with the third normalized value, and the energy consumption is positively correlated with the fourth normalized value.

[0150] In S352, the first controller may calculate a weighted sum based on the first normalized value and the first weight coefficient, the second normalized value and the second weight coefficient, the third normalized value and the third weight coefficient, the fourth normalized value and the fourth weight coefficient, the confidence value and the fifth weight coefficient.

[0151] In some embodiments, the first controller may calculate the weighted sum according to the following formula:

[0152] S_i=D_i*W1+L_i*W2+R_i*W3+E_i*W4+C_i*W5

[0153] In the above formula, S_i represents the weighted sum corresponding to the i-th second device, D_i represents the first normalized value of the i-th second device, W1 represents the first weight coefficient (such as W1=α1 under security-type anomalies), L_i represents the second normalized value of the i-th second device, W2 represents the second weight coefficient (such as W2=α2 under security-type anomalies), R_i represents the third normalized value of the i-th second device, W3 represents the third weight coefficient (such as W3=α3 under security-type anomalies), E_i represents the fourth normalized value of the i-th second device, W4 represents the fourth weight coefficient (such as W4=α4 under security-type anomalies), C_i represents the confidence value of the i-th second device, and W5 represents the fifth weight coefficient (such as W5=α5 under security-type anomalies).

[0154] For example, for the first second device, the first normalized value, the second normalized value, the third normalized value and the fourth normalized value are D_1, L_1, R_1 and E_1 respectively, the confidence value is C1, and the anomaly type is a security anomaly. Then the weighted sum S_1 corresponding to the first second device is S_1 = D_1*α1+L_1*α2+R_4*α3+E_1*α4+C_1*α5.

[0155] For another example, for the second second device, the first normalized value, the second normalized value, the third normalized value and the fourth normalized value are D_2, L_2, R_2 and E_2 respectively, the confidence value is C2, and the anomaly type is a device-type anomaly. Then the weighted sum corresponding to the second second device is S_2 = D_2*α6+L_2*α7+R_2*α8+E_2*α9+C_2*α10.

[0156] For another example, for the third second device, the first normalized value, the second normalized value, the third normalized value and the fourth normalized value are d_3, L_3, R_3 and E_3 respectively, the confidence value is c3, and the anomaly type is an environmental anomaly. Then the weighted sum corresponding to the third second device is S_3 = D_3*α11+L_3*α12+R_3*α13+E_3*α14+C_3*α15.

[0157] For another example, the first normalized value, second normalized value, third normalized value and fourth normalized value of the fourth second device are D_4, L_4, R_4 and E_4 respectively, the confidence value is C4, and the anomaly type is other anomalies, then the weighted sum S_4 corresponding to the fourth second device is S_4 = D_4*α16+L_4*α17+R_4*α18+E_4*α19+C_4*α20.

[0158] In some embodiments, the first controller can determine one or more second devices corresponding to the maximum weighted sum as the target device. For example, the first controller can arrange the weighted sums in descending order. If the preset configuration instructs the first controller to schedule one second device, the first controller can determine the second device corresponding to the maximum weighted sum as the target device. If the preset configuration indicates that the first controller needs to schedule two second devices, the first controller can determine the second devices corresponding to the weighted sums of the first two ranked devices as the target device.

[0159] In some embodiments, the first controller may record the performance data of each second device in performing similar tasks in the past, including the degree of match between the actual completion of the task and the previous confidence value. For example, in the past few tasks of a second device, its confidence value is relatively consistent with the actual completion effect of the task, indicating that its assessment of its own capabilities is relatively accurate, then in this bidding, its confidence value credibility will be higher, and the first controller can give appropriate bonus points when evaluating. For those second devices whose historical data shows a large difference between the confidence value and the actual performance, the first controller can discount its current confidence value or conduct a more rigorous evaluation. For example, a second device has always overestimated its capabilities in the past, and the actual completion effect of the task is not as good as its confidence value shows. Then in this bidding, the first controller can multiply the confidence value by a discount factor, such as 0.8, to more objectively reflect its true capabilities.

