Intelligent paying-off robot regulation and control system and method based on perceptual obstacle avoidance

Abstract

The invention discloses an intelligent paying-off robot regulation and control system and method based on perceptual obstacle avoidance, and relates to the technical field of intelligent robot control, and the method comprises the steps: collecting the historical operation data of a robot in a plurality of operation periods, and constructing an environment optimization model; before the task starts, a plurality of alternative paths with different risk characteristics are pre-generated in combination with the current operation time and the environmental risk map, the comprehensive score of each alternative path is calculated, and the calculation and storage resources of the sensing system are distributed; in the task execution process, operation data of the robot are collected in real time and compared with prediction data of the environment optimization model in real time; when the deviation exceeds a preset threshold value, dynamic path re-planning is triggered; when a risk threshold value in the health early warning matrix is reached, triggering a grading early warning mechanism; and after the task is finished, updating the environment risk map and the health early warning matrix, and dynamically adjusting the path confidence coefficient weight and the resource configuration strategy of each region.

Description

Technical Field

[0001] This invention relates to the field of intelligent robot control technology, specifically to an intelligent wire-laying robot control system and method based on perception and obstacle avoidance. Background Technology

[0002] In fields such as transportation engineering and construction, automated line laying operations are crucial for improving efficiency and accuracy. Traditional line laying robots either rely on preset fixed paths to perform tasks or only possess basic real-time obstacle avoidance capabilities. The former cannot adapt to dynamically changing working environments, while the latter typically employs a simple "perception-response" control logic, lacking the understanding and utilization of environmental patterns. Existing technical solutions mostly rely on static maps or real-time perception for path planning, lacking in-depth mining and learning of historical operational data, and are unable to predict environmental risk changes, resulting in a lack of foresight in path selection. Furthermore, perception and computing resources are usually allocated using uniform or fixed strategies, which can easily lead to resource waste or insufficient capture of key information in complex and changing environments. The correlation between the robot's health status, operational behavior, and environmental load has not been systematically modeled and analyzed, making maintenance mostly a reactive response after the fact, affecting operational continuity and equipment reliability. In existing technical solutions, most systems lack a closed-loop mechanism that continuously learns from execution results and optimizes its own decision-making model, making it difficult for the level of intelligence to evolve over time. Summary of the Invention

[0003] The purpose of this invention is to provide a control system and method for an intelligent wire-laying robot based on perception and obstacle avoidance, so as to solve the problems raised in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a control method for an intelligent wire-laying robot based on perception and obstacle avoidance, the method comprising: S100. Collect historical operating data of the robot over multiple work cycles, and construct an environmental optimization model based on the historical operating data. The environmental optimization model includes an environmental risk map reflecting the spatiotemporal distribution characteristics of obstacles, and a health warning matrix showing the correlation between operating data, energy consumption data, and fault data. Specifically, constructing the environmental optimization model includes: collecting historical operating data including obstacle avoidance event data, path deviation data, energy consumption data, and fault data; generating the environmental risk map based on the spatiotemporal distribution analysis of the obstacle avoidance event data; and constructing the health warning matrix based on the correlation analysis between various types of historical operating data. S200. Before the task begins, multiple alternative paths with different risk characteristics are pre-generated based on the current operation time and environmental risk map, and the comprehensive score of each alternative path is calculated; the computing and storage resources of the sensing system are adaptively allocated according to the risk level of the areas traversed by the alternative paths. S300: During task execution, the robot's operating data is collected in real time and compared with the predicted data of the environmental optimization model in real time. S400: When the deviation between the operating data and the predicted data exceeds the preset threshold, dynamic path replanning is triggered and abnormal events are recorded; when the operating data reaches the risk threshold in the health warning matrix, a graded warning mechanism is triggered and operating adjustment or maintenance suggestions are output; dynamic path replanning is a local path adjustment based on the latest real-time environmental perception information, while the graded warning mechanism outputs gradient suggestions from fine-tuning of operating parameters to preventive maintenance based on different risk level thresholds reached. S500. After the mission is completed, based on the operational data collected during the mission, the environmental risk map and health warning matrix are updated. The update involves incorporating new obstacle events observed during the mission into the map to correct the risk probability, and feeding back the correlation between the operational status and health status during the mission to the matrix to optimize the evaluation rules. Based on the consistency evaluation results of path execution, the path confidence weight and resource allocation strategy of each region are dynamically adjusted. The path confidence is a quantitative evaluation of the model's predictive reliability in the region based on the degree of consistency between the historical actual path and the planned path.

