Multi-bionic robot dog cross-floor three-dimensional inspection method and system

Abstract

The invention discloses a multi-bionic robot dog cross-floor three-dimensional inspection method and system, and relates to the technical field of intelligent inspection, and the method comprises the steps: issuing a task distribution list to a multi-bionic robot dog, and generating an initial inspection path; acquiring real-time environment information data through a multi-source sensor according to the initial inspection path; based on the real-time environment information data, using a Kalman filtering algorithm to fuse the data, and constructing a floor three-dimensional environment model; based on accurate positioning of multiple bionic robot dogs, a cross-floor three-dimensional inspection path is generated through a dynamic path planning algorithm and real-time environment information data. According to the method, the high-precision floor three-dimensional environment model is constructed by fusing the real-time environment information data acquired by the multi-source sensor through the Kalman filtering algorithm, so that the accuracy and the dynamic adaptability of environment perception are improved, reliable data support is provided for subsequent path planning, and the method is suitable for popularization and application. Therefore, the robot dog can safely and effectively navigate and avoid collision in a changing environment.

Description

Technical Field

[0001] The present invention relates to the field of intelligent inspection technology, and in particular to a method and system for three-dimensional inspection across floors by multiple bionic robot dogs. Background Art

[0002] In recent years, with the rapid development of multi-bionic robot dog technology, bionic robot dogs have been widely used in various fields such as industrial inspection, emergency rescue, and military missions. The design of these multi-bionic robot dogs is inspired by quadrupeds in nature. They not only have excellent maneuverability and adaptability, but also can perform high-precision tasks in complex environments. Especially for large facilities such as factories, warehouses, and commercial buildings, regular safety inspections are essential. This requires the inspection system to be able to cover large areas efficiently and accurately and respond quickly to potential risks. However, traditional inspection methods often rely on manual operation or simple automated equipment, which is not only inefficient but also difficult to cope with complex multi-story building environments.

[0003] Despite this, existing inspection technologies and systems still have significant limitations. First, current inspection multi-bionic robot dogs can usually only work within a single floor and lack the ability to conduct three-dimensional inspections across floors. This is a major challenge for scenarios that require comprehensive coverage of multi-story buildings. Second, existing path planning algorithms are mostly based on static maps and cannot adapt to environmental changes in real time, resulting in omissions or duplications during the inspection process. In addition, the lack of sensor data fusion and processing capabilities is also an urgent problem to be solved, because a single type of sensor can hardly provide enough information to build an accurate three-dimensional environmental model, and the fusion of multi-source sensor data requires efficient algorithm support. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a three-dimensional inspection method for multiple bionic robot dogs across floors, which solves the problem that efficient full-coverage inspection and real-time dynamic path adjustment cannot be achieved in multi-story buildings.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, the present invention provides a method for cross-floor three-dimensional inspection by multiple bionic robot dogs, which includes issuing a task list to multiple bionic robot dogs to generate an initial inspection path; based on the initial inspection path, collecting real-time environmental information data through multi-source sensors; based on the real-time environmental information data, using the Kalman filter algorithm to fuse the data to construct a three-dimensional floor environment model; based on the three-dimensional floor environment model, using the A* algorithm combined with deep reinforcement learning to perform path planning to generate a floor inspection path; based on the floor inspection path, multiple bionic robot dogs perform cross-floor inspection switching, and use multi-source sensors to generate precise positioning of multiple bionic robot dogs; based on the precise positioning of multiple bionic robot dogs, a dynamic path planning algorithm and real-time environmental information data are used to generate a cross-floor three-dimensional inspection path.

[0008] As a preferred solution of the method for three-dimensional inspection of multiple floors by multiple bionic robot dogs of the present invention, the specific steps of issuing a task list to multiple bionic robot dogs and generating an initial inspection path are as follows:

[0009] Multiple bionic robot dogs start inspections and obtain a task list based on the issued building structure drawings and inspection requirements;

[0010] Generate the total estimated path cost using the A* algorithm based on the building structure diagram and task list;

[0011] Through deep reinforcement learning, the Q-learning formula is used to update the path selection and generate the value of the state action;

[0012] Based on the total estimated path cost and the value of state actions, the A* algorithm combined with deep reinforcement learning is used to generate the initial inspection path.

