Self-adaptive control method and system for unmanned forcible entry robot based on environment perception

Abstract

The invention discloses a self-adaptive control method and system for an unmanned forcible entry robot based on environmental perception, and relates to the related field of robot control, and the method comprises the steps: dynamically obtaining the target information of a target region through a perception device group; performing clustering analysis on the target environment information by taking the target IMU information as a constraint to obtain a target clustering result, and extracting a first clustering cluster; introducing a step-by-step fusion mechanism to perform fusion analysis on the environment information in the first cluster to obtain first sensing information; based on a first mapping relation between the first IMU information and the first perception information, constructing a target environment visual graph of the target area; and performing adaptive control on the target unmanned forcible entry robot according to the target environment visual graph after the rear-end closed-loop optimization. The technical problem that due to the fact that an existing unmanned forcible entry robot is poor in environmental adaptability during control, the working capacity and precision in the complex environment are insufficient is solved, and the technical effects that the environmental adaptability of the robot is enhanced, and the working capacity and precision in the complex environment are improved are achieved.

Description

Technical Field

[0001] This application relates to the field of robot control, and in particular, to an adaptive control method and system for an unmanned demolition robot based on environmental perception. Background Art

[0002] The control of an unmanned demolition robot based on environmental perception is used to solve problems such as autonomous navigation, precise demolition, and environmental adaptation of the unmanned demolition robot when operating in a complex and changeable environment. Existing navigation and rule-based decision-making methods cannot comprehensively and accurately obtain environmental information, resulting in the robot being difficult to adapt to the complex and changeable environment, with low operation efficiency or failure, which limits the operation ability and accuracy of the robot.

[0003] In the current related technologies, there are technical problems in the control of unmanned demolition robots, such as poor environmental adaptability, resulting in insufficient operation ability and accuracy in complex environments. Summary of the Invention

[0004] By providing an adaptive control method and system for an unmanned demolition robot based on environmental perception, this application uses a perception device group to dynamically obtain target information of the target area, including target IMU information and target environmental information, performs clustering analysis on the target environmental information with the target IMU information as a constraint, introduces a hierarchical fusion mechanism to perform fusion analysis on the environmental information, constructs an environmental view of the target area, and adaptively controls the robot according to the environmental view of the target area optimized by the backend closed-loop, etc. technical means, achieving the technical effect of enhancing the environmental adaptability of the robot and improving the operation ability and accuracy in complex environments.

[0005] This application provides an adaptive control method for an unmanned demolition robot based on environmental perception, including: dynamically obtaining target information of the target area through a perception device group, where the target information includes target IMU information and target environmental information, and the perception device group is carried on the target unmanned demolition robot; performing clustering analysis on the target environmental information with the target IMU information as a constraint to obtain a target clustering result, and extracting the first clustering cluster corresponding to the first IMU information in the target clustering result; introducing a hierarchical fusion mechanism to perform fusion analysis on the environmental information in the first clustering cluster to obtain the first perception information corresponding to the first IMU information; constructing an environmental view of the target area based on the first mapping relationship between the first IMU information and the first perception information; and adaptively controlling the target unmanned demolition robot according to the environmental view of the target area optimized by the backend closed-loop.

[0006] In a possible implementation, the following processing is performed: The perception device group includes an IMU sensor, an image collector, a lidar, a gas sensor, and an electronic skin sensor; the target acceleration and target angular velocity of the target unmanned demolition robot are dynamically monitored through the IMU sensor, and the target IMU information is formed; the target environmental image is dynamically monitored through the image collector; the target point cloud data is dynamically monitored through the lidar; the target harmful gas concentration is dynamically monitored through the gas sensor; the target temperature, target humidity, and target pressure are dynamically monitored through the electronic skin sensor, and the target perception data is formed; the target environmental information is formed based on the target environmental image, the target point cloud data, the target harmful gas concentration, and the target perception data; the target IMU information and the target environmental information form the target information.

[0007] In a possible implementation, clustering analysis is performed on the target environmental information with the target IMU information as a constraint to obtain a target clustering result, and the first clustering cluster corresponding to the first IMU information in the target clustering result is extracted, and the following processing is performed: The first environmental information at the first time and the second environmental information at the second time in the target environmental information are randomly extracted; it is determined whether the second IMU information at the second time conforms to a predetermined condition, where the first IMU information refers to the IMU information at the first time; if it conforms to the predetermined condition, the first environmental information and the second environmental information are clustered into the first clustering cluster.

[0008] In a possible implementation, a hierarchical fusion mechanism is introduced to perform fusion analysis on the environmental information in the first clustering cluster to obtain the first perception information corresponding to the first IMU information, and the following processing is performed: Any device in the perception device group is obtained; the first arbitrary monitoring data set of the arbitrary device is matched in the first clustering cluster, and the mean value of the first arbitrary monitoring data set is taken, denoted as the first arbitrary monitoring value; the predetermined arbitrary influence structure of the arbitrary device is retrieved, and sampling aggregation is performed on the predetermined arbitrary influence structure to obtain a first arbitrary influence factor; according to the hierarchical fusion mechanism, the first arbitrary influence factor and the first arbitrary monitoring value are combined to obtain a primary fusion value; according to the hierarchical fusion mechanism, the arbitrary sensing weight of the arbitrary device is obtained, and the primary fusion value is fused to obtain a secondary fusion value; the first perception information is formed according to the corresponding relationship between the arbitrary device and the secondary fusion value.

