Efficient inspection control method and system for inspection robot

Abstract

The invention discloses an efficient inspection control method and system for an inspection robot, and the method comprises the steps: building a digital three-dimensional geographic model based on the three-dimensional point cloud data, obtained by a radar sensor, of an inspection region; establishing a trajectory prediction model based on space-time constraint based on the historical motion trajectory of the moving obstacle in the inspection area, and predicting the motion trajectory of the moving obstacle in the future 5 seconds; dynamically updating the environment database, and periodically refreshing the space coordinates and the prediction track of the moving object; based on the digital three-dimensional geographic model and the environment database data, according to the priority of the inspection tasks and the equipment state data, scheduling weights are added to the inspection tasks; and calculating and generating a global optimal inspection route of the inspection robot based on an A * algorithm and the scheduling weight. The method has the advantages that by combining a radar sensor, track prediction and an A * algorithm, task scheduling and path planning of the inspection robot are optimized, and efficient, safe and intelligent automatic inspection is achieved.

Description

Technical Field

[0001] The present invention relates to robot control technology, and in particular to a high-efficiency inspection control method and system for an inspection robot. Background Art

[0002] Inspection robots are widely used in the power industry, petrochemicals, transportation, urban infrastructure, and other fields. In the power industry, they regularly check the operating status of substations, transmission lines, and other equipment to identify potential faults. In the petrochemical industry, they can enter high-risk, hard-to-reach areas for inspection, reducing the risk of manual entry into hazardous environments. With the development of 5G technology, inspection robots can also achieve remote control and real-time data transmission, improving the real-time nature of inspection work and remote management capabilities.

[0003] The inspection control methods of inspection robots currently on the market rely on traditional path planning algorithms and lack the ability to adapt to dynamic environments, especially in dealing with moving obstacles. These methods usually rely on static maps or preset paths and cannot effectively predict and avoid sudden moving obstacles, resulting in low inspection efficiency of robots in complex environments and even collisions. Many existing systems respond slowly to environmental changes and are unable to update environmental information in real time. Especially in environments where obstacles change frequently and dynamically, this may cause the robot's path planning to fail or even the task to be completed. In addition, existing methods often lack dynamic scheduling of task priorities and equipment status, and are unable to intelligently adjust the execution order of inspection tasks according to actual conditions, resulting in low resource utilization and extended inspection time. Summary of the Invention

[0004] In order to improve the existing inspection control methods and systems for inspection robots, an efficient inspection control method and system for inspection robots are provided. This method combines radar sensors, trajectory prediction models and A* algorithms to achieve efficient path planning and task scheduling for inspection robots in complex environments. The dynamically updated environmental database and intelligent path adjustment make the inspection process more autonomous, safe and efficient.

[0005] In order to achieve the above objects, the technical solution adopted by the present invention is: An efficient inspection control method for an inspection robot includes: Based on the 3D point cloud data of the inspection area acquired by the radar sensor, a digital 3D geographic model of the inspection area is constructed, which includes terrain elevation, fixed equipment coordinates, and obstacle distribution. Based on the historical movement trajectories of mobile obstacles in the inspection area, a trajectory prediction model based on time and space constraints is established to predict the movement trajectory of mobile obstacles within the next 5 seconds; Based on the acquired motion trajectory data, the dynamically updated environment database periodically refreshes the spatial coordinates and predicted trajectory of the moving object; Based on the digital 3D geographic model and environmental database data, each inspection task is assigned a scheduling weight according to the inspection task priority and equipment status data; Based on the A* algorithm and scheduling weights, the global optimal inspection route of the inspection robot is calculated and generated.

