Method and system for multi-path dynamic visual field inspection of quadruped robot cluster and application thereof

Abstract

The application discloses a kind of four-legged robot cluster multi-path dynamic visual field's inspection system, method and application, in the technical solution of the present application, the four-legged robot cluster multi-path dynamic visual field's inspection system by cloud platform server and by multiple four-legged robots consisting of robot cluster is constituted, wherein robot cluster adopts random election mechanism and task point polling distribution mechanism, and the master robot responsible for path exploration is generated by election, other slave robot focuses on inspection, when master robot detects multiple-path intersection or confirms early warning, pause and mark current position as node or early warning task point, and reassign all robots to new task point based on global task state, the re-election method of master robot and the global dynamic allocation method of task point are used in the application, efficient, full coverage, high robustness inspection of robot cluster in unknown multi-path environment is realized, especially suitable for underground pipe gallery, substation and other complex scenes.

Description

Technical Field

[0001] This invention relates to the field of robotics and automated inspection, specifically to an inspection method, system, and application for a quadruped robot swarm with multi-path dynamic field of view. Background Technology

[0002] With the development of robotics technology, using mobile robots to replace manual labor for area inspection has become a research hotspot in industry. Quadruped robots, due to their excellent terrain adaptability, have shown great application potential in complex environments. However, the inspection efficiency of a single robot is limited, and the coverage area is small. Existing technologies employ robot swarm collaborative inspection, but these often use centralized pre-programmed path planning, which has the following drawbacks in complex multi-path environments (such as underground utility tunnels and substations) where the path is unknown or dynamically changing:

[0003] Complex path planning: The central controller needs to know the complete environmental map in advance or process the perception data of all robots in real time for global replanning, resulting in high computational load, high communication load, and significant system latency; Poor robustness: If the central controller fails or communication is interrupted, the entire cluster will be paralyzed, and the failure of a single robot may also prevent the area it is responsible for from being inspected; Insufficient adaptability: It is difficult to achieve efficient and full-coverage exploration in unknown environments. When encountering emergencies (such as equipment warnings), there is a lack of dynamic task adjustment mechanisms, which affects overall efficiency; Field of view obstruction: When the cluster moves, the robot in front will obstruct the sensor field of view of the robot behind, affecting the accuracy of path perception and obstacle recognition; Therefore, there is an urgent need for a quadruped robot cluster inspection solution that can adapt to unknown and complex environments and has high robustness and high efficiency. Summary of the Invention

[0004] In order to address the problems mentioned in the background art, the purpose of this invention is to overcome the shortcomings of the prior art and provide a decentralized, highly adaptive, and robust inspection method, system, and application for multi-path dynamic vision of quadruped robot swarms.

[0005] To achieve the above objectives, this invention provides a multi-path dynamic vision inspection system for a quadruped robot swarm. In this system, the system includes a cloud platform server and a robot swarm composed of multiple quadruped robots. The cloud platform server establishes a remote communication connection with the robot swarm to store and update inspection information uploaded by the robot swarm. Each quadruped robot in the swarm is equipped with a path perception module, an inspection perception module, a data processing and communication module, and a movement control and status management module. The path perception module performs three-dimensional environmental perception and obstacle distance measurement to determine the movement path. The module includes a path perception unit integrating a multi-line lidar, a vision sensor, and a position sensor, and a position adjustment unit for raising and lowering the path perception unit to adjust the field of view. The inspection perception module is configured according to the specific inspection scenario and is used to collect specific inspection target data. The data processing and communication module is used to receive data from each sensor and is responsible for communication between the quadruped robots and between the quadruped robots and the cloud platform server. The movement control and state management module is used to control the movement of the quadruped robots and manage their real-time state. The real-time state of the quadruped robots in the robot cluster includes: State 0: Idle state; State 1: Working state.

[0006] Furthermore, in the technical solution of this invention, the robot cluster is configured as follows: during the inspection process, a master robot is elected through a random election mechanism, and the other robots besides the master robot are slave robots. The master robot is responsible for communication between the robot cluster and the cloud platform server, as well as for controlling the movement of the robot cluster and the real-time status of the quadruped robots in the robot cluster. The master robot enables its path perception module but does not enable the inspection perception module. The master robot uses its path perception module to perform three-dimensional perception and obstacle distance measurement to determine the movement path, specifically including:

[0007] When the perceived unobstructed travel direction is a single one, the robot cluster is controlled to travel along the path of that unobstructed travel direction. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 1.