[0009] In a possible implementation, the following processing is performed: form a set of influencing factors for any of the devices; extract a first factor from the set of influencing factors, and the first factor has an identifier of a first weight; use the first factor as a vertex and the first weight as a connecting line to form an arbitrary undirected graph of any of the devices; use the arbitrary undirected graph as the predetermined arbitrary influence structure.

[0010] In a possible implementation, the following processing is performed: obtain the historical sensing records of the devices of the same type as any of the devices; use any historical sensing data of any of the devices in the historical sensing records as the dependent variable; use the historical data of the first factor corresponding to the first factor as the independent variable; perform a correlation analysis on the independent variable and the dependent variable to obtain a correlation index, and use the correlation index as the first weight.

[0011] In a possible implementation, based on the first mapping relationship between the first IMU information and the first sensing information, a target environment view of the target area is constructed. The following processing is performed: obtain a target pose neighborhood of the first IMU information, and the target pose neighborhood includes any neighborhood IMU information; obtain any sensing information corresponding to the any neighborhood IMU information, and perform multi-hop fusion of the any sensing information with the first sensing information to obtain first target sensing information; use the first target sensing information to replace the first sensing information to obtain the target environment view.

[0012] This application also provides an adaptive control system for an unmanned demolition robot based on environmental perception, including: a target information acquisition module, configured to dynamically acquire target information of a target area through a sensing device group, where the target information includes target IMU information and target environment information, and the sensing device group is carried on a target unmanned demolition robot; a clustering analysis module, configured to perform clustering analysis on the target environment information with the target IMU information as a constraint to obtain a target clustering result, and extract a first clustering cluster corresponding to the first IMU information in the target clustering result; a first sensing information acquisition module, configured to introduce a hierarchical fusion mechanism to perform fusion analysis on the environmental information in the first clustering cluster to obtain first sensing information corresponding to the first IMU information; a target environment view construction module, configured to construct a target environment view of the target area based on the first mapping relationship between the first IMU information and the first sensing information; an adaptive control module, configured to perform adaptive control on the target unmanned demolition robot according to the target environment view after backend closed-loop optimization.

[0013] The adaptive control method and system for an unmanned demolition robot based on environmental perception proposed in this application first dynamically obtain the target information of the target area through a perception device group. Among them, the target information includes target IMU information and target environmental information, and the perception device group is carried on the target unmanned demolition robot. Then, clustering analysis is performed on the target environmental information with the target IMU information as a constraint to obtain a target clustering result, and the first clustering cluster corresponding to the first IMU information in the target clustering result is extracted. Next, a hierarchical fusion mechanism is introduced to perform fusion analysis on the environmental information in the first clustering cluster to obtain the first perception information corresponding to the first IMU information. Then, based on the first mapping relationship between the first IMU information and the first perception information, a target environmental view of the target area is constructed. Finally, the target unmanned demolition robot is adaptively controlled according to the target environmental view optimized by the backend closed loop. The technical effect of enhancing the robot's environmental adaptability and improving the operation ability and accuracy in complex environments is achieved. Description of the Drawings

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings of the embodiments of the present invention will be briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the application. It should be understood that the operations in the front or below do not necessarily need to be executed precisely in sequence. On the contrary, according to needs, they can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several operations can be removed from these processes.

[0015] Figure 1 It is a schematic flowchart of the adaptive control method for an unmanned demolition robot based on environmental perception provided by an embodiment of this application.

[0016] Figure 2 It is a schematic structural diagram of the adaptive control system for an unmanned demolition robot based on environmental perception provided by an embodiment of this application.

[0017] Description of the reference numerals: Target information acquisition module 10, clustering analysis module 20, first perception information acquisition module 30, target environmental view construction module 40, adaptive control module 50. Detailed Embodiments

[0018] The above description is only an overview of the technical solutions of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the description. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the detailed embodiments of this application are specifically given below.

[0019] In order to make the objectives, technical solutions, and advantages of this application clearer, the following will further describe this application in detail with reference to the accompanying drawings. The described embodiments should not be construed as limitations on this application. All other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.

[0020] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict. The terms "first / second" involved are only used to distinguish similar objects and do not represent a specific order for the objects. The terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application.

[0021] An adaptive control method for an unmanned demolition robot based on environmental perception is provided in an embodiment of this application, as Figure 1 shown. The method includes: Step S100: Dynamically obtain target information of a target area through a perception device group, where the target information includes target IMU information and target environmental information, and the perception device group is mounted on a target unmanned demolition robot.