[0006] Preferably, the construction of a digital three-dimensional geographic model of the inspection area including terrain elevation, fixed equipment coordinates, and obstacle distribution based on the three-dimensional point cloud data of the inspection area acquired by the radar sensor specifically includes: Use radar sensors to scan the inspection area, obtain high-precision three-dimensional point cloud data, and process and register the point cloud data; The ground segmentation algorithm is used to separate ground points from non-ground points, extract the terrain surface, and interpolate the ground point cloud data to generate a continuous elevation data model; Segment fixed equipment based on point cloud data and generate a 3D coordinate position dataset of the equipment; By comparing time-series point cloud data, static obstacles are identified based on elevation differences or geometric features, and their location coordinates and dimensions are marked; Based on the above data, a digital three-dimensional geographic model of the inspection area is constructed, which includes terrain elevation, fixed equipment coordinates and obstacle distribution.

[0007] Preferably, the step of establishing a trajectory prediction model based on time and space constraints based on the historical motion trajectory of the mobile obstacle in the inspection area and predicting the motion trajectory of the mobile obstacle within the next 5 seconds specifically includes: Based on the processed time-series point cloud data, the historical motion trajectory data of mobile obstacles in the inspection area is obtained; Based on historical motion trajectory data, extract kinematic spatiotemporal features, including velocity, acceleration, and direction angle, and obtain spatiotemporal constraint features; Input the historical motion trajectory feature data into the LSTM-based neural network model, train it with spatiotemporal constraint features, and obtain the trained trajectory prediction model; The real-time time-series point cloud data is input into the trajectory prediction model to predict the movement trajectory data of the moving obstacle in the next 5 seconds.

[0008] Preferably, the extraction of kinematic spatiotemporal features based on historical motion trajectory data, including velocity, acceleration, and direction angle, and the acquisition of spatiotemporal constraint features specifically include: The loss function is used to penalize excessive acceleration, limiting the acceleration of the predicted trajectory points to no more than the threshold; The curvature penalty term is used to limit the turning radius of the obstacle. Based on the environmental data in the digitized 3D geographic model, a binary obstacle mask is defined to prevent the predicted trajectory from entering the obstacle area.

[0009] Preferably, the dynamically updated environment database based on the acquired motion trajectory data and periodically refreshing the spatial coordinates and predicted trajectory of the moving object specifically includes: Based on the predicted movement trajectory data of moving obstacles in the next 5 seconds, the movement status of objects in the inspection area is updated in real time; Based on a fixed time interval, the latest coordinates of each moving obstacle are written into the database and a timestamp is added to each obstacle. Real-time monitoring of whether the predicted trajectory overlaps with static obstacles or other moving object trajectories. If a conflict exists, trajectory replanning is triggered and the prediction result is updated; Based on the predicted motion trajectory data and the actual motion trajectory data, the prediction error is calculated, and the model is optimized based on incremental learning.

[0010] Preferably, the adding of scheduling weights to each inspection task based on the digital three-dimensional geographic model and the environmental database data according to the inspection task priority and the equipment status data specifically includes: Add priorities to inspection tasks based on their attributes; Based on the terrain elevation, fixed equipment coordinates and obstacle distribution data in the three-dimensional geographic model and the status parameter data of the inspection robot, a scheduling weight factor is added to each inspection task.

[0011] Preferably, the calculation and generation of the global optimal inspection route of the inspection robot based on the A* algorithm and the scheduling weight specifically includes: Build a set of inspection target points based on each inspection task. Each task point contains three-dimensional coordinates, scheduling weight, and time window constraints. Discretize the inspection area into topological nodes, define the connection relationship between nodes based on the robot's mobility, and construct the cost function g(n) and heuristic function h(n); Based on the A* algorithm, create an open list and a closed list, add the starting point to the open list, and set it to f(n)=g(n)+h(n); Take the node n with the smallest f(n) from the open list. If n is the end point, backtrack the path and end. Get all the reachable neighbors of n according to the movement rule. If the task weight or environment data is updated, recalculate g(n) and h(n) of the affected nodes and refresh the open list; After the iteration is completed, an ordered sequence of nodes and passing task points are output as the global optimal inspection route for the inspection robot.