[0013] As a preferred solution of the multi-bionic robot dog cross-floor three-dimensional inspection method of the present invention, wherein: according to the initial inspection path, real-time environmental information data is collected through multi-source sensors, and the specific steps are as follows:

[0014] According to the initial inspection path, multiple bionic robot dogs move along the path and activate lidar sensors, camera sensors, ultrasonic sensors and inertial measurement unit sensors to collect real-time environmental information data;

[0015] Collect real-time environmental information data including obstacle location, crowd density, temperature and humidity, air quality, sound level, light intensity, magnetic field strength, posture and motion status and GPS location.

[0016] As a preferred solution of the multi-bionic robot dog cross-floor three-dimensional inspection method of the present invention, wherein: based on real-time environmental information data, the Kalman filter algorithm is used to fuse data to construct a three-dimensional floor environment model. The specific steps are as follows:

[0017] Based on real-time environmental information data, the Kalman filter algorithm is used for fusion to generate a state vector;

[0018] By combining the state vector and the position coordinates of multiple bionic robot dogs in three-dimensional space, a three-dimensional grid diagram of the floor is constructed;

[0019] Based on the three-dimensional grid map of the floor, the occupancy probability of multiple bionic robot dogs in each grid cell and the average value of environmental parameters are calculated;

[0020] Based on the occupancy probability of multiple bionic robot dogs in grid cells and the average values of environmental parameters, a three-dimensional floor environment model is generated.

[0021] As a preferred solution of the multi-bionic robot dog cross-floor three-dimensional inspection method of the present invention, wherein: based on the three-dimensional floor environment model, the A* algorithm combined with deep reinforcement learning is used for path planning to generate the floor inspection path. The specific steps are as follows:

[0022] Based on the three-dimensional environment model of the floor, the A* algorithm is used to calculate the shortest path from the starting point to the end checkpoint;

[0023] Based on the generated shortest path combined with deep reinforcement learning, a reward mechanism is added to the deep reinforcement learning process to smooth the path, reduce the number of turns of the multi-bionic robot dog and optimize the inspection path;

[0024] According to the optimized inspection path, multiple bionic robot dogs comprehensively consider the current environmental status to avoid collisions, and use the deep Q network to generate floor inspection paths.

[0025] As a preferred solution of the method for three-dimensional cross-floor inspection of multiple bionic robot dogs of the present invention, wherein: according to the floor inspection path, multiple bionic robot dogs perform cross-floor inspection switching, and use multi-source sensors to generate precise positioning of multiple bionic robot dogs. The specific steps are as follows:

[0026] According to the floor inspection path, when the multi-bionic robot dog approaches the stairs and elevators, it confirms the surrounding environment through sensors and uses computer vision algorithms to achieve obstacle avoidance behavior;

[0027] The multi-bionic robot dog is equipped with a camera and an inertial measurement unit, and uses visual inertial odometry to analyze changes in feature points between consecutive image frames;

[0028] Based on the acceleration and angular velocity data provided by the inertial measurement unit, the multi-bionic robot dog estimates its own movement trajectory and obtains precise positioning of the multi-bionic robot dog.

[0029] As a preferred solution of the multi-bionic robot dog cross-floor three-dimensional inspection method of the present invention, wherein: based on the precise positioning of multiple bionic robot dogs, a cross-floor three-dimensional inspection path is generated through a dynamic path planning algorithm and real-time environmental information data. The specific steps are as follows:

[0030] Based on the precise positioning of multiple bionic robot dogs, multiple bionic robot dogs continuously collect real-time environmental information data during their cross-floor inspection routes;

[0031] When the multi-bionic robot dog discovers new obstacles and environmental changes, it optimizes the cross-floor inspection path through the collected real-time environmental information data;

[0032] A dynamic path planning algorithm is also used to calculate the optimal cross-floor inspection path based on the positions, target points and real-time environmental information data of multiple bionic robot dogs, generating a cross-floor three-dimensional inspection path.