[0008] When multiple unobstructed travel directions are detected, the robot swarm is controlled to stop at the intersection of these directions. At this point, the real-time state of all quadruped robots in the swarm is set to state 0. Simultaneously, the master robot begins node recording and task point allocation, specifically including:

[0009] Search the cloud platform server to query the location information of the intersection point and whether the location and direction information of the multiple barrier-free travel directions corresponding to the intersection point have been marked;

[0010] Unmarked: Record the location information of the intersection point and mark it as a node, then upload it to the cloud platform server; record the location and direction information of multiple accessible directions of the intersection point and sequentially mark these directions as multiple task points, then upload the marked task points and their status to the cloud platform server. The task point status includes:

[0011] Status A: No task point has been assigned, indicating that no quadruped robot has been assigned to inspect this task point direction.

[0012] Status B: Task in progress, indicating that the quadruped robot has been assigned to inspect this task point.

[0013] Status C: Task point completed, indicating that the inspection of this task point direction has been completed;

[0014] Status D: Warning task point, indicating that this task point is a warning location determined based on the inspection target data;

[0015] A polling mechanism is used to assign all quadruped robots in state 0 to multiple task points one by one, and the status of the assigned task points is uploaded to the cloud platform server.

[0016] Marked as: The polling mechanism is used to assign all quadruped robots in state 0 to multiple task points one by one, and the status of the assigned task points is uploaded to the cloud platform server.

[0017] Furthermore, in the technical solution of the present invention, the polling mechanism is represented by sequentially assigning all quadruped robots in state 0 to multiple task points one by one, specifically including:

[0018] If a task point in state A exists, query the cloud platform server and assign all quadruped robots in state 0 to all task points in state A sequentially.

[0019] When the number of quadruped robots in state 0 is greater than the number of task points in state A, each task point in state A is assigned to at least one quadruped robot, and the state flag of the task point in state A assigned to a quadruped robot changes to state B.

[0020] When the number of quadruped robots in state 0 is less than the number of task points in state A, the state of task points in state A that are not assigned to quadruped robots remains unchanged.

[0021] If the cloud platform server does not have a task point in state A but has a task point in state B, then all quadruped robots in state 0 will be assigned to all task points in state B in sequence, and the state of the task points in state B will remain unchanged.

[0022] If the cloud platform server query finds no task point in state A and no task point in state B, the inspection is complete and the process returns.

[0023] Furthermore, in the technical solution of the present invention, the slave robot activates its inspection perception module but not its path perception module. The slave robot collects specific inspection target data through its inspection perception module and sends it to the master robot. The master robot receives and analyzes the inspection target data sent by the slave robot, and further controls the movement of the robot cluster and the real-time status of the quadruped robots in the robot cluster based on the inspection target data, specifically including:

[0024] When the inspection target data reaches the warning threshold, the master control robot controls the robot cluster to stop its inspection progress on that path. At this time, the real-time status of all quadruped robots in the robot cluster is controlled to state 0. The master control robot simultaneously begins to record warnings and allocate task points, specifically including:

[0025] Record the location information corresponding to the inspection target data and mark it as a status D task point, then upload it to the cloud platform server;

[0026] Record the location and direction information of the directions on the path where the inspection has not been completed, mark them as new status A task points, and upload them to the cloud platform server;

[0027] First, a quadruped robot in state 0 is assigned to the marked task point D for early warning and monitoring. Then, the remaining quadruped robots in state 0 are assigned to multiple task points one by one using a polling mechanism, and the status of the assigned task points is uploaded to the cloud platform server.

[0028] Furthermore, in the technical solution of this invention, after all quadruped robots in state 0 are assigned to multiple task points one by one:

[0029] After all quadruped robots in state 0 are assigned to multiple task points in state A: the quadruped robots move along the directional paths of their assigned task points, and all assigned quadruped robots on different task point directional paths re-form a robot cluster and re-elect a master robot and slave robots through a random election mechanism. At this time, the real-time state of all quadruped robots in the robot clusters on different task point directional paths is controlled to state 1; the re-elected master robot and slave robot on different task point directional paths also activate the path perception module and the inspection perception module to determine the travel path and collect specific inspection target data, including the re-elected master robot controlling the travel and real-time state of its corresponding robot cluster and the quadruped robots in the robot cluster according to the inspection target data; when the re-elected master robot on the task point directional path perceives multiple unobstructed travel directions, it also records nodes and assigns task points at the intersection points of the multiple unobstructed travel directions it perceives, and at the same time changes the task point markers corresponding to the task point directional paths it previously traveled to state C.