[0012] Furthermore, an efficient inspection control system for the inspection robot is proposed, including: Data processing module: The data processing module is mainly used to obtain three-dimensional point cloud data from the radar sensor and generate a digital three-dimensional geographic model, including terrain elevation, fixed equipment coordinates and obstacle distribution; Trajectory prediction module: The trajectory prediction module is mainly used to predict the movement trajectory of the moving obstacle within the next 5 seconds by combining historical movement trajectory data with the LSTM neural network model; Environmental database update module: The environmental database update module is mainly used to dynamically update the environmental database in real time, refresh the spatial coordinates and predicted trajectory of moving objects, and ensure timely and accurate information; Task scheduling module: The task scheduling module is mainly used to assign scheduling weights to each inspection task according to the priority of the inspection task and the equipment status data, and optimize the task execution order; Path planning module: The path planning module is mainly used to calculate and generate the global optimal inspection route of the inspection robot based on the A* algorithm, combined with task scheduling weights and environmental data; Task execution monitoring module: The task execution monitoring module is mainly used to monitor the task execution status of the inspection robot in real time, and dynamically adjust the path planning and task priority according to the progress of the task; Processor: The processor is mainly used for the calculation process of each formula and the construction calculation process of each model.

[0013] Compared with the prior art, the advantages of the present invention are: By acquiring precise three-dimensional point cloud data through radar sensors and combining algorithms such as ground segmentation and obstacle recognition, a digital three-dimensional geographic model is constructed to ensure a comprehensive understanding of the inspection area. A trajectory prediction model based on LSTM neural networks and spatiotemporal constraints can accurately predict the future motion trajectory of moving obstacles, avoiding possible collisions or delays during the inspection process. This method also further improves the accuracy and flexibility of path planning by dynamically updating the environmental database and adjusting the motion state of obstacles in real time. In scheduling and path planning, the A* algorithm is combined with task priority weighting to ensure that the robot can select the optimal inspection route, maximize resource utilization, and minimize time consumption. Finally, through real-time feedback from the task execution monitoring module, the inspection plan can be dynamically adjusted according to the progress of the task, making the inspection process more intelligent, adaptable, and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A schematic diagram of the method proposed in the present invention; Figure 2 A schematic diagram of constructing a three-dimensional geographic model proposed by the present invention; Figure 3This is a schematic diagram of motion trajectory prediction proposed by the present invention; Figure 4 This is a schematic diagram of feature extraction proposed by the present invention; Figure 5 This is a schematic diagram of the environmental data refresh proposed by the present invention; Figure 6 A schematic diagram of the priority and weight settings proposed by the present invention; Figure 7 Generate a schematic diagram of the optimal inspection route proposed by the present invention; Figure 8 This is a diagram of the architecture of the electronic equipment in this solution; Figure 9 This is a schematic diagram of the computer-readable storage medium structure in this solution. DETAILED DESCRIPTION

[0015] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are merely examples, and those skilled in the art may conceive of other obvious variations.

[0016] An efficient inspection control system for inspection robots, including: Data processing module: The data processing module is mainly used to obtain three-dimensional point cloud data from the radar sensor and generate a digital three-dimensional geographic model, including terrain elevation, fixed equipment coordinates and obstacle distribution; Trajectory prediction module: The trajectory prediction module is mainly used to predict the movement trajectory of the moving obstacle within the next 5 seconds by combining historical movement trajectory data with the LSTM neural network model; Environmental database update module: The environmental database update module is mainly used to dynamically update the environmental database in real time, refresh the spatial coordinates and predicted trajectory of moving objects, and ensure timely and accurate information; Task scheduling module: The task scheduling module is mainly used to assign scheduling weights to each inspection task according to the priority of the inspection task and the equipment status data, and optimize the task execution order; Path planning module: The path planning module is mainly used to calculate and generate the global optimal inspection route of the inspection robot based on the A* algorithm, combined with task scheduling weights and environmental data; Task execution monitoring module: The task execution monitoring module is mainly used to monitor the task execution status of the inspection robot in real time, and dynamically adjust the path planning and task priority according to the progress of the task; Processor: The processor is mainly used for the calculation process of each formula and the construction calculation process of each model.