[0005] According to the above scheme, step S100 includes: S110. Collect historical operating data of the robot in multiple work cycles. The historical operating data includes obstacle avoidance event data, path deviation data, energy consumption data, and fault data. The obstacle avoidance event data includes the time of obstacle occurrence, geographical coordinates, obstacle type, and robot response behavior. The path deviation data includes quantitative data of the deviation between the robot's actual working path and the preset planned path, which is used to characterize the accuracy of the robot's trajectory control and the degree of environmental interference. The energy consumption data includes real-time power consumption data of each execution subsystem of the robot. The fault data includes the time of fault occurrence, related execution components, and fault type. S120. Based on the analysis of obstacle avoidance event data, an environmental risk map is generated. The analysis includes statistical processing of the temporal, geographical, and type distribution of obstacle occurrences. The statistical processing includes time periodicity analysis, spatial density clustering, and type frequency correlation analysis. The environmental risk map is a dynamic risk probability model generated by multi-dimensional statistical fusion of the occurrence time, spatial coordinates, and category attributes of obstacles. S130. Based on the correlation analysis of obstacle avoidance event data, path deviation data, energy consumption data and fault data in historical operation data, a health early warning matrix is ​​constructed. The health early warning matrix is ​​established by analyzing the statistical correlation between operation data and fault data, and is used as an assessment model to map the real-time operation status to the system health risk level. The correlation analysis includes identifying and quantifying the statistical correlation between specific operation mode combinations and the occurrence of specific types of faults.

[0006] According to the above scheme, step S200 includes: S210. Based on the current operation time, extract the regional risk information for the corresponding time period from the environmental risk map; based on the regional risk information, generate at least two alternative paths with differentiated risk characteristics; the alternative paths are compared in terms of path length, estimated risk accumulation of the areas they traverse, and historical environmental stability of the areas. The differences are reflected in the path's exposure to high-risk areas, the total estimated risk value, and the path smoothness. All alternative paths connect the task start point and the end point, and ensure that the robot can pass through them. S220. Calculate the comprehensive score of each alternative route. The comprehensive score is calculated by weighting the route length, the risk level of the area traversed, and the stability of the regional environment of each alternative route. S230. Based on the risk level of each area traversed by the selected alternative path, and taking the risk level as the decision variable, the system dynamically configures the frequency of sensing data acquisition and the priority of processing resource allocation for the corresponding area in the sensing system, following the principle that the higher the risk level, the greater the resource allocation intensity.

[0007] According to the above scheme, step S300 includes: S310. During the task execution process, the robot's operation data is collected in real time, including real-time obstacle avoidance event data, real-time path deviation data, real-time energy consumption data, and real-time fault status data. S320. Based on the robot's current working position and current working time, obtain the prediction data for the corresponding working time period and working area from the environmental optimization model. The prediction data includes predicted obstacle information, predicted path deviation range, predicted energy consumption range, and predicted fault risk status. S330. The running data and the predicted data are compared dimension by dimension to generate real-time deviation analysis results. The real-time deviation analysis results are structured datasets that quantify the deviation between the actual observed values ​​and the model predicted values ​​in each dimension.