[0033] In the second aspect, the present invention provides a multi-bionic robot dog cross-floor three-dimensional patrol system, including an initial path generation module, a real-time data acquisition module, a model construction module, a patrol path generation module, a precise positioning generation module and a three-dimensional patrol path generation module; the initial path generation module is used to issue a task list to multiple bionic robot dogs to generate an initial patrol path; the real-time data acquisition module is used to collect real-time environmental information data through multi-source sensors based on the initial patrol path; the model construction module is used to fuse data based on real-time environmental information data using a Kalman filter algorithm to construct a three-dimensional floor environment model; the patrol path generation module is used to perform path planning based on the three-dimensional floor environment model using an A* algorithm combined with deep reinforcement learning to generate a floor patrol path; the precise positioning generation module is used to switch multiple bionic robot dogs across floors according to the floor patrol path, and generate precise positioning of multiple bionic robot dogs using multi-source sensors; the three-dimensional patrol path generation module is used to generate a three-dimensional cross-floor patrol path based on the precise positioning of multiple bionic robot dogs through a dynamic path planning algorithm and real-time environmental information data.

[0034] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the method for three-dimensional cross-floor inspection of multiple bionic robot dogs as described in the first aspect of the present invention is implemented.

[0035] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the method for three-dimensional cross-floor inspection by multiple bionic robot dogs as described in the first aspect of the present invention is implemented.

[0036] The beneficial effects of this invention are as follows: by utilizing the Kalman filter algorithm to fuse real-time environmental information data collected by multiple sensors, a high-precision three-dimensional floor environment model is constructed. This not only improves the accuracy and dynamic adaptability of environmental perception, but also provides reliable data support for subsequent path planning, thereby ensuring that the robot dog can safely and effectively navigate in a changing environment and avoid collisions. Secondly, based on this three-dimensional environmental model, the A* algorithm combined with deep reinforcement learning is used for path planning. This not only finds the shortest path from the starting point to the end point, but also optimizes the path through a reward mechanism, reducing the number of turns and unnecessary repeated inspections, thereby improving inspection efficiency and comprehensive coverage capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] 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. 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 work.

[0038] Figure 1 This is a flow chart of the method for three-dimensional cross-floor inspection by multiple bionic robot dogs in Example 1.

[0039] Figure 2 This is a schematic diagram of the multi-bionic robot dog cross-floor three-dimensional inspection system in Example 1. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0041] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0042] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0043] Example 1, with reference to Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a method for three-dimensional inspection across floors by multiple bionic robot dogs, comprising the following steps:

[0044] S1. Send a task list to multiple bionic robot dogs to generate an initial inspection path;

[0045] Multiple bionic robot dogs start inspections and obtain a task list based on the issued building structure drawings and inspection requirements;

[0046] It should be noted that the building structure diagram includes floor plans, stair locations, elevator locations and other important landmarks; the inspection requirements define the specific content that needs to be inspected, such as safety inspections of specific areas, equipment status monitoring or finding potential problem points.

[0047] Generate the total estimated path cost using the A* algorithm based on the building structure diagram and task list;

[0048] It should be noted that the A* algorithm is used to generate the total estimated path cost expression as follows:

[0049] f(n)=g(n)+h(n);

[0050] Where f(n) represents the total estimated path cost of node n, g(n) represents the actual path cost of node n, h(n) represents the estimated path cost of node n, and n represents a node in the path.

[0051] Through deep reinforcement learning, the Q-learning formula is used to update the path selection and generate the value of the state action;

[0052] It should be noted that the Q-learning formula is used to update the path selection and generate the value of the state action, which is expressed as:

[0053]

[0054] Among them, Q represents the value of the state action, α represents the learning rate that determines the influence of the newly acquired information on the current Q value, r represents the benefit after taking the action in the current state, and γ represents the discount factor. Q ′ It represents the maximum Q value's estimate of the optimal path in the future.