[0017] See Figure 1 As shown, the efficient inspection control method for the inspection robot includes: Step 1: Based on the 3D point cloud data of the inspection area acquired by the radar sensor, a digital 3D geographic model of the inspection area is constructed, which includes terrain elevation, fixed equipment coordinates, and obstacle distribution. Step 2: Based on the historical movement trajectory of the mobile obstacles in the inspection area, a trajectory prediction model based on time and space constraints is established to predict the movement trajectory of the mobile obstacles within the next 5 seconds; Step 3: Based on the acquired motion trajectory data, the dynamically updated environment database periodically refreshes the spatial coordinates and predicted trajectory of the moving object; Step 4: Based on the digital 3D geographic model and environmental database data, assign scheduling weights to each inspection task according to inspection task priority and equipment status data; Step 5: Based on the A* algorithm and scheduling weights, calculate and generate the global optimal inspection route for the inspection robot.

[0018] See Figure 2 As shown, based on the three-dimensional point cloud data of the inspection area obtained by the radar sensor, a digital three-dimensional geographic model of the inspection area including terrain elevation, fixed equipment coordinates and obstacle distribution is constructed, which specifically includes: Use radar sensors to scan the inspection area, obtain high-precision three-dimensional point cloud data, and process and register the point cloud data; The ground segmentation algorithm is used to separate ground points from non-ground points, extract the terrain surface, and interpolate the ground point cloud data to generate a continuous elevation data model; Segment fixed equipment based on point cloud data and generate a 3D coordinate position dataset of the equipment; By comparing time-series point cloud data, static obstacles are identified based on elevation differences or geometric features, and their location coordinates and dimensions are marked; Based on the above data, a digital three-dimensional geographic model of the inspection area is constructed, which includes terrain elevation, fixed equipment coordinates and obstacle distribution.

[0019] Specifically, point cloud registration is the process of merging multiple point cloud datasets from different perspectives into a unified coordinate system. A common method is to perform point cloud registration based on an iterative closest point algorithm. To extract the terrain surface, ground points and non-ground points need to be segmented from the point cloud, and segmentation and extraction are performed using a segmentation algorithm based on a ground model. In the process of generating the elevation data model, the Kriging interpolation method is used to generate a continuous elevation data model from the ground point cloud data. The formula is:

[0020] in, Target location The estimated elevation of is the elevation of the known point, is the weight, usually calculated through the covariance model, is the average value of the known point elevations; By comparing time-series point cloud data, static obstacles can be identified based on elevation differences or geometric features. Finally, a three-dimensional digital geographic model of the inspection area can be constructed based on the extracted terrain elevation data, fixed equipment coordinates and obstacle information.

[0021] See Figure 3 As shown in the figure, based on the historical motion trajectory of the mobile obstacle in the inspection area, a trajectory prediction model based on time and space constraints is established to predict the motion trajectory of the mobile obstacle in the next 5 seconds. Specifically, the following are included: Based on the processed time-series point cloud data, the historical motion trajectory data of mobile obstacles in the inspection area is obtained; Based on historical motion trajectory data, extract kinematic spatiotemporal features, including velocity, acceleration, and direction angle, and obtain spatiotemporal constraint features; Input the historical motion trajectory feature data into the LSTM-based neural network model, train it with spatiotemporal constraint features, and obtain the trained trajectory prediction model; The real-time time-series point cloud data is input into the trajectory prediction model to predict the movement trajectory data of the moving obstacle in the next 5 seconds.