[0030] After all quadruped robots in state 0 are assigned to multiple task points in state B: the quadruped robots move forward along the directional path of their assigned task points, and all the assigned quadruped robots on the directional path of different task points automatically become slave robots and join the robot cluster that originally existed on the directional path of their task points. At this time, the real-time state of the quadruped robots is controlled to state 1.

[0031] Furthermore, in the technical solution of the present invention, the main control robot determines the travel path by performing three-dimensional perception and obstacle distance measurement through its path perception module. This further includes: when the main control robot detects that there is no direction of travel on the inspection path, the main control robot controls the robot cluster to return along the original path. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 0. During the return process of the robot cluster, each time a node position is passed, a polling mechanism is used to reassign all quadruped robots in state 0 to multiple task points one by one.

[0032] In another aspect, this invention provides an inspection method for a quadruped robot swarm with multi-path dynamic vision, employing the quadruped robot swarm multi-path dynamic vision inspection system described above. In the technical solution of this invention, the inspection method for a quadruped robot swarm with multi-path dynamic vision specifically includes the following steps:

[0033] S1. Configure a robot cluster consisting of multiple quadruped robots. Start the robot at the starting position of the inspection. Elect the master robot and slave robots through a random election mechanism. At this time, the real-time status of all quadruped robots in the robot cluster is controlled to state 0. The master robot starts to record nodes and allocate task points. The master robot first records the location information of the starting position and marks the starting position as a node and uploads it to the cloud platform server.

[0034] S2. The main control robot perceives all possible directions of travel, records the position and direction information of all possible directions of travel, and marks all possible directions of travel as multiple task points in sequence. The marked multiple task points and the status of the task points are uploaded to the cloud platform server.

[0035] S3. The main control robot uses a polling mechanism to assign all quadruped robots in state 0 to multiple task points one by one, and uploads the assigned task point status to the cloud platform server.

[0036] S4. The assigned quadruped robots move along the task point path respectively. All the assigned quadruped robots on the different task point direction paths re-form a robot cluster and re-elect the master robot and slave robot through a random election mechanism. At this time, the real-time state of all quadruped robots in the robot clusters on the different task point direction paths is controlled to state 1.

[0037] S5. When the main control robot detects multiple unobstructed travel directions again, the main control robot controls the robot cluster to stop moving at the intersection of multiple unobstructed travel directions. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 0. The main control robot performs node recording and task point allocation again, and at the same time changes the task point marker corresponding to the previously traveled task point direction path to state C.

[0038] S6. When the main control robot determines that the data of the inspection target has reached the warning threshold, the main control robot controls the robot cluster to stop the inspection on the path. At this time, the real-time status of all quadruped robots in the robot cluster is controlled to state 0, and the main control robot starts to record the warning and allocate task points.

[0039] S7. When the master robot senses that there is no direction to travel on the inspection path, the master robot controls the robot cluster to return along the original path. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 0. During the return process of the robot cluster, each time it passes a node position, the master robot re-uses the polling mechanism to assign all quadruped robots in state 0 to multiple task points one by one.

[0040] Another aspect of the present invention includes the application of a multi-path dynamic field of view inspection method for quadruped robot swarms as described above in underground integrated pipe corridors, large substations, disaster emergency search and rescue sites, and mine roadways.

[0041] Beneficial Effects: In summary, this invention provides a method, system, and application for multi-path dynamic vision inspection of quadruped robot swarms. In the technical solution of this invention, the dynamic master robot election mechanism and node-level task reset allocation rules avoid the single-point failure risk of traditional centralized architectures. Even if a single robot fails, its task can be automatically taken over by the system, ensuring the completion rate of inspection tasks. Secondly, this method exhibits excellent environmental adaptability and exploration efficiency. In the technical solution of this invention, the robot swarm can spontaneously conduct efficient and comprehensive parallel exploration in completely unknown multi-path complex environments, like a biological community, without the need for pre-mapped or complex global planning, greatly reducing the system's dependence on the environment. Furthermore, the separation of exploration and inspection tasks between the master robot and the slave robots achieves optimized allocation of computing and sensing resources. In other words, this invention can achieve automated, full-coverage, and highly reliable inspection of large and complex infrastructure and environments with lower maintenance costs and higher intelligence levels.

[0042] Other features and advantages of the present invention will be set forth in the following description. Attached Figure Description

[0043] To more clearly illustrate the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart illustrating a multi-path dynamic field of view inspection method for a quadruped robot swarm according to an embodiment of the present invention. Detailed Implementation

[0045] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0046] To address the shortcomings of existing technologies, this embodiment provides a decentralized, highly adaptive, and robust inspection method, system, and application for multi-path dynamic vision of quadruped robot swarms.