[0055] Based on the total estimated path cost and the value of state actions, the A* algorithm combined with deep reinforcement learning is used to generate the initial inspection path;

[0056] It should be noted that the A* algorithm combined with deep reinforcement learning is used to generate the initial inspection path, which is expressed as:

[0057]

[0058] Among them, P iRepresents the initial inspection path; P represents the set of all possible paths, each path consists of a series of nodes, which represent the positions where the multi-bionic robot dog can move; n represents a node in the path; Q is the information obtained from deep reinforcement learning, which reflects the estimate of the long-term expected benefit after taking a certain action.

[0059] S2. Collect real-time environmental information data through multi-source sensors based on the initial inspection route;

[0060] According to the initial inspection path, multiple bionic robot dogs move along the path and activate lidar sensors, camera sensors, ultrasonic sensors and inertial measurement unit sensors to collect real-time environmental information data;

[0061] It should be noted that as the multi-robot dog moves along a path, all of the aforementioned sensors work together to collect real-time environmental information. This data is not only used for immediate decision-making, such as avoiding obstacles or adjusting travel direction, but is also recorded for subsequent analysis.

[0062] Collect real-time environmental information data including obstacle location, crowd density, temperature and humidity, air quality, sound level, light intensity, magnetic field intensity, posture and motion status, and GPS location;

[0063] It should be noted that the Obstacle Location function identifies and locates surrounding obstacles, helping the Multi-Bionic Robot Dog plan its path to avoid collisions. The Crowd Density function estimates the number of people in a specific area by analyzing human silhouettes or movement patterns in a video stream. This is crucial for public safety monitoring or crowd management. The Temperature and Humidity function monitors changes in ambient temperature and humidity, which is useful for assessing environmental comfort or warning of abnormal conditions such as fires. The Air Quality function monitors air quality parameters in real time, contributing to health protection and environmental pollution control. The Sound Level function measures noise levels, which can be used for noise pollution monitoring or detecting abnormal sound signals such as alarms or mechanical failures. The Light Intensity function measures light intensity, which can be used to adjust the Multi-Bionic Robot Dog's camera exposure settings or automatically control lighting systems. The Magnetic Field Strength function measures the Earth's magnetic field strength, typically used to assist navigation systems in correcting for directional deviations, especially in weak GPS signal conditions. The Attitude and Motion State function provides information about the Multi-Bionic Robot Dog's attitude (tilt angle, rotation, etc.) and motion state (speed, acceleration), which is crucial for stability and precise control. The GPS Position function obtains coordinate information provided by the Global Positioning System to determine the Multi-Bionic Robot Dog's location, supporting functions such as navigation and geofencing.

[0064] S3, based on real-time environmental information data, uses the Kalman filter algorithm to fuse data and build a three-dimensional floor environment model;

[0065] Based on real-time environmental information data, the Kalman filter algorithm is used for fusion to generate a state vector;

[0066] It should be noted that the Kalman filter algorithm is used for fusion to generate the state vector, which is expressed as:

[0067] x k =Ax k-1 +Bu k +w k ;

[0068] Among them, x k represents the state vector at time k, A represents the state transfer matrix, x k-1 represents the state vector at time k-1, B represents the control input matrix, u k represents the control vector at time k, w k represents the process noise at time k.

[0069] By combining the state vector and the position coordinates of multiple bionic robot dogs in three-dimensional space, a three-dimensional grid diagram of the floor is constructed;

[0070] It should be noted that the position information in the state vector allows for precise positioning of the multi-bionic robot dog in three-dimensional space and, accordingly, adjustment of the relative positions of the point cloud data. The state vector also contains pose information, which is crucial for correctly interpreting sensor data. For example, if the multi-bionic robot dog is tilted, the image captured by the camera must also be adjusted accordingly to accurately reflect the real-world situation.