[0008] According to the above scheme, step S400 includes: S410. Obtain the real-time deviation analysis results and compare them with the corresponding first preset threshold. If the deviation of at least one dimension exceeds the first preset threshold, it is determined to be an abnormal operation event. S420. When an abnormal operation event is determined to occur, dynamic path replanning based on the current environmental information is triggered, and the relevant data of the abnormal operation event is recorded in the abnormal event database. The abnormal event database is used to store data of sudden events that the model has not fully predicted. S430. Input the real-time collected operational data into the health early warning matrix to assess the risk status. When the assessment result reaches the second preset threshold, trigger the graded early warning mechanism. S440. Based on the warning level corresponding to the graded warning mechanism, generate and output targeted robot operation parameter adjustment instructions or system component maintenance suggestions; the specific content of the instructions or suggestions is directly related to the triggered warning level.

[0009] According to the above scheme, step S500 includes: S510. After the task is completed, extract all the runtime data collected during the execution of this task; S520. Based on the obstacle avoidance event data in the operational data, update the obstacle risk level information of the corresponding spatiotemporal region in the environmental risk map. The basis for the update is the impact of this event on the historical obstacle statistical characteristics of the region. S530. Based on all operational data from this task, update the correlation between behavioral data, energy consumption data, and fault data in the health warning matrix; S540. Calculate the consistency assessment results between the actual execution path and the planned path for this task, and dynamically adjust the path confidence weights of each region in the environmental risk map based on the consistency assessment results. The path confidence weights are adjusted in reverse based on the consistency assessment results. S550. Based on the updated environmental risk map and combined with the path confidence weights, update the resource allocation strategy of the perception system in the corresponding area.

[0010] A control system for an intelligent wire-laying robot based on perception and obstacle avoidance, comprising: a model building module, a dynamic programming module, a real-time comparison module, an event response module, and an update and optimization module. The model building module is used to collect historical operational data of the robot and build an environmental optimization model based on the historical operational data. The environmental optimization model includes an environmental risk map and a health warning matrix. The model building module includes a data acquisition unit, a map construction unit, and a matrix construction unit. The data acquisition unit is used to collect historical and real-time operational data of the robot over multiple work cycles. The operational data includes obstacle avoidance event data, path deviation data, energy consumption data, and fault data. The map construction unit is used to perform spatiotemporal distribution analysis on obstacle avoidance event data to generate and update the environmental risk map. The matrix construction unit is used to perform correlation analysis on obstacle avoidance event data, path deviation data, energy consumption data, and fault data in the operational data to build a health warning matrix. The dynamic programming module is used to generate and evaluate multiple alternative paths based on the environmental risk map and the current operation time before the task begins, and dynamically allocate the computing and storage resources of the sensing system according to the risk level of the selected path. The real-time comparison module is used to collect running data in real time during task execution and compare it with the prediction data provided by the environmental optimization model to generate real-time deviation analysis results. The event response module is used to trigger dynamic path replanning and anomaly recording based on real-time deviation analysis results, and to trigger graded early warnings and maintenance recommendations based on the risk assessment results of the operational data in the health warning matrix. The update and optimization module is used to update the environmental risk map and health warning matrix based on the data from this task after the task is completed, and to adjust the path confidence weight and resource allocation strategy according to the path execution consistency assessment results.

[0011] According to the above scheme, the dynamic planning module includes a path planning unit, a path evaluation unit, and a resource allocation unit. The path planning unit is used to extract regional risk information from the environmental risk map based on the current operation time and generate at least two alternative paths with differentiated risk characteristics. The route assessment unit is used to assign a weighted score to each candidate route based on route length, risk level of the areas traversed, and regional environmental stability. The resource allocation unit is used to dynamically configure the sensing data acquisition frequency and processing resource allocation priority of the corresponding areas in the sensing system according to the risk level of each area traversed by the selected path.

[0012] According to the above scheme, the event response module includes an exception handling unit, a health warning unit, and an instruction output unit; The anomaly handling unit is used to determine an abnormal operation event when the real-time deviation analysis result exceeds the first preset threshold, trigger dynamic path replanning, and record the event data to the abnormal event database. The health early warning unit is used to input real-time collected operational data into the health early warning matrix for risk assessment. When the assessment result reaches the second preset threshold, a graded early warning mechanism is triggered. The instruction output unit is used to generate and output corresponding robot operation parameter adjustment instructions or system component maintenance suggestions based on the triggered warning level.