[0160] In some embodiments, the first controller can obtain the historical task records of the second device, which include the historical completion results and historical confidence values ​​of each historical task completed by the second device, and determine the confidence level based on the historical completion results and historical confidence values. The confidence level is used to indicate the degree of match between the historical completion results and the historical confidence values, and the confidence value of the second device in the bidding information is adjusted based on the confidence level.

[0161] For example, in a historical task, if the historical completion result is success and the confidence value is greater than a preset confidence threshold, the first controller marks the historical task as 1.

[0162] For another example, in a historical task, if the historical completion result is success and the confidence value is less than a preset confidence threshold, the first controller marks this historical task as 0.

[0163] For another example, in a historical task, if the historical completion result is failure and the confidence value is greater than a preset confidence threshold, the first controller marks this historical task as 0.

[0164] For another example, in a historical task, if the historical completion result is failure and the confidence value is less than the preset confidence threshold, the first controller marks this historical task as 1.

[0165] In some embodiments, the first controller may add up the mark values ​​of each historical task of each second device, and then divide the sum by the number of historical tasks to obtain the confidence level corresponding to each second device.

[0166] In some embodiments, the first controller can determine whether the confidence value corresponding to each second device is greater than or equal to a preset confidence threshold. If it is greater than or equal to, the first controller can multiply the confidence value of the second device by a bonus coefficient. The bonus coefficient can be any value greater than 1. If the product is greater than 1, the confidence value is determined to be 1. If it is less than, the first controller can multiply the confidence value of the second device by a discount coefficient. The discount coefficient can be any value less than 1.

[0167] Therefore, this embodiment can flexibly adjust the confidence value of each second device according to the historical task execution status of each second device, thereby preventing the second device from misjudging its ability to perform similar tasks, thereby ensuring that the second device with a high degree of match between historical completion results and historical confidence values ​​is in an advantageous position in the competition.

[0168] In some embodiments, the first controller can obtain a first area image of the abnormal area, determine first feature information of the abnormal area based on the first area image, the first feature information includes at least one of shadow distribution information, geometric shape features and reflective area information, and determine lighting requirement information of the abnormal area based on the feature information.

[0169] In this embodiment, before the target device provides lighting for the abnormal area, the first controller may control the image acquisition device to photograph the abnormal area to acquire a first area image of the abnormal area.

[0170] In some embodiments, when the target device provides lighting for the abnormal area, the first controller can control the image acquisition device to photograph the abnormal area to obtain a second area image of the abnormal area, and determine whether the lighting requirements are met based on the second area image. When the second area image determines that the lighting requirements are not met, the second feature information of the abnormal area is determined based on the second area image, and the second feature information includes at least one of shadow distribution information, geometric shape features and reflective area information. The fill light information is determined based on the second feature information, the fill light information is encapsulated in a bidding request and the bidding request is sent to multiple second devices.

[0171] In this embodiment, the second region image is an image captured by the image acquisition device while the target device is providing illumination for the abnormal region. Shadow distribution information is used to characterize the distribution of shadow areas within the second region image. The first controller may employ an image segmentation and recognition algorithm or other image processing algorithm to segment and recognize the second region image to obtain the shadow distribution information. Geometric shape features are used to characterize the geometric shape of the abnormal region. Reflective region information is used to characterize reflective areas within the abnormal region.

[0172] In some embodiments, the fill light information includes a fill light angle and / or a fill light position.

[0173] It is understandable that for some abnormal areas with special geometric features, the lighting provided by the target device may not be able to completely cover the abnormal area, resulting in the loss of some details. This embodiment can request multiple second devices to bid based on the fill light information when the lighting provided by the target device cannot completely cover the abnormal area, so that the multiple second devices generate bidding information and send it to the first device. The first device then determines the fill light device from the multiple second devices based on the bidding information, and sends a task execution instruction to the fill light device, so that the fill light device fills in the area with missing lighting according to the task execution instruction, thereby ensuring that the first device can clearly capture all the details of the abnormal area, thereby improving the inspection effect.