[0071] Based on the three-dimensional grid map of the floor, the occupancy probability of multiple bionic robot dogs in each grid cell and the average value of environmental parameters are calculated;

[0072] It should be noted that based on each grid cell, the Bayesian update rule is used to update its occupancy probability, which is expressed as:

[0073] l j =L hit (x δ )-L miss (q δ );

[0074] Among them, l j represents the log-odds ratio of grid cell j, L hit (q δ ) The logarithmic probability gain applied when the sensor detects an obstacle, L miss (x δ ) The log-probability impairment applied when the sensor detects free space, q δ represents the position of the δth sensor reading or measurement point;

[0075] Converting the log odds back to probability, the expression is:

[0076]

[0077] Among them, p j represents the probability that grid cell j is occupied, Ensures that the result of the entire score is always between 0 and 1;

[0078] For the average value of environmental parameters in a grid cell, there are m measurements e1, e2, ..., e m , the expression is:

[0079]

[0080] Among them, E j represents the average value of the jth group of data, m represents the number of elements in the set, e η represents the nth observation value, and n indicates that the summation range starts from n equals 1 until n equals m.

[0081] Generate a three-dimensional floor environment model based on the occupancy probability of multiple bionic robot dogs in the grid cells and the average value of environmental parameters;

[0082] It should be noted that a three-dimensional environment model refers to the digital reproduction of the physical space of the real world through computer technology, thereby creating a three-dimensional representation that can be viewed and interacted with in a virtual environment.

[0083] S4. Based on the three-dimensional floor environment model, use the A* algorithm combined with deep reinforcement learning for path planning to generate floor inspection paths;

[0084] Based on the three-dimensional environment model of the floor, the A* algorithm is used to calculate the shortest path from the starting point to the end checkpoint;

[0085] It should be noted that the A* algorithm is a heuristic search algorithm that selects the next node to explore by estimating the sum of the cost from the current node to the target node and the actual cost from the start node to the current node.

[0086] Based on the generated shortest path combined with deep reinforcement learning, a reward mechanism is added to the deep reinforcement learning process to smooth the path, reduce the number of turns of the multi-bionic robot dog and optimize the inspection path;

[0087] It should be noted that the generated shortest path is combined with the deep reinforcement learning training process;

[0088] The initial path is fed into deep reinforcement learning as a starting point;

[0089] At each time step, an action is selected based on the current state, and after executing the action, the new state and the reward obtained are observed;

[0090] Update the policy network based on the experience gained so that it can better predict future behavior;

[0091] As training progresses, gradually adjust the path to make it smoother and with fewer turns.

[0092] According to the optimized inspection path, multiple bionic robot dogs comprehensively consider the current environmental status to avoid collisions and use the deep Q network to generate floor inspection paths;

[0093] It should be noted that building a deep Q-network involves creating a neural network to approximate the action-value function. This network accepts the state of the environment as input and outputs the expected reward for each possible action taken in that state. By stabilizing the training process through techniques such as experience replay and a fixed Q-target, effective policies can be learned, enabling the agent to make optimal decisions in complex environments. For example, multiple bionic robot dogs can optimize their paths to avoid collisions and maintain appropriate spacing during inspection tasks. All of this is based on a self-improvement mechanism based on data collected from interactions with the environment.

[0094] S5. Based on the floor inspection path, multiple bionic robot dogs perform cross-floor inspection switching and use multi-source sensors to generate precise positioning of multiple bionic robot dogs;

[0095] According to the floor inspection path, when the multi-bionic robot dog approaches the stairs and elevators, it confirms the surrounding environment through sensors and uses computer vision algorithms to achieve obstacle avoidance behavior;

[0096] It should be noted that if a stationary obstacle is detected, the robot can choose to circumvent it and continue forward. If it encounters a dynamic obstacle (such as a pedestrian), it must predict its trajectory and make appropriate avoidance decisions accordingly. When approaching stairs or elevator entrances, the robot must pay special attention to the edge of the steps, the position of the handrails, and the status of the elevator door switches to ensure a smooth transition or wait for the right moment to enter the elevator.