[0022] Specifically, by matching continuous timestamps with point cloud data, the spatial positions of the moving target at different time points are connected to form a historical motion trajectory. Kinematic spatiotemporal features, including velocity, acceleration, and azimuth, are extracted from the historical trajectory data. Velocity is the rate at which the obstacle position changes over time and can be calculated by position differentials. Acceleration is the rate at which velocity changes over time and can be calculated by velocity differentials. The azimuth is the angle of the obstacle relative to a certain reference direction and can usually be calculated using the velocity vector. The extracted kinematic features and spatiotemporal constraint features are input into the LSTM model to form a training dataset for model training to minimize the error between the predicted position and the actual position. After the model training is completed, real-time time-series point cloud data can be input for trajectory prediction to obtain motion trajectory data for a period of time in the future (such as 5 seconds).

[0023] See Figure 4 As shown in the figure, based on the historical motion trajectory data, the kinematic spatiotemporal features are extracted, including velocity, acceleration, and direction angle, and the spatiotemporal constraint features are obtained, including: The loss function is used to penalize excessive acceleration, limiting the acceleration of the predicted trajectory points to no more than the threshold; The curvature penalty term is used to limit the turning radius of the obstacle. Based on the environmental data in the digitized 3D geographic model, a binary obstacle mask is defined to prevent the predicted trajectory from entering the obstacle area.

[0024] Specifically, if the acceleration exceeds the threshold, a penalty term is added to the loss function, the formula is;

[0025] in, is the cumulative penalty term, is the instantaneous acceleration of the obstacle at time t, is the preset acceleration threshold; If the curvature is too large, it means that the obstacle's turning radius is too small. To avoid excessive turning, a curvature penalty term can be added to the loss function. In order to ensure that the predicted trajectory of the obstacle will not enter the known obstacle area, a binary obstacle mask is defined using the environmental data in the digital 3D geographic model. The mask indicates which areas are obstacles and which areas are open areas.

[0026] See Figure 5 As shown, based on the acquired motion trajectory data, the dynamically updated environment database periodically refreshes the spatial coordinates and predicted trajectory of the moving object, specifically including: Based on the predicted movement trajectory data of moving obstacles in the next 5 seconds, the movement status of objects in the inspection area is updated in real time; Based on a fixed time interval, the latest coordinates of each moving obstacle are written into the database and a timestamp is added to each obstacle. Real-time monitoring of whether the predicted trajectory overlaps with static obstacles or other moving object trajectories. If a conflict exists, trajectory replanning is triggered and the prediction result is updated; Based on the predicted motion trajectory data and the actual motion trajectory data, the prediction error is calculated, and the model is optimized based on incremental learning.

[0027] Specifically, the system obtains the latest point cloud data from the radar sensor, identifies obstacles in the current area based on the point cloud data, and obtains their current motion state by comparing it with historical trajectory data. Based on the three-dimensional coordinates of the moving obstacle, the system calculates the motion state information and updates the object's state. At each fixed time interval, the latest obstacle coordinates are saved to the database. Each record contains information such as obstacle identifier, timestamp, and coordinates. A timestamp is also added to the coordinate data of each obstacle to ensure that the data is stored in a time series, which is convenient for subsequent analysis and monitoring. Based on the defined obstacle mask, the system determines whether a collision has occurred by calculating the overlap between the predicted obstacle position and the static obstacle mask area. It also checks whether the trajectories of other moving objects overlap with the currently predicted trajectory. If a collision occurs or the trajectories overlap, the obstacle's trajectory is recalculated to ensure it avoids static obstacles and other moving objects. By comparing the predicted motion trajectory with the actual observed motion trajectory, the prediction error is calculated, and incremental learning of the model is performed. By adding the latest prediction error data to the training set, the model weight is continuously adjusted to improve the accuracy of the model.

[0028] See Figure 6 As shown in the figure, based on the digital 3D geographic model and environmental database data, and according to the inspection task priority and equipment status data, the scheduling weight is added to each inspection task, specifically including: Add priorities to inspection tasks based on their attributes; Based on the terrain elevation, fixed equipment coordinates and obstacle distribution data in the three-dimensional geographic model and the status parameter data of the inspection robot, a scheduling weight factor is added to each inspection task.