[0047] This embodiment provides a multi-path dynamic field of view inspection system for a quadruped robot swarm. In this embodiment, the multi-path dynamic field of view inspection system for a quadruped robot swarm includes a cloud platform server and a robot swarm consisting of multiple quadruped robots. The cloud platform server establishes a remote communication connection with the robot swarm to store and update the inspection information uploaded by the robot swarm. The inspection information uploaded by the robot swarm includes the nodes marked by each master robot and their location information, the marked task points and their location and direction information, and the task point status information.

[0048] Specifically, in the robot swarm, each quadruped robot is equipped with a path perception module, an inspection perception module, a data processing and communication module, and a movement control and state management module. The path perception module is used to perform three-dimensional perception of the environment and obstacle distance measurement to determine the movement path. The path perception module includes a path perception unit integrating multi-line LiDAR, vision sensor, and position sensor, as well as a position adjustment unit for raising and lowering the path perception unit to adjust the field of view. The inspection perception module is configured according to the inspection scenario and is used to collect specific inspection target data. The data processing and communication module is used to receive data from various sensors and is responsible for communication between quadruped robots and between quadruped robots and cloud platform servers. The movement control and state management module is used to control the movement of quadruped robots and manage the real-time status of quadruped robots.

[0049] Specifically, in this embodiment, the quadruped robots communicate with each other using WiFi Mesh self-organizing network communication, and the quadruped robots communicate with the cloud platform server using 4G and 5G network communication.

[0050] Specifically, in this embodiment, the path perception module includes a multi-line lidar (16-line and 32-line lidar), a visual sensor (monocular and binocular cameras), a position sensor (GPS module and IMU inertial navigation unit), and a position adjustment unit (servo motor and lead screw mechanism).

[0051] Specifically, in this embodiment, the specific inspection target data collected by the inspection perception module refers to the following data depending on the inspection scenario: image data, temperature data, gas data, and sound data. The inspection perception module is configured with a camera, an infrared thermal imager, a gas sensor, and an acoustic sensor according to different inspection scenarios.

[0052] In this embodiment, the robot cluster is configured such that, during the inspection process, a master robot is elected through a random election mechanism, and the other robots besides the master robot are slave robots. The master robot is responsible for communication between the robot cluster and the cloud platform server, as well as for controlling the movement of the robot cluster and the real-time status of the quadruped robots in the robot cluster. The real-time status of the quadruped robots in the robot cluster includes:

[0053] State 0: Idle state, that is, a quadruped robot in state 0 is an idle state that can be assigned.

[0054] State 1: Working state, that is, the quadruped robot in state 1 is a quadruped robot that cannot be assigned a working state;

[0055] In this system, the master robot activates its path perception module but disables its inspection perception module. The master robot uses its path perception module for 3D perception and obstacle distance measurement to determine its path. The slave robot activates its inspection perception module but disables its path perception module. The slave robot uses its inspection perception module to collect specific inspection target data and sends it to the master robot. The master robot receives and analyzes the inspection target data sent by the slave robot and further controls the movement of the robot cluster and the real-time status of the quadruped robots within the cluster based on this data. It should be noted that the master robot analyzes the inspection target data sent by the slave robot based on warning thresholds. For example, if the inspection target data is image data, temperature data, gas data, or sound data, the corresponding warning thresholds analyzed by the master robot would be the critical similarity between the image data and the warning image, the warning critical temperature for the temperature data, the warning critical content for the gas data, and the warning critical wavelength / frequency / amplitude for the sound data, etc.

[0056] Specifically, in this embodiment, the main control robot determines its travel path by performing three-dimensional perception and obstacle distance measurement through its path perception module, including:

[0057] When the perceived unobstructed travel direction is a single one, the robot cluster is controlled to travel along the path of that unobstructed travel direction. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 1, which is the working state of performing inspection.

[0058] When multiple unobstructed travel directions are detected, the robot swarm stops moving at the intersection of these directions. At this point, the real-time state of all quadruped robots in the swarm is set to state 0, indicating an idle state where inspection has stopped. Simultaneously, the main control robot begins node recording and task allocation, specifically including:

[0059] First, search the cloud platform server to query the location information of the intersection point and whether the location and direction information of the multiple barrier-free travel directions corresponding to the intersection point have been marked;

[0060] When not marked:

[0061] Record the location information of the intersection point and mark it as a node, then upload it to the cloud platform server.