[0097] The multi-bionic robot dog is equipped with a camera and an inertial measurement unit, and uses visual inertial odometry to analyze changes in feature points between consecutive image frames;

[0098] It should be noted that in actual operation, when the multi-bionic robot dog moves, it continuously acquires new image frames and searches for common feature points between these frames. If the position of the feature points shifts between frames, it indicates that the multi-bionic robot dog has moved in a straight line during this period. If the feature points are accompanied by rotation, it means that the multi-bionic robot dog has simultaneously turned.

[0099] Based on the acceleration and angular velocity data provided by the inertial measurement unit, the multi-bionic robot dog estimates its own movement trajectory and obtains its precise positioning;

[0100] It should be noted that the accelerometer measures the linear acceleration of the multi-bionic robot dog relative to the inertial reference frame. Using the angular velocity data provided by the gyroscope, the multi-bionic robot dog's posture is updated through numerical integration. This method can determine the multi-bionic robot dog's roll, pitch, and yaw angles at any moment.

[0101] S6, based on the precise positioning of multiple bionic robot dogs, generates a three-dimensional inspection path across floors through dynamic path planning algorithms and real-time environmental information data;

[0102] Based on the precise positioning of multiple bionic robot dogs, multiple bionic robot dogs continuously collect real-time environmental information data during their cross-floor inspection routes;

[0103] It should be noted that the real-time environmental information data collected is used by cameras to capture visual information, help identify and classify objects, monitor human activities, and build or update environmental maps. LiDAR is used to provide accurate distance measurements and generate three-dimensional point cloud maps, which help understand the spatial layout and detect dynamic obstacles. Ultrasonic sensors are suitable for close-range detection, especially in low-light conditions or transparent materials. Infrared sensors can be used for temperature detection to detect abnormal heat sources, such as overheating of electrical equipment, which is particularly important in power facility inspections. In addition to positioning, IMU can also monitor the robot's posture changes to ensure its stability and safety. Microphone arrays are used to collect sound signals and can be used for noise monitoring or triggering of specific audio events, such as the sound of a fire alarm.

[0104] When the multi-bionic robot dog discovers new obstacles and environmental changes, it optimizes the cross-floor inspection path through the collected real-time environmental information data;

[0105] It should be noted that if the obstacle is small and located near the current path, a simple local obstacle avoidance algorithm can be used to instantly adjust the path, bypassing the obstacle without straying too far from the overall goal. For larger environmental changes or multiple obstacles, a more complex path planning algorithm may be required to recalculate the optimal path from the current location to the destination. During this process, the optimal or suboptimal solution is selected as the new inspection route, taking into account factors such as battery consumption and time efficiency.

[0106] A dynamic path planning algorithm is used to calculate the optimal cross-floor inspection path based on the positions, target points, and real-time environmental information of multiple bionic robot dogs, generating a three-dimensional cross-floor inspection path.

[0107] It should be noted that dynamic path planning algorithms are those that adjust the robot's path based on continuously updated environmental information during its motion. Unlike static path planning, these algorithms can respond to dynamic changes in the environment, such as the appearance of new obstacles or other moving objects. Cross-floor inspections involve multi-dimensional spatial navigation, including horizontally through corridors and rooms, and vertically through stairways or elevators. Therefore, path planning must not only consider the shortest path within a plane but also rationally arrange the movements of going up and down stairs, taking into account the structural characteristics of the building.

[0108] This embodiment also provides a multi-bionic robot dog cross-floor three-dimensional inspection system, including: an initial path generation module, a real-time data acquisition module, a model construction module, a patrol path generation module, a precise positioning generation module and a three-dimensional patrol path generation module; the initial path generation module is used to issue a task list to multiple bionic robot dogs to generate an initial patrol path; the real-time data acquisition module is used to collect real-time environmental information data through multi-source sensors based on the initial patrol path; the model construction module is used to fuse data based on the real-time environmental information data using the Kalman filter algorithm to construct a three-dimensional floor environment model; the patrol path generation module is used to perform path planning based on the three-dimensional floor environment model using the A* algorithm combined with deep reinforcement learning to generate a floor patrol path; the precise positioning generation module is used to switch multiple bionic robot dogs across floors for patrol according to the floor patrol path, and use multi-source sensors to generate precise positioning of multiple bionic robot dogs; the three-dimensional patrol path generation module is used to generate a three-dimensional cross-floor patrol path based on the precise positioning of multiple bionic robot dogs through a dynamic path planning algorithm and real-time environmental information data.