[0174] In some embodiments, the first controller can obtain first brightness information sent by the target device, the first brightness information is the first brightness information of the abnormal area collected by the target device, and / or obtain the regional lighting image of the abnormal area to determine the second brightness information of the abnormal area, generate lighting adjustment information based on the first brightness information and / or the second brightness information, and send the lighting adjustment information to the target device so that the target device adjusts the lighting of the abnormal area according to the lighting adjustment information.

[0175] In this embodiment, the first controller can generate lighting adjustment information based on the first brightness information, or can generate lighting adjustment information based on the second brightness information, or can generate lighting adjustment information based on the first brightness information and the second brightness information. The first brightness information is brightness information collected by the target device through the light sensor sensing the brightness of the abnormal area. The first controller can use brightness detection algorithms such as the average pixel value method and the weighted average method to process the regional lighting image to obtain the second brightness information. The lighting adjustment information is used to dynamically compensate for the brightness of the abnormal area. Since there may be differences between the first brightness information and the second brightness information, the first controller can determine brightness difference information based on the first brightness information and the second brightness information, and then generate lighting adjustment information based on the brightness difference information.

[0176] Therefore, this embodiment can dynamically compensate the brightness of the abnormal area according to the brightness of the abnormal area detected by the target device and the brightness of the abnormal area detected by the first controller, ensuring that the brightness of the abnormal area can meet the requirement of the image acquisition device to clearly capture the image of the abnormal area.

[0177] See also Figure 5 , an embodiment of the present invention provides an inspection method based on an AI agent, the inspection method is applied to a second device, such as Figure 5 As shown, the inspection method includes:

[0178] S51. Obtain a bidding request sent by the first device. The bidding request is generated when the first device detects that the ambient lighting conditions of the abnormal area do not meet the preset conditions. The abnormal area is determined by the first device based on the target area image of the area to be inspected. The bidding request includes lighting requirement information of the abnormal area and the area location of the abnormal area.

[0179] S52: Generate bidding information according to the bidding request.

[0180] S53: Send the bidding information to the first device, so that the first device determines a target device from the plurality of second devices according to the bidding information.

[0181] S54: When receiving the task execution instruction sent by the first device, provide lighting for the abnormal area according to the area location and lighting requirement information.

[0182] Therefore, on the one hand, by controlling the second device through the first device to provide lighting for the abnormal area, the lighting energy consumption of the first device can be transferred to the second device, thereby saving the energy consumption of the first device and further increasing the battery life of the first device. Since the first device does not need to provide lighting for the abnormal area, it can also avoid exposure of the first device and ensure the safety of the first device. On the other hand, by selecting the second device that best matches the lighting needs to provide lighting for the abnormal area, the lighting needs can be met to the greatest extent.

[0183] In some embodiments, the second controller can obtain a bidding request sent by the first device, the bidding request includes fill light information, generate bidding information based on the bidding request and send it to the first device, so that the first device determines the fill light device from multiple second devices based on the bidding information, obtains the task execution instruction sent by the first device, and fills light for the abnormal area based on the fill light information.

[0184] Therefore, on the one hand, this embodiment can dispatch other second devices to supplement the area where lighting is missing when the lighting provided by the target device cannot completely cover the abnormal area, thereby ensuring that the first device can clearly capture all the details of the abnormal area, thereby improving the inspection effect.

[0185] In some embodiments, the second controller controls the light sensor to collect first brightness information of the abnormal area and sends it to the first device, so that the first device generates lighting adjustment information based on the first brightness information and / or the second brightness information. The second brightness information is determined by the first device based on the second area image of the abnormal area, the lighting adjustment information sent by the first device is obtained, and the lighting of the abnormal area is adjusted according to the lighting adjustment information.

[0186] Therefore, this embodiment can dynamically compensate the brightness of the abnormal area according to the brightness of the abnormal area detected by the target device and the brightness of the abnormal area detected by the first controller, ensuring that the brightness of the abnormal area can meet the requirement of the image acquisition device to clearly capture the image of the abnormal area.