[0062] Record the location and direction information of multiple accessible directions at the intersection point, and sequentially mark these multiple accessible directions as multiple task points. Upload the marked task points and their status to the cloud platform server. The task point status includes:

[0063] Status A: No task point has been assigned, indicating that no quadruped robot has been assigned to inspect this task point direction.

[0064] Status B: Task in progress, indicating that the quadruped robot has been assigned to inspect this task point.

[0065] Status C: Task point completed, indicating that the inspection of this task point direction has been completed;

[0066] Status D: Warning task point, indicating that this task point is a warning location determined based on the inspection target data;

[0067] Next, a polling mechanism is used to assign all quadruped robots in state 0 to multiple task points, including the assignment of task points in state 0 and the assignment of task points in state 1, and the assigned task point status is uploaded to the cloud platform server.

[0068] When already marked:

[0069] A polling mechanism is used to assign all quadruped robots in state 0 to multiple task points one by one, including the assignment of task points in state 0 and the assignment of task points in state 1, and the assigned task point status is uploaded to the cloud platform server.

[0070] Specifically, in this embodiment, the polling mechanism is represented by sequentially assigning all quadruped robots in state 0 to multiple task points one by one, specifically including:

[0071] By querying the cloud platform server, when task points in state A exist, all quadruped robots in state 0 are sequentially assigned to all task points in state A:

[0072] When the number of quadruped robots in state 0 is greater than the number of task points in state A, each task point in state A is assigned to at least one quadruped robot, and the state marker of the task point in state A assigned to a quadruped robot is changed to state B.

[0073] When the number of quadruped robots in state 0 is less than the number of task points in state A, the state of task points in state A that are not assigned to quadruped robots remains unchanged.

[0074] It should be noted that the sequential allocation can be exemplified as follows: For example, 8 quadruped robots in state 0 are sequentially allocated to 3 task points in state A. First, one quadruped robot in state 0 is allocated to the first task point in state A, then another quadruped robot in state 0 is allocated to the second task point in state A, then another quadruped robot in state 0 is allocated to the third task point in state A, then another quadruped robot in state 0 is allocated to the first task point in state A, then another quadruped robot in state 0 is allocated to the second task point in state A, then another quadruped robot in state 0 is allocated to the third task point in state A, then another quadruped robot in state 0 is allocated to the first task point in state A, and then another quadruped robot in state 0 is allocated to the second task point in state A. At this time, the 3 task points in state A are allocated to 3, 3 and 2 quadruped robots respectively.

[0075] By querying the cloud platform server, if there is no task point in state A but there is a task point in state B, all quadruped robots in state 0 will be assigned to all task points in state B in sequence, and the state of the task points in state B will remain unchanged.

[0076] If the cloud platform server does not have a task point in state A and does not have a task point in state B, the inspection is completed and the process is returned.

[0077] Specifically, in this embodiment, after all quadruped robots in state 0 are assigned to multiple task points one by one:

[0078] In this process, after all quadruped robots in state 0 are assigned to multiple task points in state A: the quadruped robots move forward along the directional path of their assigned task points, and all the assigned quadruped robots on the directional path of different task points re-form a robot cluster and re-elect the master robot and slave robot through a random election mechanism. At this time, the real-time state of all quadruped robots in the robot cluster on the directional path of different task points is controlled to state 1, which is the working state of inspection.

[0079] Specifically, in this embodiment, the newly elected master robot and slave robot on different task point direction paths also respectively enable the path perception module and the inspection perception module to determine the travel path and collect specific inspection target data. This includes the newly elected master robot controlling the travel and real-time status of its corresponding robot cluster and the quadruped robots in the robot cluster according to the inspection target data. When the newly elected master robot on the task point direction path perceives multiple unobstructed travel directions, it also records the nodes and assigns task points at the intersection of the multiple unobstructed travel directions it perceives. At the same time, the task point mark corresponding to the previously traveled task point direction path is changed to state C, which means that the path of the previously traveled task point direction has been inspected. Here, the previously traveled task point has been inspected specifically means that the path from the task point to the subsequent new node has been inspected.

[0080] In this process, after all quadruped robots in state 0 are assigned to multiple task points in state B, the quadruped robots move along the directional path of their assigned task points. All the assigned quadruped robots on the directional path of different task points automatically become slave robots and join the robot cluster that originally existed on the directional path of their task points. At this time, the real-time state of the quadruped robots is controlled to state 1, which is the working state of performing inspection.