[0109] This embodiment also provides a computer device suitable for the method of three-dimensional cross-floor inspection of multiple bionic robot dogs, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the three-dimensional cross-floor inspection method of multiple bionic robot dogs proposed in the above embodiment.

[0110] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0111] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for realizing a three-dimensional cross-floor inspection of multiple bionic robot dogs as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0112] In summary, the present invention utilizes a Kalman filter algorithm to fuse real-time environmental information data collected by multiple sensors to construct a high-precision three-dimensional floor environment model. This not only improves the accuracy and dynamic adaptability of environmental perception, but also provides reliable data support for subsequent path planning, thereby ensuring that the robot dog can safely and effectively navigate and avoid collisions in a changing environment. Secondly, based on this three-dimensional environmental model, the A* algorithm is combined with deep reinforcement learning for path planning. This not only finds the shortest path from the starting point to the end point, but also optimizes the path through a reward mechanism, reducing the number of turns and unnecessary repeated inspections, thereby improving inspection efficiency and comprehensive coverage capabilities.

[0113] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for three-dimensional inspection across floors using multiple bionic robot dogs, characterized by: include, Send task lists to multiple bionic robot dogs and generate initial inspection routes; Based on the initial inspection route, real-time environmental information data is collected through multi-source sensors; Based on real-time environmental information data, the Kalman filter algorithm is used to fuse data and build a three-dimensional floor environment model; Based on the three-dimensional floor environment model, the A* algorithm combined with deep reinforcement learning is used for path planning to generate floor inspection paths; Based on the floor inspection path, multiple bionic robot dogs conduct cross-floor inspection switching, and use multi-source sensors to generate precise positioning of multiple bionic robot dogs; Based on the precise positioning of multiple bionic robot dogs, a three-dimensional inspection path across floors is generated through dynamic path planning algorithms and real-time environmental information data.

2. The method for three-dimensional inspection of multiple floors by multiple bionic robot dogs according to claim 1, characterized in that: The specific steps of issuing a task list to multiple bionic robot dogs and generating an initial inspection path are as follows: Multiple bionic robot dogs start inspections and obtain a task list based on the issued building structure drawings and inspection requirements; Generate the total estimated path cost using the A* algorithm based on the building structure diagram and task list; Through deep reinforcement learning, the Q-learning formula is used to update the path selection and generate the value of the state action; Based on the total estimated path cost and the value of state actions, the A* algorithm combined with deep reinforcement learning is used to generate the initial inspection path.

3. The method for three-dimensional inspection of multiple floors by multiple bionic robot dogs as claimed in claim 2, characterized in that: According to the initial inspection path, real-time environmental information data is collected through multi-source sensors. The specific steps are as follows: According to the initial inspection path, multiple bionic robot dogs move along the path and activate lidar sensors, camera sensors, ultrasonic sensors and inertial measurement unit sensors to collect real-time environmental information data; Collect real-time environmental information data including obstacle location, crowd density, temperature and humidity, air quality, sound level, light intensity, magnetic field strength, posture and motion status and GPS location.

4. The method for three-dimensional inspection of multiple floors by multiple bionic robot dogs as claimed in claim 3, characterized in that: Based on real-time environmental information data, the Kalman filter algorithm is used to fuse data to construct a three-dimensional floor environment model. The specific steps are as follows: Based on real-time environmental information data, the Kalman filter algorithm is used for fusion to generate a state vector; By combining the state vector and the position coordinates of multiple bionic robot dogs in three-dimensional space, a three-dimensional grid diagram of the floor is constructed; Based on the three-dimensional grid map of the floor, the occupancy probability of multiple bionic robot dogs in each grid cell and the average value of environmental parameters are calculated; Based on the occupancy probability of multiple bionic robot dogs in grid cells and the average values of environmental parameters, a three-dimensional floor environment model is generated.