[0029] Specifically, inspection tasks usually have multiple attributes, such as task importance, urgency, and time required. The importance of each task can be evaluated based on the nature of the task. For example, some tasks may involve critical equipment, while other tasks may involve general facilities. The importance value can be set as a weight ranging from 0 to 1. The urgency of the task, such as equipment failure and accident risk, can be evaluated by the time constraint of the task. The time required for the inspection task is usually determined by the scope and complexity of the task. By calculating the priority of tasks by weight, all tasks are sorted according to the calculated priority. Tasks with higher priority values are considered more important and will be scheduled first. Combining geographic information and the robot's status, a task scheduling weight factor is defined to reflect the difficulty of executing each task in the current environment. The scheduling weight factor can be weighted based on the following factors: Distance weight: The distance between the task and the robot's current position. The farther the task, the more difficult it is to schedule, and the higher the weight factor. Terrain influence: Terrain elevation affects the difficulty of inspection tasks. If the task needs to be performed in a higher or steeper place, the difficulty weight can be increased; Obstacle impact: The presence of obstacles will increase the complexity of task execution. If the task path passes through an obstacle area, the impact weight of the obstacle will be increased; Robot status: The robot's status also has a significant impact on task scheduling. For example, if the robot's battery is low or the remaining time for a task is short, tasks that are closer or easier to complete should be prioritized. Based on each task's priority and scheduling weight factor, each task is assigned a final scheduling value.

[0030] See Figure 7 As shown in the figure, based on the A* algorithm and scheduling weights, the global optimal inspection route of the inspection robot is calculated and generated, specifically including: Build a set of inspection target points based on each inspection task. Each task point contains three-dimensional coordinates, scheduling weight, and time window constraints. Discretize the inspection area into topological nodes, define the connection relationship between nodes based on the robot's mobility, and construct the cost function g(n) and heuristic function h(n); Based on the A* algorithm, create an open list and a closed list, add the starting point to the open list, and set it to f(n)=g(n)+h(n); Take the node n with the smallest f(n) from the open list. If n is the end point, backtrack the path and end. Get all the reachable neighbors of n according to the movement rule. If the task weight or environment data is updated, recalculate g(n) and h(n) of the affected nodes and refresh the open list; After the iteration is completed, an ordered sequence of nodes and passing task points are output as the global optimal inspection route for the inspection robot.

[0031] Specifically, the inspection area is discretized into several topological nodes. Each node represents a discretized location with three-dimensional coordinates. Based on the robot's mobility, it is determined whether there is a reachable path between two nodes. If there is a path, they are connected. The cost of the connection is usually calculated based on distance, obstacles and other environmental factors. In the A* algorithm, we need to define two functions to guide the path search: Cost function g(n): represents the actual cost from the starting point to node n, usually the path length, time consumption, and energy consumption; Heuristic function h(n): represents the expected cost from node n to the target node, usually estimated using a heuristic algorithm (such as Manhattan distance or Euclidean distance); Obtaining all reachable neighbors of n specifically includes: If the neighbor node is already in the closed list, skip the node; If the neighbor node is not in the open list, add it to the open list and calculate its g(n) and h(n); If the neighbor node is already in the open list, check whether a lower g(n) value can be obtained through the current node. If so, update g(n) and recalculate its f(n); After reaching the target node, trace back from the target node to the starting point, record each node in the path and the task points passed through, and form the global optimal inspection route.

[0032] Furthermore, the method according to the embodiment of the present application can also be used with the aid of Figure 8 The electronic device architecture shown in FIG. Figure 8 As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, a read-only memory (ROM) 503, a random access memory (RAM) 504, a communication port 505 connected to a network, an input / output component 506, a hard disk 507, etc. The storage device in the electronic device 500, such as the ROM 503 or the hard disk 507, may store the efficient inspection control method and system for the inspection robot provided in this application. The electronic device 500 may also include a terminal interface 508. Of course, Figure 8 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 8 One or more components of an electronic device are shown.