[0081] Specifically, in this embodiment, the master robot further controls the movement of the robot cluster and the real-time status of the quadruped robots in the robot cluster based on the inspection target data, including:

[0082] When the data of the inspected target reaches the warning threshold, the master robot controls the robot cluster to stop its inspection movement on that path. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 0, which is the idle state where inspection has stopped. The master robot simultaneously begins to record warnings and allocate task points, specifically including:

[0083] Record the location information corresponding to the inspection target data and mark it as a status D task point, then upload it to the cloud platform server;

[0084] Record the location and direction information of the directions on the path where the inspection has not been completed, mark them as new status A task points, and upload them to the cloud platform server;

[0085] First, a quadruped robot in state 0 is assigned to the marked state D task point for early warning and monitoring. Then, the remaining quadruped robots in state 0 are assigned to multiple task points one by one using a polling mechanism, including the assignment of task points in state 0 and task points in state 1. The status of the assigned task points is then uploaded to the cloud platform server.

[0086] Specifically, in this embodiment, the main control robot determines the travel path by performing three-dimensional perception and obstacle distance measurement through its path perception module. This also includes: when it senses that there is no direction to travel on the inspection path, the main control robot controls the robot cluster to return along the original path. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 0, which is the idle state of stopping inspection. During the return process of the robot cluster, each time it passes a node position, the polling mechanism is used to reassign all quadruped robots in state 0 to multiple task points one by one, including the assignment of task points in state 0 and task points in state 1.

[0087] This embodiment also includes an inspection method for a quadruped robot swarm with multi-path dynamic vision, employing the quadruped robot swarm multi-path dynamic vision inspection system described above. Figure 1 This is a flowchart illustrating a multi-path dynamic field-of-view inspection method for a quadruped robot swarm according to an embodiment of the present invention. Figure 1 As shown, in this embodiment, the inspection method for a quadruped robot swarm with multi-path dynamic vision specifically includes the following steps:

[0088] S1. Configure a robot cluster consisting of multiple quadruped robots. Start the robot at the starting position of the inspection. Elect the master robot and slave robots through a random election mechanism. At this time, the real-time status of all quadruped robots in the robot cluster is controlled to state 0. The master robot starts to record nodes and allocate task points. The master robot first records the location information of the starting position and marks the starting position as a node and uploads it to the cloud platform server.

[0089] S2. The main control robot perceives all possible directions of travel, records the position and direction information of all possible directions of travel, and marks all possible directions of travel as multiple task points in sequence. The marked multiple task points and the status of the task points are uploaded to the cloud platform server.

[0090] S3. The main control robot uses a polling mechanism to assign all quadruped robots in state 0 to multiple task points one by one, and uploads the assigned task point status to the cloud platform server.

[0091] S4. The assigned quadruped robots move along the task point path respectively. All the assigned quadruped robots on the different task point direction paths re-form a robot cluster and re-elect the master robot and slave robot through a random election mechanism. At this time, the real-time state of all quadruped robots in the robot clusters on the different task point direction paths is controlled to state 1.

[0092] S5. When the main control robot detects multiple unobstructed travel directions again, the main control robot controls the robot cluster to stop moving at the intersection of multiple unobstructed travel directions. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 0. The main control robot performs node recording and task point allocation again, and at the same time changes the task point marker corresponding to the previously traveled task point direction path to state C.

[0093] S6. When the main control robot determines that the data of the inspection target has reached the warning threshold, the main control robot controls the robot cluster to stop the inspection on the path. At this time, the real-time status of all quadruped robots in the robot cluster is controlled to state 0, and the main control robot starts to record the warning and allocate task points.

[0094] S7. When the master robot senses that there is no direction to travel on the inspection path, the master robot controls the robot cluster to return along the original path. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 0. During the return process of the robot cluster, each time it passes a node position, the master robot re-uses the polling mechanism to assign all quadruped robots in state 0 to multiple task points one by one.

[0095] This embodiment also includes the application of a multi-path dynamic field of view inspection method for quadruped robot swarms as described above in underground integrated pipe corridors, large substations, disaster emergency search and rescue sites, and mine roadways.