5. The method for three-dimensional inspection of multiple floors by multiple bionic robot dogs as claimed in claim 4, characterized in that: Based on the floor three-dimensional environment model, the A* algorithm is combined with deep reinforcement learning for path planning to generate a floor inspection path. The specific steps are as follows: Based on the three-dimensional environment model of the floor, the A* algorithm is used to calculate the shortest path from the starting point to the end checkpoint; Based on the generated shortest path combined with deep reinforcement learning, a reward mechanism is added to the deep reinforcement learning process to smooth the path, reduce the number of turns of the multi-bionic robot dog and optimize the inspection path; According to the optimized inspection path, multiple bionic robot dogs comprehensively consider the current environmental status to avoid collisions, and use the deep Q network to generate floor inspection paths.

6. The method for three-dimensional inspection of multiple floors by multiple bionic robot dogs according to claim 5, characterized in that: According to the floor inspection path, multiple bionic robot dogs perform cross-floor inspection switching and use multi-source sensors to generate precise positioning of multiple bionic robot dogs. The specific steps are as follows: According to the floor inspection path, when the multi-bionic robot dog approaches the stairs and elevators, it confirms the surrounding environment through sensors and uses computer vision algorithms to achieve obstacle avoidance behavior; The multi-bionic robot dog is equipped with a camera and an inertial measurement unit, and uses visual inertial odometry to analyze changes in feature points between consecutive image frames; Based on the acceleration and angular velocity data provided by the inertial measurement unit, the multi-bionic robot dog estimates its own movement trajectory and obtains precise positioning of the multi-bionic robot dog.

7. The method for three-dimensional inspection of multiple floors by multiple bionic robot dogs according to claim 6, characterized in that: The method is based on the precise positioning of multiple bionic robot dogs, and generates a three-dimensional inspection path across floors through a dynamic path planning algorithm and real-time environmental information data. The specific steps are as follows: Based on the precise positioning of multiple bionic robot dogs, multiple bionic robot dogs continuously collect real-time environmental information data during their cross-floor inspection routes; When the multi-bionic robot dog discovers new obstacles and environmental changes, it optimizes the cross-floor inspection path through the collected real-time environmental information data; A dynamic path planning algorithm is also used to calculate the optimal cross-floor inspection path based on the positions, target points and real-time environmental information data of multiple bionic robot dogs, generating a cross-floor three-dimensional inspection path.

8. A multi-bionic robot dog cross-floor 3D inspection system, based on the multi-bionic robot dog cross-floor 3D inspection method according to any one of claims 1 to 7, characterized in that: It includes an initial path generation module, a real-time data acquisition module, a model building module, an inspection path generation module, a precise positioning generation module, and a three-dimensional inspection path generation module; The initial path generation module is used to distribute task lists to multiple bionic robot dogs and generate initial inspection paths; Real-time data acquisition module, used to collect real-time environmental information data through multi-source sensors according to the initial inspection path; The model building module is used to build a three-dimensional floor environment model based on real-time environmental information data and the Kalman filter algorithm to fuse data; The inspection path generation module is used to generate floor inspection paths based on the three-dimensional floor environment model using the A* algorithm combined with deep reinforcement learning for path planning; The precise positioning generation module is used to switch multiple bionic robot dogs across floors according to the floor inspection path, and uses multi-source sensors to generate precise positioning of multiple bionic robot dogs; The three-dimensional inspection path generation module is used to generate a three-dimensional inspection path across floors based on the precise positioning of multiple bionic robot dogs through dynamic path planning algorithms and real-time environmental information data.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the multi-bionic robot dog cross-floor three-dimensional inspection method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-bionic robot dog cross-floor three-dimensional inspection method according to any one of claims 1 to 7 are implemented.

Images