[0022] Specifically, the perception device group includes a camera, a lidar (LiDAR), an inertial measurement unit (IMU), a depth sensor, etc., which are integrated on the target unmanned demolition robot. During the movement of the target unmanned demolition robot, these perception devices continuously collect data. For example, the camera captures images, the lidar generates point cloud data, and the IMU provides information about the robot's attitude, acceleration, and angular velocity. Read data from the interfaces of each perception device to obtain the target IMU information and the target environmental information, where the target IMU information includes acceleration, angular velocity, attitude, etc.; the target environmental information includes obstacle positions, surface types, lighting conditions, etc.

[0023] In a possible implementation, step S100 further includes step S110, and the group of sensing devices includes an IMU sensor, an image collector, a lidar, a gas sensor, and an electronic skin sensor. Specifically, each sensor in the group of sensing devices is connected to the control system of the target unmanned demolition robot through a specific interface. The control system regularly or on demand reads data from each sensor and stores this data in an appropriate data structure. Among them, the IMU sensor is used to measure the motion state information of the target unmanned demolition robot, such as acceleration, angular velocity, and attitude. The image collector, such as a camera, is used to capture the image information of the target area to understand the objects, obstacles, etc. in the environment. The lidar emits laser light and receives the reflected signal to generate the point cloud data of the target area, which is used to construct a three-dimensional model of the environment. The gas sensor is used to detect the concentration of harmful gases in the target area, such as methane, carbon monoxide, etc., to ensure the safety of the robot operation. The electronic skin sensor mimics the sensing function of the human skin and can monitor physical parameters such as the temperature, humidity, and pressure of the environment.

[0024] Step S120, dynamically monitor the target acceleration and target angular velocity of the target unmanned demolition robot through the IMU sensor, and form the target IMU information. Specifically, the IMU sensor continuously monitors the motion state of the target unmanned demolition robot and outputs acceleration and angular velocity data. The control system receives this data and combines it into the target IMU information, which is stored in a data structure, such as an array or a structure.

[0025] Step S130, dynamically monitor the target environment image through the image collector. Specifically, the image collector (such as a camera) captures the image of the target area. The control system receives the image data and preprocesses it using an image processing library (such as OpenCV), such as denoising, enhancing the contrast, etc.

[0026] Step S140, dynamically monitor the target point cloud data through the lidar. Specifically, the lidar emits laser beams and receives the reflected signals to generate the point cloud data of the target area. The control system receives the point cloud data and preprocesses it using a point cloud processing library (such as PCL), such as denoising, filtering, etc.

[0027] Step S150, dynamically monitor the concentration of the target harmful gas through the gas sensor. Specifically, the gas sensor detects the concentration of the harmful gas in the target area. The control system receives the gas concentration data and stores it.

[0028] Step S160, dynamically monitor the target temperature, target humidity, and target pressure through the electronic skin sensor, and form the target sensing data. Specifically, the electronic skin sensor monitors the temperature, humidity, and pressure of the environment. The control system receives this sensing data and stores it.

[0029] Step S170, compose the target environmental information based on the target environmental image, the target point cloud data, the target harmful gas concentration, and the target perception data. Specifically, the control system creates a data structure (such as a structure or a class) to store the target environmental information. The target environmental image, the target point cloud data, the target harmful gas concentration, and the target perception data are used as member variables of this data structure.

[0030] Step S180, the target IMU information and the target environmental information compose the target information. Specifically, the control system creates a data structure (such as a structure or a class) to store the target information. The target IMU information and the target environmental information are used as member variables of this data structure. This implementation method provides comprehensive and accurate environmental information for the target unmanned demolition robot through multi-sensor fusion and real-time data monitoring, improves the robot's perception ability and adaptability, and helps the robot make more informed decisions in complex environments.

[0031] Step S200, perform clustering analysis on the target environmental information with the target IMU information as a constraint to obtain a target clustering result, and extract the first clustering cluster corresponding to the first IMU information in the target clustering result.

[0032] Specifically, use a clustering algorithm (such as K-means, DBSCAN, etc.) to group the target environmental information so that information with high similarity is grouped into the same group. During the clustering process, consider the robot's motion state provided by the IMU information, such as changes in acceleration and angular velocity, and use the IMU information as a constraint condition in the clustering process by weighting or adjusting the distance metric to ensure that the clustering result conforms to the actual motion situation of the robot. For example, when the IMU data shows that the robot is moving rapidly, the environmental information obtained at similar time points is grouped into one category.

[0033] In a possible implementation manner, when performing clustering analysis on the target environmental information with the target IMU information as a constraint to obtain a target clustering result and extracting the first clustering cluster corresponding to the first IMU information in the target clustering result, step S200 further includes step S210, randomly extract the first environmental information at the first time and the second environmental information at the second time from the target environmental information. Specifically, randomly select environmental information at two different time points from the target environmental information database through a random number generator, and denote them as the first environmental information (corresponding to the first time) and the second environmental information (corresponding to the second time). These environmental information includes images, point cloud data, harmful gas concentration, and perception data (temperature, humidity, pressure), etc.