[0187] In some embodiments, before sending the bidding information to the first device, the second controller can determine whether the bidding conditions are met based on the lighting demand information. If the bidding conditions are met, the second controller can generate bidding information based on the lighting demand information and send the bidding information to the first device. If the bidding conditions are not met, the second controller may not participate in the bidding.

[0188] Therefore, the second device can decide whether to participate in the bidding based on the lighting demand, which is beneficial for the first device to quickly collect bidding information that meets the bidding conditions, thereby improving inspection efficiency.

[0189] In some embodiments, the lighting demand information includes target light intensity and target coverage range. The second controller can determine whether the maximum light intensity of the second light source module is greater than or equal to the target light intensity and whether the maximum coverage range is greater than or equal to the target coverage range. If the maximum light intensity is greater than or equal to the target light intensity and the maximum coverage range is greater than or equal to the target coverage range, the second controller can determine that the bidding conditions are met; otherwise, the second controller can determine that the bidding conditions are not met.

[0190] It should be noted that, in each of the above-mentioned embodiments, there is not necessarily a certain order between the above-mentioned steps. A person of ordinary skill in the art can understand from the description of the embodiments of the present invention that, in different embodiments, the above-mentioned steps may have different execution orders, that is, they may be executed in parallel, may be executed interchangeably, and so on.

[0191] As another aspect of an embodiment of the present invention, an AI-agent-based inspection device is provided. The AI-agent-based inspection device can be a software module comprising a number of instructions stored in a memory, which a processor can access and execute to implement the AI-agent-based inspection method described in each of the aforementioned embodiments.

[0192] In some embodiments, the AI-agent-based inspection device can be constructed from hardware devices. For example, the AI-agent-based inspection device can be constructed from one or more chips, and the chips can work in coordination with each other to complete the AI-agent-based inspection method described in the above embodiments. For another example, the AI-agent-based inspection device can also be constructed from a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine), a programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination of these components.

[0193] In some embodiments, see Figure 6 The AI ​​agent-based inspection device 600 provided in an embodiment of the present invention includes a first acquisition module 601, a first determination module 602, a first generation module 603, a first sending module 604, a second determination module 605 and a second sending module 606.

[0194] The first acquisition module 601 is used to acquire the target area image of the area to be inspected, the first determination module 602 is used to determine the abnormal area based on the target area image, the first generation module 603 is used to generate a bidding request when the ambient lighting conditions in the abnormal area do not meet the preset conditions, the bidding request includes the lighting demand information of the abnormal area and the area location of the abnormal area, the first sending module 604 is used to send the bidding request to multiple second devices, so that the multiple second devices generate bidding information according to the bidding request and send it to the first device, the second determination module 605 is used to determine the target device from the multiple second devices according to the bidding information when receiving the bidding information sent by the second device, and the second sending module 606 is used to send a task execution instruction to the target device, so that when the target device receives the task execution instruction, it provides lighting for the abnormal area according to the area location and lighting demand information.

[0195] In some embodiments, the bidding information includes at least energy consumption, and the second determining module 605 is specifically configured to filter a target device from a plurality of second devices according to energy consumption and ambient light conditions.

[0196] In some embodiments, see Figure 7 The AI ​​agent-based inspection device 600 also includes a third determination module 607.

[0197] The third determination module 607 is used to determine the abnormality type according to the target area image.

[0198] In some embodiments, the bidding information includes at least a confidence value, and the second determination module 605 is specifically configured to filter a target device from a plurality of second devices according to the confidence value and the abnormality type.

[0199] In some embodiments, each type of bid is configured with a weight coefficient corresponding to the bid information, see Figure 7 The second determination module 605 includes a first determination unit 6051 , a second determination unit 6052 , a calculation unit 6053 and a third determination unit 6054 .

[0200] The first determination unit 6051 is used to determine the normalized parameter value corresponding to the bidding information based on the bidding information, the second determination unit 6052 is used to determine the normalized parameter value corresponding to the bidding information based on the bidding information, the calculation unit 6053 is used to calculate the weighted sum based on the normalized parameter value corresponding to the bidding information and the weight coefficient corresponding to the bidding information, and the third determination unit 6054 is used to filter the target device from multiple second devices based on the weighted sum.