[0013] According to the above scheme, the update and optimization module includes a model update unit, a confidence adjustment unit, and a strategy optimization unit; The model update unit is used to update the obstacle risk level information in the environmental risk map and the correlation in the health warning matrix based on the operational data collected in this task. The confidence adjustment unit is used to dynamically adjust the path confidence weight of each region in the environmental risk map based on the consistency assessment results between the actual execution path and the planned path of this task. The strategy optimization unit is used to update the perception system resource configuration strategy for the corresponding area based on the updated environmental risk map and path confidence weights.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention constructs an environmental risk map that integrates spatiotemporal features, enabling robots to learn environmental patterns based on historical data, predict risks in different time periods and regions before the task begins, and generate optimized paths with differentiated risk characteristics, thereby improving the robot's adaptability and decision-making foresight in dynamic environments. 2. This invention dynamically configures the acquisition frequency and processing resource priority of the sensing system according to the path risk level, thereby achieving efficient utilization of computing and storage resources; 3. By establishing and updating a health early warning matrix of related behaviors, energy consumption and faults, this invention can conduct real-time assessment and risk warning of the system's health status and output targeted maintenance suggestions, transforming passive maintenance into proactive prevention, thereby improving reliability and service life. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the steps of a control method for an intelligent wire-laying robot based on perception and obstacle avoidance according to the present invention. Figure 2 This is a schematic diagram of the control system for an intelligent wire-laying robot based on perception and obstacle avoidance according to the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example: Figures 1-2 As shown, the present invention provides a technical solution, a control method for an intelligent wire-laying robot based on perception and obstacle avoidance, the method comprising: S100: Collect historical operating data of the robot in multiple work cycles, and build an environmental optimization model based on the historical operating data. The environmental optimization model includes an environmental risk map that reflects the spatiotemporal distribution characteristics of obstacles, and a health warning matrix that shows the correlation between operating data, energy consumption data and fault data. Specifically, step S100 includes: S110. Collect historical operating data of the robot in multiple work cycles. The historical operating data includes obstacle avoidance event data, path deviation data, energy consumption data, and fault data. The obstacle avoidance event data includes the time of obstacle occurrence, geographical coordinates, obstacle type, and robot response behavior. The path deviation data includes quantitative data of the deviation between the robot's actual working path and the preset planned path, which is used to characterize the accuracy of the robot's trajectory control and the degree of environmental interference. The energy consumption data includes real-time power consumption data of each execution subsystem of the robot. The fault data includes the time of fault occurrence, related execution components, and fault type. For example: The robot has completed N complete work cycles, where N is a positive integer. Retrieve the historical running data for these N cycles from the storage unit, including: Obstacle avoidance event data: {Timestamp: 2025-01-01 14:30:05, Coordinates: (X=105.3, Y=207.8), Obstacle type: Temporary pile, Response behavior: Stop suddenly and then detour to the left}; Path deviation data: At a certain time t, the preset path point is P. plan (t)=(x p ,y p The actual location obtained through the positioning system is P. real (t)=(x r ,y r ), where x and y represent the abscissa and ordinate in a two-dimensional plane coordinate system, respectively, P plan (t) represents the location of the preset or planned path point at time t, P real Let (t) represent the robot's actual position obtained by the positioning system at time t; then the path deviation d(t) at this time is calculated as the Euclidean distance d(t) = sqrt((x)). r -x p ) 2 +(y r -y p ) 2 The deviation data for the entire task cycle is a sequence {d(t1), d(t2), ..., d(tm)}, where m represents the total number of deviation data. Energy consumption data: The current I and voltage V of the drive motor, main controller, and sensor module are recorded at the sampling period. The real-time power consumption P(t) = I(t) × V(t) is recorded as {timestamp: 2025-01-01 14:30:00, motor power consumption: 150W, controller power consumption: 25W, lidar power consumption: 30W}. Fault data: {Occurrence time: 2025-01-01, Related component: Left drive wheel encoder, Fault type: Signal loss}; This is only an example and is not a limitation. S120. Based on the analysis of obstacle avoidance event data, an environmental risk map is generated. The analysis includes statistical processing of the temporal, geographical, and type distribution of obstacle occurrences. The statistical processing includes time periodicity analysis, spatial density clustering, and type frequency correlation analysis. The environmental risk map is a dynamic risk probability model generated by multi-dimensional statistical fusion of the occurrence time, spatial coordinates, and category attributes of obstacles. For example: The system performs statistical analysis on all obstacle avoidance event data collected over the past N periods; Time distribution analysis: Divide the day into multiple time periods and count the total frequency of obstacle events in each time period; for example, if the statistics show that the average number of obstacle events occurs twice as often as in other time periods between 14:00 and 16:00, then assign a higher time risk coefficient R to this time period. time ; Geographic distribution analysis: The work area is gridded, for example, into 1m×1m grids; spatial density clustering algorithm is used to analyze the clustering of obstacle event coordinates; the number C of obstacle events occurring in each grid is counted. ij Where i and j are grid indices; the spatial risk coefficient R of the grid space (i,j) is represented by the number of obstacle events C occurring in each grid cell. ij After normalization, we obtain R. space (i,j)=C ij / max(C all ); where max(C all ) represents the maximum number of obstacle events counted for all grids within the N historical operation cycles; Type distribution analysis: The frequency of occurrence of various obstacles is statistically analyzed, and the difficulty of obstacle avoidance is assessed. For example, the risk weight is set as 1.0 for moving vehicles, 0.7 for fixed facilities, and 0.5 for temporary storage materials. Combined with the response behaviors in the event, the type risk coefficient R is comprehensively evaluated. type ; In the environmental risk map, for a specific time period T and grid (i,j), the comprehensive risk level R(T,i,j) is calculated using a fusion function, such as a weighted average: R(T,i,j) = w1 × R time (T)+w2×R space (i,j)+w3×R type (avg); where R(T,i,j) represents the predicted risk level at time period T and location (i,j); R time (T) represents the time risk coefficient for time period T; R space (i,j) represents the spatial risk coefficient of grid (i,j); R type(avg) represents the average risk