[0034] Step S220: Determine whether the second IMU information at the second time conforms to a predetermined condition, where the first IMU information refers to the IMU information at the first time. Specifically, through mathematical operations such as calculating the Euclidean distance or Manhattan distance between two IMU information vectors, compare whether the IMU information at the second time (i.e., the second IMU information) and the IMU information at the first time (i.e., the first IMU information) meet the predetermined condition. The predetermined condition is a threshold set in advance, used to determine whether two IMU information are close enough to consider that the corresponding environmental information is similar. For example, the difference in acceleration and angular velocity is within a certain threshold range.

[0035] Step S230: If it conforms to the predetermined condition, cluster the first environmental information and the second environmental information into the first cluster. Specifically, the clustering operation can be implemented by creating a new data structure in memory, which is used to store the environmental information clustered together. If the second IMU information conforms to the predetermined condition with the first IMU information, cluster the first environmental information and the second environmental information into a cluster, denoted as the first cluster. That is, the environmental information perceived under similar motion states is considered similar and can be grouped into one category. This implementation method can exclude the environmental information differences caused by the change of the robot's motion state by introducing the IMU information as a constraint condition, improving the accuracy of clustering.

[0036] Step S300: Introduce a hierarchical fusion mechanism to perform fusion analysis on the environmental information in the first cluster to obtain the first perception information corresponding to the first IMU information.

[0037] Specifically, the first cluster is the clustering result associated with the first IMU information. Steps such as data alignment, weighted average, and maximum likelihood estimation are used to perform hierarchical fusion on the data in the first cluster. That is, in the fusion process, first fuse the low-level information, and then use the fusion result as the basis for higher-level fusion, and finally output the fused first perception information. The first perception information is a more accurate set of environmental features. For example, the data from different lidar scans can be fused first, and then these fusion results can be fused with the camera data.

[0038] In a possible implementation, introduce a hierarchical fusion mechanism to perform fusion analysis on the environmental information in the first cluster to obtain the first perception information corresponding to the first IMU information. Step S300 further includes step S310: Obtain any device in the perception device group. Specifically, a traversal algorithm is used to select a device from the list of the perception device group (which stores the identifiers and related information of all perception devices) as the device currently being processed.

[0039] Step S320: Match the first arbitrary monitoring data set of the arbitrary device in the first clustering cluster, and take the mean value of the first arbitrary monitoring data set, denoted as the first arbitrary monitoring value. Specifically, the first clustering cluster stores environmental information that meets specific IMU information constraint conditions. In the first clustering cluster, search for all monitoring data sets corresponding to the current device (such as image data, point cloud data, harmful gas concentration data, temperature / humidity / pressure data, etc.), and calculate the mean value of these data as the first arbitrary monitoring value of the device.

[0040] Step S330: Retrieve the predetermined arbitrary influence structure of the arbitrary device, and perform sampling aggregation on the predetermined arbitrary influence structure to obtain the first arbitrary influence factor. Specifically, retrieve the predetermined influence structure related to the device (the predetermined influence structure can be a parameter set, a mathematical model, or a function used to describe the relationship between device monitoring data and environmental information), and perform sampling and aggregation operations on these influence structures, including steps such as data extraction, weighted averaging, and model prediction, to obtain the first arbitrary influence factor.

[0041] Step S340: According to the hierarchical fusion mechanism, combine the first arbitrary influence factor and the first arbitrary monitoring value to obtain the first-level fusion value. Specifically, combine the first arbitrary influence factor and the first arbitrary monitoring value (which can be through multiplication, addition, division, or exponential operations, etc.) to obtain the first-level fusion value, which reflects the comprehensive result of the device monitoring data after considering the predetermined influence structure.

[0042] Step S350: Obtain the arbitrary sensing weight of the arbitrary device according to the hierarchical fusion mechanism, and fuse the first-level fusion value to obtain the second-level fusion value. Specifically, obtain the sensing weight of the current device (the sensing weight is a preset value based on factors such as the accuracy, reliability, and importance of the device, used to describe the relative importance of the device in the overall fusion process), and use this weight to fuse the first-level fusion value to obtain the second-level fusion value.

[0043] Step S360: Compose the first perception information according to the corresponding relationship between the arbitrary device and the second-level fusion value. Specifically, organize the second-level fusion values corresponding to all devices according to the device identifier (a unique identifier used to distinguish different devices) or other identification information to form the first perception information. This information is a comprehensive environmental perception result that combines the monitoring data of multiple devices and the predetermined influence structure, and can be a data structure such as an array, a list, or a dictionary, etc., used to store and represent the fused environmental information. This implementation method reduces the uncertainty and error of single-device data by combining the monitoring data of multiple devices and the predetermined influence structure, and improves the accuracy of the overall data.