[0096] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-path dynamic field-of-view inspection system for a quadruped robot swarm, comprising a cloud platform server and a robot swarm consisting of multiple quadruped robots, characterized in that, The cloud platform server establishes a remote communication connection with the robot cluster to store and update the inspection information uploaded by the robot cluster. In the robot swarm, each quadruped robot is equipped with: The path perception module is used to perform three-dimensional perception of the environment and obstacle ranging to determine the travel path. The path perception module includes a path perception unit integrating a multi-line lidar, a vision sensor and a position sensor, and a position adjustment unit for raising and lowering the path perception unit to adjust the field of view. The inspection perception module is configured according to the specific inspection scenario and is used to collect specific inspection target data. The data processing and communication module is used to receive data from various sensors and is responsible for communication between the quadruped robots and communication between the quadruped robots and the cloud platform server. The movement control and status management module is used to control the movement of the quadruped robot and manage the real-time status of the quadruped robot. The robot cluster is configured as follows: During the inspection process, a master robot is elected through a random election mechanism, and the other robots besides the master robot are slave robots. The master robot is responsible for communication between the robot cluster and the cloud platform server, as well as for controlling the movement of the robot cluster and the real-time status of the quadruped robots in the robot cluster. The master robot activates its path perception module but disables its inspection perception module. The master robot uses its path perception module to perform three-dimensional perception and obstacle distance measurement to determine its travel path. The slave robot activates its inspection perception module but disables its path perception module. The slave robot uses its inspection perception module to collect specific inspection target data and sends it to the master robot. The master robot receives and analyzes the inspection target data sent by the slave robot, and further controls the movement of the robot cluster and the real-time status of the quadruped robots in the robot cluster based on the inspection target data.

2. The inspection system for a quadruped robot swarm with multi-path dynamic field of view according to claim 1, characterized in that, The real-time status of the quadruped robots in the robot cluster includes: State 0: Idle state; Status 1: Working status.

3. The inspection system for a quadruped robot swarm with multi-path dynamic field of view according to claim 2, characterized in that, The main control robot determines its travel path through three-dimensional perception and obstacle distance measurement via its path perception module, specifically including: When the perceived unobstructed travel direction is a single one, the robot cluster is controlled to travel along the path of that unobstructed travel direction. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 1. When multiple unobstructed travel directions are detected, the robot swarm is controlled to stop at the intersection of these directions. At this point, the real-time state of all quadruped robots in the swarm is set to state 0. Simultaneously, the master robot begins node recording and task point allocation, specifically including: Search the cloud platform server to query the location information of the intersection point and whether the location and direction information of the multiple barrier-free travel directions corresponding to the intersection point have been marked; Unmarked: Record the location information of the intersection point and mark it as a node, then upload it to the cloud platform server. Record the location and direction information of multiple accessible directions at the intersection point, and sequentially mark these multiple accessible directions as multiple task points. Upload the marked task points and their status to the cloud platform server. The task point status includes: Status A: No task point has been assigned, indicating that no quadruped robot has been assigned to inspect this task point direction. Status B: Task in progress, indicating that the quadruped robot has been assigned to inspect this task point. Status C: Task point completed, indicating that the inspection of this task point direction has been completed; Status D: Warning task point, indicating that this task point is a warning location determined based on the inspection target data; A polling mechanism is used to assign all quadruped robots in state 0 to multiple task points one by one, and the status of the assigned task points is uploaded to the cloud platform server. Marked as follows: A polling mechanism is used to assign all quadruped robots in state 0 to multiple task points one by one, and the status of the assigned task points is uploaded to the cloud platform server.

4. The inspection system for a quadruped robot swarm with multi-path dynamic field of view according to claim 3, characterized in that, The polling mechanism refers to sequentially assigning all quadruped robots in state 0 to multiple task points one by one, specifically including: If a task point in state A exists, query the cloud platform server and assign all quadruped robots in state 0 to all task points in state A sequentially. When the number of quadruped robots in state 0 is greater than the number of task points in state A, each task point in state A is assigned to at least one quadruped robot, and the state flag of the task point in state A assigned to a quadruped robot changes to state B. When the number of quadruped robots in state 0 is less than the number of task points in state A, the state of task points in state A that are not assigned to quadruped robots remains unchanged. If the cloud platform server does not have a task point in state A but has a task point in state B, then all quadruped robots in state 0 will be assigned to all task points in state B in sequence, and the state of the task points in state B will remain unchanged. If the cloud platform server query finds no task point in state A and no task point in state B, the inspection is complete and the process returns.