coefficient of the type of historical obstacle events in the grid; w1, w2 and w3 are weight coefficients, and w1+w2+w3=1; in this embodiment, the weight coefficients are determined by reverse optimization through historical data: a segment of historical data is selected as the validation set, and the accuracy of path prediction is used as the optimization objective. The optimal combination of w1, w2 and w3 is solved using grid search or optimization algorithms. This is only an example and is not a limitation. S130. Based on the correlation analysis of obstacle avoidance event data, path deviation data, energy consumption data and fault data in historical operation data, a health early warning matrix is ​​constructed. The health early warning matrix is ​​established by analyzing the statistical correlation between operation data and fault data, and is used as an assessment model to map real-time operation status to system health risk level. The correlation analysis includes identifying and quantifying the statistical correlation between specific operation mode combinations and the occurrence of specific types of faults. For example, through association rule mining or statistical analysis, the following patterns were discovered: Rule 1: If the number of emergency stop obstacle avoidances is greater than 5 within 10 minutes, and the average power consumption fluctuation rate of the drive motor is greater than 15% within the same period, the probability of drive motor overheating failure in the following 24 hours increases to 40%; Rule 2: If the root mean square error of the path deviation is higher than the threshold L for more than 5 minutes, the probability of positioning module abnormal failure in subsequent tasks increases to 30%; This is only an example and is not a limitation. The health warning matrix encodes rules; for example, it takes real-time operational data as input, including behavior vector B, energy consumption vector E, and path deviation vector D, and uses a pre-trained risk assessment function F. hazard The comprehensive health risk score H is calculated, expressed by the formula H=F hazard (B,E,D); where the range of H is set to [0,1], and a higher value represents a greater health risk to the system; the correlation between the matrices is reflected in the function F. hazard In the parameters and structure of function F hazard This is obtained by learning the relationship between the behavior vector B, energy consumption vector E, and path deviation vector D in historical data and the fault label F; where fault label F=1 indicates a fault, and fault label F=0 indicates no fault. Furthermore, the pre-trained evaluation function F hazardThe specific steps include: collecting a large amount of historical task data, and tagging each piece of data with a fault label F, that is, the data within a period of time before and after the occurrence of a fault is marked as F = 1, and the normal operation data is marked as F = 0; extracting or constructing features for fault prediction from the original operation data to form a feature vector (B, E, D); selecting logistic regression, support vector machine, random forest or neural network, and training the model using the prepared historical data set. By adjusting the model parameters, the model can accurately distinguish the normal state from the fault precursor state; after the training is completed, run the model on an independent validation set to obtain a prediction score H; according to the business's tolerance for false alarms and missed alarms, determine the second preset threshold for triggering an early warning on the distribution of H. Here, only an example is given for illustration and no limitation is made; S200. Before the task starts, combine the current operation time with the environmental risk map to pre-generate multiple alternative paths with different risk characteristics, and calculate the comprehensive scores of each alternative path; adaptively allocate the computing and storage resources of the sensing system according to the risk levels of the areas passed by the alternative paths; Specifically, step S200 includes: S210. According to the current operation time, extract the regional risk information for the corresponding period from the environmental risk map; based on the regional risk information, generate at least two alternative paths with different risk characteristics; the alternative paths form a differential comparison in terms of path length, the estimated cumulative risk of the areas traversed, and the historical environmental stability of the areas. The difference is reflected in the exposure degree of the path to high-risk areas, the total estimated risk value, and the path smoothness. All alternative paths connect the task start point and the end point and ensure that the robot can pass through; For example: the current time is 14:30, and the plan is to operate from the start point A to the end point B; extract the regional risk information R(14:00 - 16:00, i, j) for the period from 14:00 to 16:00 from the environmental risk map; after considering the risk cost, the path planning algorithm generates two paths: Path 1 (fast): length 80 meters, passing through 3 high-risk grids (R > 0.8), estimated cumulative risk ΣR1 = 2.5; Path 2 (safe): length 95 meters, only passing through 1 medium-risk grid (0.5 < R < 0.8), estimated cumulative risk ΣR2 = 0.6. Here, only an example is given for illustration and no limitation is made; S220. Calculate the comprehensive scores of each alternative path, and the comprehensive scores are calculated by weighted calculation based on the path length, the risk levels of the areas passed by, and the regional environmental stability of each alternative path; For example, in this embodiment, the scoring function is: Score = α × (1 / L) + β × (1 / ΣR) + γ × Stability; where L represents the path length, ΣR represents the cumulative risk, Stability represents the average score of historical execution consistency in the areas traversed by the path (higher consistency results in a higher Stability value), and α, β, and γ are weighting coefficients. These weighting coefficients are obtained by analyzing data from historical successful tasks to fit the optimal weight combination for overall task performance, or by using a multi-objective optimization algorithm to generate a Pareto-optimal weighting scheme for decision-makers to choose from. After calculation and analysis, path 1 scores 75 and path 2 scores 82. This is only an example and is not a limitation. Furthermore, the Stability score, which measures the historical consistency of the routes through the regions, is based on statistical data on the deviations between the actual and planned routes in all historical tasks within that region. For example, the inverse of the mean or the inverse of the standard deviation of the historical route deviations in that region can be used as the stability score. A higher Stability score indicates a smaller mean deviation and less fluctuation. This is just an example and is not a limitation. S230. Based on the risk level of each area traversed by the selected alternative path, and taking the risk level as the decision variable, the system dynamically configures the frequency of sensing data collection and the priority of processing resource allocation for the corresponding area in the sensing system, following the principle that the higher the risk level, the greater the intensity of resource allocation. For example: Based on the comprehensive score, path 2 is selected; for the only medium-risk grid on path 2, the system increases the scanning frequency of the LiDAR in that area from 10Hz to 20Hz and sets the CPU priority of the point cloud data processing task in that area to high; for other low-risk grids, the scanning frequency of 10Hz and normal priority are maintained; this is only an example and is not a limitation.