[0044] In a possible implementation, step S330 further includes step S331 of forming an influencing factor set for any device. Specifically, each device in the sensing device group is traversed, and for the device currently being processed, all factors that may affect the accuracy of the monitoring data of the device are extracted from a pre-stored database or knowledge base. These factors include, but are not limited to, the physical characteristics of the device itself (such as sensitivity, precision, response time, etc.), environmental factors (such as temperature, humidity, electromagnetic interference, etc.), and operating conditions (such as the position, attitude, motion state of the device, etc.). All these factors are collected and form an influencing factor set.

[0045] Step S332 of extracting a first factor from the influencing factor set, and the first factor has an identifier of a first weight. Specifically, in the influencing factor set, each factor is extracted one by one, and the weight assigned to each factor (i.e., the first weight) is checked. Among them, the weight is determined according to the degree of influence of the factor on the accuracy of the device monitoring data and can be obtained through experimental data.

[0046] Step S333 of forming an arbitrary undirected graph of any device with the first factor as the vertex and the first weight as the connecting line. Specifically, each factor in the influencing factor set is used as a vertex of the undirected graph, and the weight of each factor is used as the connecting line (or edge) connecting these vertices to form an undirected graph representing the device influencing factors and their corresponding weight values. This undirected graph is used to evaluate the influence of different factors on the device monitoring data.

[0047] Step S334 of using the arbitrary undirected graph as the predetermined arbitrary influence structure. Specifically, after the undirected graph is constructed, it is stored as the predetermined influence structure of the device. This structure is used to evaluate the influence of different factors on the device monitoring data and adjust the processing strategy of the monitoring data accordingly. This implementation method can comprehensively consider various factors that affect the accuracy of the device monitoring data by constructing the predetermined influence structure (i.e., the undirected graph) of the device, and perform more accurate data processing and analysis based on this, improving the accuracy and reliability of the monitoring data. At the same time, by pre-constructing and storing the influence structure, the time cost of analysis and calculation is also reduced, improving the operating efficiency of the entire system.

[0048] In a possible implementation, step S332 further includes step S3321 of obtaining the historical sensing records of the same-type devices of any device. Specifically, the storage system (such as a database or a file system) is accessed to retrieve the historical sensing records of all devices of the same type as the arbitrary device currently being processed (devices similar to the arbitrary device in terms of function, model, or manufacturer, etc.). These records include the sensing data of the device at different time points and the relevant environmental factor records.

[0049] Step S3322: Use any historical sensing data of any device in the historical sensing record as the dependent variable, and use the historical data of the first factor corresponding to the first factor as the independent variable. Perform a correlation analysis on the independent variable and the dependent variable to obtain a correlation index, and use the correlation index as the first weight. Specifically, set any historical sensing data of any device as the dependent variable (Y), and set the historical data of the first factor (such as temperature, humidity, etc.) corresponding to the first factor as the independent variable (X). Ensure that the timestamps of X and Y match, that is, for each pair of data points of X and Y, they are recorded at the same time point. Use statistical methods (such as Pearson correlation coefficient, Spearman rank correlation coefficient, etc.) to calculate the correlation index between X and Y. This index measures the strength of the linear relationship between X and Y. Use the calculated correlation index as the first weight, which reflects the influence degree of the first factor on the sensing data of any device. This implementation method accurately evaluates the influence of different factors on the device sensing data through correlation analysis based on historical data, improving the accuracy of fusion analysis.

[0050] Step S400: Construct a target environment view of the target area based on the first mapping relationship between the first IMU information and the first sensing information.

[0051] Specifically, the first mapping relationship refers to the correspondence between the IMU information and the fused sensing information. For example, the timestamp in the IMU data can be associated with the feature points or object positions in the sensing information. According to the mapping relationship, integrate the first IMU information and the first sensing information into a data structure. Use graph theory algorithms (such as Dijkstra algorithm, A* search, etc.) to construct the viewable graph, where the nodes represent important positions or features in the environment, the edges represent the spatial relationships between these positions or features, and the settings of the nodes and edges are used to reflect the actual structure of the environment and the movement ability of the robot.

[0052] In a possible implementation, a target environment view of the target area is constructed based on the first mapping relationship between the first IMU information and the first perception information. Step S400 further includes step S410 of obtaining a target pose neighborhood of the first IMU information, where the target pose neighborhood includes any neighborhood IMU information. Specifically, access the storage system to retrieve historical IMU data associated with the target IMU information (i.e., the first IMU information). This data includes the position (coordinates) and attitude (orientation) information of the robot at different time points. Based on the retrieved historical IMU data, calculate the pose neighborhood of the target IMU information. The pose neighborhood refers to a set of poses that are close to the target pose (i.e., the pose represented by the first IMU information) in terms of time and space, and is obtained by analyzing the spatial proximity and temporal continuity of the pose data. According to a preset proximity criterion (such as a distance threshold, a time window, etc.), select the poses belonging to the target pose neighborhood from all historical poses. These selected poses and their corresponding IMU information constitute the target pose neighborhood.