5. The inspection system for a quadruped robot swarm with multi-path dynamic field of view according to claim 4, characterized in that, After all the quadruped robots in state 0 are assigned to multiple task points one by one: After all the quadruped robots in state 0 are assigned to multiple task points in state A: The quadruped robots move along the directional paths of their assigned task points. All the quadruped robots assigned on the directional paths of different task points re-form a robot cluster and re-elect the master robot and slave robots through a random election mechanism. At this time, the real-time state of all quadruped robots in the robot clusters on the directional paths of different task points is controlled to state 1. The newly elected master robot and slave robot on different task point directions also use the path perception module and inspection perception module respectively to determine the travel path and collect specific inspection target data, including the newly elected master robot controlling the travel and real-time status of its corresponding robot cluster and the quadruped robots in the robot cluster according to the inspection target data. When the newly elected master robot on the task point direction path perceives multiple unobstructed travel directions, it also records the nodes and assigns task points at the intersection of the multiple unobstructed travel directions it perceives, and at the same time changes the task point mark corresponding to the task point direction path it previously traveled to state C. After all the quadruped robots in state 0 are assigned to multiple task points in state B: The quadruped robots move along the directional paths of their assigned task points. All the quadruped robots assigned on the directional paths of different task points automatically transform into slave robots and join the robot cluster that originally existed on the directional path of their task point. At this time, the real-time state of the quadruped robots is controlled to state 1.

6. The inspection system for a quadruped robot swarm with multi-path dynamic field of view according to claim 5, characterized in that, The main control robot further controls the movement of the robot cluster and the real-time status of the quadruped robots in the robot cluster based on the inspection target data, specifically including: When the inspection target data reaches the warning threshold, the master control robot controls the robot cluster to stop its inspection progress on that path. At this time, the real-time status of all quadruped robots in the robot cluster is controlled to state 0. The master control robot simultaneously begins to record warnings and allocate task points, specifically including: Record the location information corresponding to the inspection target data and mark it as a status D task point, then upload it to the cloud platform server; Record the location and direction information of the directions on the path where the inspection has not been completed, mark them as new status A task points, and upload them to the cloud platform server; First, a quadruped robot in state 0 is assigned to the marked task point D for early warning and monitoring. Then, the remaining quadruped robots in state 0 are assigned to multiple task points one by one using a polling mechanism, and the status of the assigned task points is uploaded to the cloud platform server.

7. The inspection system for a quadruped robot swarm with multi-path dynamic field of view according to claim 3, characterized in that, The main control robot uses its path perception module to perform three-dimensional perception and obstacle distance measurement to determine the travel path. This also includes: when the robot detects that there is no direction to travel on the inspection path, the main control robot controls the robot cluster to return along the original path. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 0. During the return process of the robot cluster, each time a node position is passed, a polling mechanism is used to reassign all quadruped robots in state 0 to multiple task points one by one.

8. A method for multi-path dynamic field of view inspection of a quadruped robot swarm, characterized in that, The inspection system for a quadruped robot swarm with multi-path dynamic field of view according to any one of claims 1-7 specifically includes the following steps: S1. Configure a robot cluster consisting of multiple quadruped robots. Start the robot at the starting position of the inspection. Elect the master robot and slave robots through a random election mechanism. At this time, the real-time status of all quadruped robots in the robot cluster is controlled to state 0. The master robot starts to record nodes and allocate task points. The master robot first records the location information of the starting position and marks the starting position as a node and uploads it to the cloud platform server. S2. The main control robot perceives all possible directions of travel, records the position and direction information of all possible directions of travel, and marks all possible directions of travel as multiple task points in sequence. The marked multiple task points and the status of the task points are uploaded to the cloud platform server. S3. The main control robot uses a polling mechanism to assign all quadruped robots in state 0 to multiple task points one by one, and uploads the assigned task point status to the cloud platform server. S4. The assigned quadruped robots move along the task point path respectively. All the assigned quadruped robots on the different task point direction paths re-form a robot cluster and re-elect the master robot and slave robot through a random election mechanism. At this time, the real-time state of all quadruped robots in the robot clusters on the different task point direction paths is controlled to state 1. S5. When the main control robot detects multiple unobstructed travel directions again, the main control robot controls the robot cluster to stop moving at the intersection of multiple unobstructed travel directions. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 0. The main control robot performs node recording and task point allocation again, and at the same time changes the task point marker corresponding to the previously traveled task point direction path to state C. S6. When the main control robot determines that the data of the inspection target has reached the warning threshold, the main control robot controls the robot cluster to stop the inspection on the path. At this time, the real-time status of all quadruped robots in the robot cluster is controlled to state 0, and the main control robot starts to record the warning and allocate task points. S7. When the master robot senses that there is no direction to travel on the inspection path, the master robot controls the robot cluster to return along the original path. At this time, the real-time state of all quadruped robots in the robot cluster is controlled to state 0. During the return process of the robot cluster, each time it passes a node position, the master robot re-uses the polling mechanism to assign all quadruped robots in state 0 to multiple task points one by one.

9. The application of a multi-path dynamic field of view inspection method for a quadruped robot swarm as described in claim 8 in underground integrated pipe corridors, large substations, disaster emergency search and rescue sites, and mine roadways.

Images