[0018] S300: During task execution, the robot's operating data is collected in real time and compared with the predicted data of the environmental optimization model in real time. Specifically, step S300 includes: S310. During the task execution process, the robot's operation data is collected in real time, including real-time obstacle avoidance event data, real-time path deviation data, real-time energy consumption data, and real-time fault status data. For example: The robot travels along path 2 and collects data in real time; at time t, the following data is collected: no new obstacle avoidance events, real-time position deviation from the planned point d(t) = 0.12m, current total power consumption P(t) = 205W, and no fault codes. S320. Based on the robot's current working position and current working time, obtain the prediction data for the corresponding working time period and working area from the environmental optimization model. The prediction data includes predicted obstacle information, predicted path deviation range, predicted energy consumption range, and predicted fault risk status. For example: Based on the robot's position at time t and the current time, query the environmental optimization model to obtain predicted data: the predicted obstacle probability P. obs =0.1, predicted path deviation range [0.05m, 0.15m], predicted energy consumption range [180W, 220W], predicted health risk H pred =0.05; S330. The running data and the predicted data are compared dimension by dimension to generate real-time deviation analysis results. The real-time deviation analysis results are structured datasets that quantify the deviation between the actual observed values ​​and the model predicted values ​​in each dimension. For example: the actual deviation d(t) = 0.12m is within the prediction range [0.05m, 0.15m]; the actual power consumption of 205W is within the prediction range [180W, 220W]; there is no actual fault, and the prediction risk is low; the deviation analysis result is generated: {path deviation: 0, energy consumption deviation: 0, ...}; this is just an example and is not a limitation.