[0053] Step S420 of obtaining any perception information corresponding to the any neighborhood IMU information and performing multi-hop fusion on the first perception information with the any perception information to obtain first target perception information. Specifically, for each neighborhood IMU information in the target pose neighborhood, retrieve the corresponding perception information. This perception information is obtained by a perception device group (such as an image collector, a lidar, etc.) at the same time point or a close time point. Use a multi-hop fusion algorithm to fuse the target perception information (i.e., the first perception information) with the neighborhood perception information. Multi-hop fusion means iteratively fusing the perception information from different poses (i.e., different time points or positions) to construct a more global and accurate perception map, including analyzing the spatial consistency, temporal continuity, and feature matching of the perception information. After multi-hop fusion, output the fused perception information, i.e., the first target perception information.

[0054] Step S430 of replacing the first perception information with the first target perception information to obtain the target environment view. Specifically, replace the first perception information in the target environment view with the first target perception information, that is, use the more global and accurate perception information to construct the target environment view. After replacing the perception information, update the target environment view, including recalculating and redrawing the nodes, edges, and features in the view to reflect the new perception information. Finally, output the updated target environment view for use in the adaptive control step. This implementation method constructs a more complete and accurate perception map by introducing the target pose neighborhood and the multi-hop fusion mechanism, fusing the perception information from different time points and positions, improving the accuracy and globality of the target environment view, and ensuring that the target unmanned demolition robot can work stably and reliably in various environments.

[0055] Step S500, adaptively control the target unmanned demolition robot according to the target environment viewable image optimized by the backend closed-loop.

[0056] Specifically, the backend closed-loop optimization is an iterative process. The SLAM (Simultaneous Localization and Mapping) algorithm is used to correct the accumulated errors by comparing the current position of the robot with the previously visited positions, and optimize the target environment viewable image. Based on the optimized viewable image, the target unmanned demolition robot can adjust its motion strategy to adapt to the changes in the environment. For example, it can avoid obstacles, select the optimal path, etc. The control command is sent to the actuator of the target unmanned demolition robot to achieve adaptive control. In the embodiment of the present application, a perception device group is carried to dynamically obtain the target information of the target area, including the target IMU information and the target environment information. The target environment information is clustered and analyzed with the target IMU information as a constraint, and a hierarchical fusion mechanism is introduced to fuse and analyze the environment information, construct the environment viewable image of the target area, and adaptively control the robot according to the environment viewable image optimized by the backend closed-loop. By using such technical means, the technical effects of enhancing the robot's environmental adaptability and improving the operation ability and accuracy in complex environments are achieved.

[0057] In the above, reference is made to Figure 1 The adaptive control method of the unmanned demolition robot based on environmental perception according to the embodiment of the present invention is described in detail. Next, reference will be made to Figure 2 Describe the adaptive control system of the unmanned demolition robot based on environmental perception according to the embodiment of the present invention.

[0058] The adaptive control system of the unmanned demolition robot based on environmental perception according to the embodiment of the present invention is used to solve the technical problems existing in the prior art, such as poor environmental adaptability, resulting in insufficient operation ability and accuracy in complex environments, and achieves the technical effects of enhancing the robot's environmental adaptability and improving the operation ability and accuracy in complex environments. The adaptive control system of the unmanned demolition robot based on environmental perception includes: a target information acquisition module 10, a clustering analysis module 20, a first perception information acquisition module 30, a target environment viewable image construction module 40, and an adaptive control module 50.

[0059] The target information acquisition module 10 is used to dynamically acquire the target information of the target area through the perception device group. Among them, the target information includes target IMU information and target environment information, and the perception device group is carried on the target unmanned demolition robot; the clustering analysis module 20 is used to perform clustering analysis on the target environment information with the target IMU information as a constraint to obtain a target clustering result, and extract the first clustering cluster corresponding to the first IMU information in the target clustering result; the first perception information acquisition module 30 is used to introduce a hierarchical fusion mechanism to perform fusion analysis on the environment information in the first clustering cluster to obtain the first perception information corresponding to the first IMU information; the target environment view construction module 40 is used to construct a target environment view of the target area based on the first mapping relationship between the first IMU information and the first perception information; the adaptive control module 50 is used to perform adaptive control on the target unmanned demolition robot according to the target environment view after closed-loop optimization at the backend.

[0060] Next, the specific configuration of the target information acquisition module 10 will be described in detail. As described above, the target information acquisition module 10 may further include: a perception device group building unit for building a perception device group, where the perception device group includes an IMU sensor, an image collector, a lidar, a gas sensor, and an electronic skin sensor; a target IMU information acquisition unit for dynamically monitoring the target acceleration and target angular velocity of the target unmanned demolition robot through the IMU sensor and forming the target IMU information; a target environment image acquisition unit for dynamically monitoring a target environment image through the image collector; a target point cloud data acquisition unit for dynamically monitoring target point cloud data through the lidar; a target harmful gas concentration acquisition unit for dynamically monitoring the target harmful gas concentration through the gas sensor; a target perception data acquisition unit for dynamically monitoring the target temperature, target humidity, and target pressure through the electronic skin sensor and forming target perception data; a target environment information integration unit for forming the target environment information based on the target environment image, the target point cloud data, the target harmful gas concentration, and the target perception data; a target information integration unit for forming the target information from the target IMU information and the target environment information.