[0201] In some embodiments, see Figure 9 The AI ​​agent-based inspection device 600 further includes a second acquisition module 608 , a fourth determination module 609 and a fifth determination module 610 .

[0202] The second acquisition module 608 is used to obtain a first area image of the abnormal area, the fourth determination module 609 is used to determine first characteristic information of the abnormal area based on the first area image, the first characteristic information includes at least one of shadow distribution information, geometric shape characteristics and reflective area information, and the fifth determination module 610 is used to determine lighting requirement information of the abnormal area based on the characteristic information.

[0203] In some embodiments, see Figure 10 The AI-agent-based inspection device 600 further includes a third acquisition module 611 , a first judgment module 612 , a sixth determination module 613 , a seventh determination module 614 and a third sending module 615 .

[0204] The third acquisition module 611 is used to obtain a second area image of the abnormal area when the target device provides lighting for the abnormal area. The first judgment module 612 is used to judge whether the lighting requirements are met based on the second area image. The sixth determination module 613 is used to determine second feature information of the abnormal area based on the second area image when it is judged that the second area image does not meet the lighting requirements. The second feature information includes at least one of shadow distribution information, geometric shape features and reflective area information. The seventh determination module 614 is used to determine fill light information based on the second feature information. The third sending module 615 is used to encapsulate the fill light information in a bidding request and send the bidding request to multiple second devices.

[0205] In some embodiments, see Figure 11The AI ​​agent-based inspection device 600 also includes a fourth acquisition module 616 , an eighth determination module 617 and a second generation module 618 .

[0206] The fourth acquisition module 616 is used to obtain the first brightness information sent by the target device, where the first brightness information is the brightness information of the abnormal area collected by the target device. The eighth determination module 617 is used to obtain the second area image of the abnormal area and determine the second brightness information of the abnormal area based on the second area image. The second generation module 618 is used to generate lighting adjustment information based on the first brightness information and / or the second brightness information and send the lighting adjustment information to the target device, so that the target device adjusts the lighting of the abnormal area according to the lighting adjustment information.

[0207] In some embodiments, see Figure 12 Another embodiment of the present invention provides an AI agent-based inspection device 600 including a fifth acquisition module 619 , a third generation module 620 , a fourth sending module 621 and a lighting providing module 622 .

[0208] The fifth acquisition module 619 is used to obtain the bidding request sent by the first device. The bidding request is generated by the first device when it detects that the ambient lighting conditions of the abnormal area do not meet the preset conditions. The abnormal area is determined by the first device based on the target area image of the area to be inspected. The bidding request includes the lighting requirement information of the abnormal area and the area location of the abnormal area. The third generation module 620 is used to generate bidding information based on the bidding request. The fourth sending module 621 is used to send the bidding information to the first device so that the first device determines the target device from multiple second devices based on the bidding information. The lighting providing module 622 is used to provide lighting for the abnormal area based on the area location and lighting requirement information when receiving the task execution instruction sent by the first device.

[0209] In some embodiments, see Figure 13 The AI-agent-based inspection device 600 also includes a collection control module 623 , a fifth sending module 624 , a sixth acquisition module 625 and a lighting adjustment module 626 .

[0210] The acquisition control module 623 is used to control the light sensor to collect the first brightness information of the abnormal area. The fifth sending module 624 is used to send the first brightness information to the first device so that the first device generates lighting adjustment information based on the first brightness information and / or the second brightness information. The second brightness information is determined by the first device based on the second area image of the abnormal area. The sixth acquisition module 625 is used to obtain the lighting adjustment information sent by the first device. The lighting adjustment module 626 is used to adjust the lighting of the abnormal area according to the lighting adjustment information.