[0033] Figure 9 This is a schematic diagram of the computer-readable storage medium structure provided by an embodiment of the present application. Figure 9 , a computer-readable storage medium 600 according to one embodiment of the present application is shown. Computer-readable instructions are stored on the computer-readable storage medium 600. When the computer-readable instructions are executed by a processor, the efficient inspection control method and system for an inspection robot according to an embodiment of the present application described with reference to the above figures can be executed. The storage medium 600 includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory. Non-volatile memory may include, for example, read-only memory (ROM), a hard disk, a flash memory, etc.

[0034] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0035] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0036] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An efficient inspection control method for an inspection robot, characterized in that: include: Based on the 3D point cloud data of the inspection area acquired by the radar sensor, a digital 3D geographic model of the inspection area is constructed, which includes terrain elevation, fixed equipment coordinates, and obstacle distribution. Based on the historical movement trajectories of mobile obstacles in the inspection area, a trajectory prediction model based on time and space constraints is established to predict the movement trajectory of mobile obstacles within the next 5 seconds; Based on the acquired motion trajectory data, the dynamically updated environment database periodically refreshes the spatial coordinates and predicted trajectory of the moving object; Based on the digital 3D geographic model and environmental database data, each inspection task is assigned a scheduling weight according to the inspection task priority and equipment status data; Based on the A* algorithm and scheduling weights, the global optimal inspection route of the inspection robot is calculated and generated.

2. The efficient inspection control method for an inspection robot according to claim 1, characterized in that: The construction of a digital three-dimensional geographic model of the inspection area including terrain elevation, fixed equipment coordinates, and obstacle distribution based on the three-dimensional point cloud data of the inspection area acquired by the radar sensor specifically includes: Use radar sensors to scan the inspection area, obtain high-precision three-dimensional point cloud data, and process and register the point cloud data; The ground segmentation algorithm is used to separate ground points from non-ground points, extract the terrain surface, and interpolate the ground point cloud data to generate a continuous elevation data model; Segment fixed equipment based on point cloud data and generate a 3D coordinate position dataset of the equipment; By comparing time-series point cloud data, static obstacles are identified based on elevation differences or geometric features, and their location coordinates and dimensions are marked; Based on the above data, a digital three-dimensional geographic model of the inspection area is constructed, which includes terrain elevation, fixed equipment coordinates and obstacle distribution.

3. The efficient inspection control method for an inspection robot according to claim 1, characterized in that: The method of establishing a trajectory prediction model based on time and space constraints based on the historical movement trajectory of the mobile obstacle in the inspection area and predicting the movement trajectory of the mobile obstacle within the next 5 seconds specifically includes: Based on the processed time-series point cloud data, the historical motion trajectory data of mobile obstacles in the inspection area is obtained; Based on historical motion trajectory data, extract kinematic spatiotemporal features, including velocity, acceleration, and direction angle, and obtain spatiotemporal constraint features; Input the historical motion trajectory feature data into the LSTM-based neural network model, train it with spatiotemporal constraint features, and obtain the trained trajectory prediction model; The real-time time-series point cloud data is input into the trajectory prediction model to predict the movement trajectory data of the moving obstacle in the next 5 seconds.

4. The efficient inspection control method for an inspection robot according to claim 3, characterized in that: The extraction of kinematic spatiotemporal features based on historical motion trajectory data, including velocity, acceleration, and direction angle, and the acquisition of spatiotemporal constraint features specifically include: The loss function is used to penalize excessive acceleration, limiting the acceleration of the predicted trajectory points to no more than the threshold; The curvature penalty term is used to limit the turning radius of the obstacle. Based on the environmental data in the digitized 3D geographic model, a binary obstacle mask is defined to prevent the predicted trajectory from entering the obstacle area.