[0061] Next, the specific configuration of the clustering analysis module 20 will be described in detail. As described above, the target environmental information is clustered and analyzed with the target IMU information as a constraint to obtain a target clustering result, and the first clustering cluster corresponding to the first IMU information in the target clustering result is extracted. The clustering analysis module 20 may further include: an environmental information extraction unit for randomly extracting the first environmental information at the first time and the second environmental information at the second time from the target environmental information; a judgment unit for judging whether the second IMU information at the second time and the first IMU information meet a predetermined condition, where the first IMU information refers to the IMU information at the first time; a clustering unit for clustering the first environmental information and the second environmental information into the first clustering cluster if they meet the predetermined condition.

[0062] Next, the specific configuration of the first perception information acquisition module 30 will be described in detail. As described above, a hierarchical fusion mechanism is introduced to perform fusion analysis on the environmental information in the first clustering cluster to obtain the first perception information corresponding to the first IMU information. The first perception information acquisition module 30 may further include: an arbitrary device acquisition unit for acquiring any device in the perception device group; a first arbitrary monitoring value acquisition unit for matching the first arbitrary monitoring data set of the arbitrary device in the first clustering cluster and taking the mean of the first arbitrary monitoring data set, denoted as the first arbitrary monitoring value; a sampling aggregation unit for retrieving the predetermined arbitrary influence structure of the arbitrary device and performing sampling aggregation on the predetermined arbitrary influence structure to obtain a first arbitrary influence factor; a hierarchical fusion unit for obtaining a first-level fusion value by combining the first arbitrary influence factor and the first arbitrary monitoring value according to the hierarchical fusion mechanism, obtaining the arbitrary sensing weight of the arbitrary device according to the hierarchical fusion mechanism, and performing fusion on the first-level fusion value to obtain a second-level fusion value; a first perception information formation unit for forming the first perception information according to the corresponding relationship between the arbitrary device and the second-level fusion value.

[0063] Among them, the sampling aggregation unit may further include: an influence factor set formation subunit for forming the influence factor set of the arbitrary device; a first factor extraction subunit for extracting the first factor in the influence factor set, and the first factor has an identifier of the first weight; an arbitrary undirected graph generation subunit for forming an arbitrary undirected graph of the arbitrary device with the first factor as the vertex and the first weight as the connection line; an arbitrary influence structure predetermination subunit for using the arbitrary undirected graph as the predetermined arbitrary influence structure.

[0064] Among them, the first factor extraction subunit may further include: a historical sensing record acquisition component configured to acquire historical sensing records of the same type of devices as the any device; a correlation analysis component configured to use any historical sensing data of the any device in the historical sensing records as the dependent variable and the historical data of the first factor corresponding to the first factor as the independent variable, perform a correlation analysis on the independent variable and the dependent variable to obtain a correlation index, and use the correlation index as the first weight.

[0065] Next, the specific configuration of the target environment view construction module 40 will be described in detail. As described above, based on the first mapping relationship between the first IMU information and the first perception information, the target environment view of the target area is constructed. The target environment view construction module 40 may further include: a target pose neighborhood acquisition unit configured to acquire a target pose neighborhood of the first IMU information, where the target pose neighborhood includes any neighborhood IMU information; a multi-hop fusion unit configured to acquire any perception information corresponding to the any neighborhood IMU information, and perform multi-hop fusion on the first perception information with the any perception information to obtain first target perception information; a target environment view acquisition unit configured to replace the first perception information with the first target perception information to obtain the target environment view.

[0066] The adaptive control system for an unmanned demolition robot based on environmental perception provided in the embodiments of the present invention can execute the method for adaptive control of an unmanned demolition robot based on environmental perception provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0067] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The included various units and modules are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of the functional units are only for the convenience of mutual distinction and do not limit the protection scope of the present invention.

[0068] The above specific embodiments do not constitute a limitation on the protection scope of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the protection scope of this application. In some cases, the actions or steps recited in this application can be executed in an order different from that in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or consecutive order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. An adaptive control method for an unmanned demolition robot based on environmental perception, characterized in that: include: Dynamically obtain target information of the target area through a sensing device group, wherein the target information includes target IMU information and target environment information, and the sensing device group is mounted on the target unmanned demolition robot; Performing cluster analysis on the target environment information with the target IMU information as a constraint to obtain a target clustering result, and extracting a first cluster corresponding to the first IMU information in the target clustering result; Introducing a step-by-step fusion mechanism to perform fusion analysis on the environmental information in the first cluster to obtain first perception information corresponding to the first IMU information; Constructing a target environment visual graph of the target area based on a first mapping relationship between the first IMU information and the first perception information; The target unmanned demolition robot is adaptively controlled according to the target environment visual map after back-end closed-loop optimization.