[0211] It should be noted that the aforementioned AI-agent-based inspection device can execute the AI-agent-based inspection method provided in the embodiments of the present invention, and possesses the corresponding functional modules and beneficial effects of the execution method. For technical details not fully described in the embodiments of the AI-agent-based inspection device, please refer to the AI-agent-based inspection method provided in the embodiments of the present invention.

[0212] See also Figure 14 , Figure 14 A hardware structure diagram of a controller is provided for an embodiment of the present invention. The controller may be the first controller as described above, or the second controller as described above. Figure 14 As shown, the controller 1400 includes one or more processors 1401 and a memory 1402. Figure 14 A processor 1401 is taken as an example.

[0213] Processor 1401 is configured to support the computer device in executing the corresponding functions of the method in the above method embodiment. Processor 1401 can be a central processing unit (CPU), a network processor (NP), a hardware chip, or any combination thereof. The above hardware chip can be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The above PLD can be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0214] Memory 1402 is used to store program code. Memory 1402 may include volatile memory (VM), such as random access memory (RAM); non-volatile memory (NVM), such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid-state drive (SSD); or a combination of the aforementioned types of memory.

[0215] Memory 1402 can be used to store non-volatile software programs, non-volatile computer executable programs, and modules, such as the program instructions / modules corresponding to the AI-agent-based inspection method in the embodiments of the present invention. Processor 1401 executes the non-volatile software programs, instructions, and modules stored in memory 1402 to execute the various functional applications and data processing of the AI-agent-based inspection method and the AI-agent-based inspection device, thereby implementing the functions of the various modules or units of the AI-agent-based inspection method and the AI-agent-based inspection device provided in the above-mentioned method embodiments.

[0216] Memory 1402 may include a program storage area and a data storage area. The program storage area may store an operating system and application programs required for at least one function. The data storage area may store data generated based on the use of the AI-based inspection device. In some embodiments, memory 1402 may optionally include a remote memory device relative to the processor. These remote memories may be connected to the inspection device via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0217] The one or more modules are stored in the memory 1402. When executed by the one or more processors 1401, the AI ​​agent-based inspection method in any of the above-mentioned method embodiments is executed, for example, the method steps described in the above-mentioned method embodiments are executed to realize the functions of the modules described in the above-mentioned device embodiments.

[0218] An embodiment of the present invention further provides a storage medium storing a computer program. The computer program includes program instructions. When executed by a computer, the program instructions enable the computer to execute the method described in the above embodiment.

[0219] It will be understood by those skilled in the art that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware through a computer program. The program can be stored in a computer-readable storage medium, and when executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0220] Finally, it should be noted that the present invention can be implemented in many different forms and is not limited to the embodiments described in this specification. These embodiments are not intended to be additional limitations on the content of the present invention. The purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive. In addition, under the thinking of the present invention, the above-mentioned technical features continue to be combined with each other, and there are many other changes in different aspects of the present invention as described above, all of which are considered to be within the scope of the description of the present invention. Furthermore, it is clear to those skilled in the art that improvements or changes can be made based on the above description, and all such improvements and changes should fall within the scope of protection of the claims appended to the present invention.

Claims

1. An inspection method based on an AI agent, applied to a first device, characterized in that: include: Acquire a target area image of the area to be inspected; determining an abnormal area according to the target area image; When the ambient lighting condition of the abnormal area does not meet the preset condition, generating a bidding request, the bidding request including the lighting demand information of the abnormal area and the area location of the abnormal area; Sending the bidding request to multiple second devices, so that the multiple second devices generate bidding information according to the bidding request and send the bidding information to the first device; Upon receiving the bidding information sent by the second device, determining a target device from the plurality of second devices according to the bidding information; A task execution instruction is sent to the target device, so that when the target device receives the task execution instruction, it provides lighting for the abnormal area according to the area location and the lighting requirement information.

2. The inspection method based on AI agent according to claim 1, characterized in that: The bidding information includes at least one of energy consumption, confidence value, light intensity, coverage range and device location.

3. The inspection method based on AI agent according to claim 1, characterized in that: The bidding information includes at least energy consumption; Determining a target device from a plurality of second devices according to the bidding information includes: A target device is selected from the plurality of second devices according to the energy consumption and the ambient light condition.