5. The efficient inspection control method for an inspection robot according to claim 1, characterized in that: The dynamically updated environment database based on the acquired motion trajectory data and periodically refreshing the spatial coordinates and predicted trajectory of the moving object specifically includes: Based on the predicted movement trajectory data of moving obstacles in the next 5 seconds, the movement status of objects in the inspection area is updated in real time; Based on a fixed time interval, the latest coordinates of each moving obstacle are written into the database and a timestamp is added to each obstacle. Real-time monitoring of whether the predicted trajectory overlaps with static obstacles or other moving object trajectories. If a conflict exists, trajectory replanning is triggered and the prediction result is updated; Based on the predicted motion trajectory data and the actual motion trajectory data, the prediction error is calculated, and the model is optimized based on incremental learning.

6. The efficient inspection control method for an inspection robot according to claim 1, characterized in that: The method of adding a scheduling weight to each inspection task based on the digital three-dimensional geographic model and environmental database data and according to the inspection task priority and equipment status data specifically includes: Add priorities to inspection tasks based on their attributes; Based on the terrain elevation, fixed equipment coordinates and obstacle distribution data in the three-dimensional geographic model and the status parameter data of the inspection robot, a scheduling weight factor is added to each inspection task.

7. The efficient inspection control method for an inspection robot according to claim 1, characterized in that: The calculation and generation of the global optimal inspection route of the inspection robot based on the A* algorithm and the scheduling weight specifically includes: Build a set of inspection target points based on each inspection task. Each task point contains three-dimensional coordinates, scheduling weight, and time window constraints. Discretize the inspection area into topological nodes, define the connection relationship between nodes based on the robot's mobility, and construct the cost function g(n) and heuristic function h(n); Based on the A* algorithm, create an open list and a closed list, add the starting point to the open list, and set it to f(n)=g(n)+h(n); Take the node n with the smallest f(n) from the open list. If n is the end point, backtrack the path and end. Get all the reachable neighbors of n according to the movement rule. If the task weight or environment data is updated, recalculate g(n) and h(n) of the affected nodes and refresh the open list; After the iteration is completed, an ordered sequence of nodes and passing task points are output as the global optimal inspection route for the inspection robot.

8. In combination with an efficient inspection control system for an inspection robot, the method for implementing the efficient inspection control method for an inspection robot according to any one of claims 1 to 7 is characterized in that: include: Data processing module: The data processing module is mainly used to obtain three-dimensional point cloud data from the radar sensor and generate a digital three-dimensional geographic model, including terrain elevation, fixed equipment coordinates and obstacle distribution; Trajectory prediction module: The trajectory prediction module is mainly used to predict the movement trajectory of the moving obstacle within the next 5 seconds by combining historical movement trajectory data with the LSTM neural network model; Environmental database update module: The environmental database update module is mainly used to dynamically update the environmental database in real time, refresh the spatial coordinates and predicted trajectory of moving objects, and ensure timely and accurate information; Task scheduling module: The task scheduling module is mainly used to assign scheduling weights to each inspection task according to the priority of the inspection task and the equipment status data, and optimize the task execution order; Path planning module: The path planning module is mainly used to calculate and generate the global optimal inspection route of the inspection robot based on the A* algorithm, combined with task scheduling weights and environmental data; Task execution monitoring module: The task execution monitoring module is mainly used to monitor the task execution status of the inspection robot in real time, and dynamically adjust the path planning and task priority according to the progress of the task; Processor: The processor is mainly used for the calculation process of each formula and the construction calculation process of each model.

9. An electronic device, characterized in that: include: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the efficient inspection control method for an inspection robot as described in any one of claims 1-7.

10. A computer-readable storage medium storing computer-readable instructions, characterized in that: When the computer-readable instructions are executed by a processor, the efficient inspection control method for an inspection robot according to any one of claims 1 to 7 is implemented.

Images