2. The method for adaptive control of an unmanned demolition robot based on environmental perception according to claim 1 is characterized in that: The sensing device group includes an IMU sensor, an image collector, a laser radar, a gas sensor and an electronic skin sensor; The target acceleration and target angular velocity of the target unmanned demolition robot are obtained by dynamic monitoring of the IMU sensor, and constitute the target IMU information; Dynamically monitor and obtain the target environment image through the image collector; Obtaining target point cloud data through dynamic monitoring of the laser radar; Dynamically monitor the target harmful gas concentration through the gas sensor; Dynamically monitoring the target temperature, target humidity and target pressure through the electronic skin sensor and forming target perception data; The target environment information is formed based on the target environment image, the target point cloud data, the target harmful gas concentration and the target perception data; The target IMU information and the target environment information constitute the target information.

3. The adaptive control method of the unmanned demolition robot based on environmental perception according to claim 1 is characterized in that: The target environment information is clustered and analyzed with the target IMU information as a constraint to obtain a target clustering result, and a first cluster corresponding to the first IMU information in the target clustering result is extracted, including: Randomly extracting first environmental information at a first time and second environmental information at a second time from the target environmental information; Determine whether the second IMU information at the second time and the first IMU information meet a predetermined condition, wherein the first IMU information refers to the IMU information at the first time; If the predetermined condition is met, the first environment information and the second environment information are clustered into the first cluster.

4. The method for adaptive control of an unmanned demolition robot based on environmental perception according to claim 3 is characterized in that: Introducing a step-by-step fusion mechanism to perform fusion analysis on the environmental information in the first cluster to obtain first perception information corresponding to the first IMU information, including: Obtain any device in the sensing device group; Matching a first arbitrary monitoring data set of the arbitrary device in the first cluster, and taking a mean value of the first arbitrary monitoring data set, recorded as a first arbitrary monitoring value; Retrieving a predetermined arbitrary impact structure of the arbitrary device, and performing sampling and aggregation on the predetermined arbitrary impact structure to obtain a first arbitrary impact factor; According to the step-by-step fusion mechanism, combining the first arbitrary influencing factor and the first arbitrary monitoring value to obtain a first-level fusion value; Obtaining any sensor weight of any device according to the step-by-step fusion mechanism, and fusing the first-level fusion value to obtain a second-level fusion value; The first perception information is composed according to the correspondence between the arbitrary device and the secondary fusion value.

5. The method for adaptive control of an unmanned demolition robot based on environmental perception according to claim 4 is characterized in that: include: Build a set of influencing factors of the arbitrary device; Extracting a first factor from the influencing factor set, wherein the first factor has an identifier of a first weight; Using the first factor as a vertex and the first weight as a connecting line, an arbitrary undirected graph of the arbitrary device is formed; The arbitrary undirected graph is used as the predetermined arbitrary influence structure.

6. The method for adaptive control of an unmanned demolition robot based on environmental perception according to claim 5 is characterized in that: include: Obtain historical sensing records of similar devices of the arbitrary device; Taking any historical sensor data of any device in the historical sensor record as a dependent variable; The first factor historical data corresponding to the first factor is used as an independent variable; A correlation analysis is performed on the independent variable and the dependent variable to obtain a correlation index, and the correlation index is used as the first weight.

7. The method for adaptive control of an unmanned demolition robot based on environmental perception according to claim 1 is characterized in that: Constructing a target environment visual map of the target area based on a first mapping relationship between the first IMU information and the first perception information, including: Acquire a target pose neighborhood of the first IMU information, wherein the target pose neighborhood includes any neighborhood IMU information; Acquire any perception information corresponding to the arbitrary neighborhood IMU information, and perform multi-hop fusion on the first perception information with the arbitrary perception information to obtain first target perception information; The first perception information is replaced by the first target perception information to obtain the target environment visual map.

8. The adaptive control system of unmanned demolition robot based on environmental perception is characterized by: The system is used to implement the method for adaptive control of an unmanned demolition robot based on environmental perception according to any one of claims 1 to 7, and the system comprises: A target information acquisition module, used to dynamically acquire target information of a target area through a sensing device group, wherein the target information includes target IMU information and target environment information, and the sensing device group is mounted on a target unmanned demolition robot; A cluster analysis module, used to perform cluster analysis on the target environment information with the target IMU information as a constraint, obtain a target clustering result, and extract a first cluster corresponding to the first IMU information in the target clustering result; A first perception information acquisition module, used to introduce a step-by-step fusion mechanism to perform fusion analysis on the environmental information in the first cluster to obtain first perception information corresponding to the first IMU information; A target environment visible graph construction module, configured to construct a target environment visible graph of the target area based on a first mapping relationship between the first IMU information and the first perception information; The adaptive control module is used to adaptively control the target unmanned demolition robot according to the target environment visual map after back-end closed-loop optimization.

Images