4. The inspection method based on AI agent according to claim 1, characterized in that: Also includes: determining the abnormality type according to the target area image; The bidding information at least includes a confidence value; Determining a target device from a plurality of second devices according to the bidding information includes: A target device is screened from the plurality of second devices according to the confidence value and the abnormality type.

5. The inspection method based on AI agent according to claim 1, characterized in that: Also includes: determining an abnormality type according to the target area image, each abnormality type being configured with a weight coefficient corresponding to the bidding information; Determining a target device from a plurality of second devices according to the bidding information includes: Determining a normalized parameter value corresponding to the bidding information according to the bidding information; Calculating a weighted sum based on the normalized parameter value corresponding to the bidding information and the weight coefficient corresponding to the bidding information; A target device is screened from the plurality of second devices according to the weighted sum.

6. The inspection method based on AI agent according to claim 1, characterized in that: Also includes: Acquiring a first regional image of the abnormal region; determining first feature information of the abnormal area based on the first area image, where the first feature information includes at least one of shadow distribution information, geometric shape features, and reflective area information; The lighting requirement information of the abnormal area is determined according to the characteristic information.

7. The inspection method based on AI agent according to claim 1, characterized in that: Also includes: acquiring a second region image of the abnormal region when the target device provides illumination for the abnormal region; determining whether lighting requirements are met according to the second area image; If the second region image determines that the lighting requirement is not met, determining second feature information of the abnormal region based on the second region image, where the second feature information includes at least one of shadow distribution information, geometric shape characteristics, and reflective region information; determining fill light information according to the second characteristic information; The fill light information is encapsulated in a bidding request and the bidding request is sent to multiple second devices.

8. The inspection method based on AI agent according to claim 1, characterized in that: Also includes: Acquire first brightness information sent by the target device, where the first brightness information is brightness information of the abnormal area collected by the target device; and / or acquiring a second region image of the abnormal region and determining second brightness information of the abnormal region according to the second region image; Illumination adjustment information is generated according to the first brightness information and / or the second brightness information, and the illumination adjustment information is sent to the target device, so that the target device adjusts the illumination of the abnormal area according to the illumination adjustment information.

9. An inspection method based on an AI agent, applied to a second device, characterized in that: include: Obtaining a bidding request sent by a first device, the bidding request being generated by the first device when it detects that ambient lighting conditions in an abnormal area do not meet a preset condition, the abnormal area being determined by the first device based on a target area image of the area to be inspected, the bidding request including lighting requirement information and a location of the abnormal area; generating bidding information according to the bidding request; sending the bidding information to the first device, so that the first device determines a target device from a plurality of second devices according to the bidding information; When receiving the task execution instruction sent by the first device, lighting is provided for the abnormal area according to the area location and the lighting requirement information.

10. A first device, characterized in that: The first device includes a memory and a processor, the memory is connected to the processor, and the processor is used to execute one or more computer programs stored in the memory. When the processor executes the one or more computer programs, the first device implements the AI ​​agent-based inspection method as described in any one of claims 1 to 8.

11. A second device, characterized in that: The second device includes a memory and a processor, the memory is connected to the processor, and the processor is used to execute one or more computer programs stored in the memory. When the processor executes the one or more computer programs, the second device implements the AI ​​agent-based inspection method as described in claim 9.

12. A patrol inspection system, characterized in that: include: The first device as claimed in claim 10; as well as A plurality of the second devices as claimed in claim 11, wherein the plurality of the second devices are communicatively connected to the first device respectively.

13. The inspection system according to claim 12, characterized in that: The first device is an aerial flying device, and the second device is a ground walking device; or The first device is a ground-moving device, and the second device is an aerial device; or The first device and the second device are both aerial devices; or Both the first device and the second device are ground walking devices.

14. A storage medium, characterized in that The storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor, the processor executes the AI ​​agent-based inspection method according to any one of claims 1 to 8 or claim 9.

Images