[0019] S400: When the deviation between the operating data and the predicted data exceeds the preset threshold, dynamic path replanning is triggered and abnormal events are recorded; when the operating data reaches the risk threshold in the health warning matrix, a graded warning mechanism is triggered and operating adjustment or maintenance suggestions are output; dynamic path replanning is a local path adjustment based on the latest real-time environmental perception information, while the graded warning mechanism outputs gradient suggestions from fine-tuning of operating parameters to preventive maintenance based on different risk level thresholds reached. Specifically, step S400 includes: S410. Obtain the real-time deviation analysis results and compare them with the corresponding first preset threshold. If the deviation of at least one dimension exceeds the first preset threshold, it is determined to be an abnormal operation event. For example, at another moment, the robot suddenly detects an unpredictable moving obstacle, causing an emergency stop obstacle avoidance event, and the actual power consumption instantly spikes to 250W. Comparison reveals that the deviations in both the obstacle avoidance event dimension and the energy consumption dimension exceed a first preset threshold, which the system determines as an abnormal operation event. In this embodiment, the first preset threshold is set as the upper and lower boundaries of the prediction interval itself. For example, if the predicted energy consumption interval is [180W, 220W], then the threshold is 180W and 220W; exceeding this boundary is considered abnormal. For dimensions without a clearly defined interval, the threshold is set by calculating the standard deviation of historical normal data fluctuations in that dimension; for example, setting the threshold as Y±k. σ, where Y represents the predicted value, k is a constant, and σ is the standard deviation; these are adjusted according to the requirements for sensitivity to anomalies; both k and σ are obtained through historical data analysis during the model building phase; this is only an example and is not a limitation. S420. When an abnormal operation event is determined to occur, dynamic path replanning based on the current environmental information is triggered, and the relevant data of the abnormal operation event is recorded in the abnormal event database. The abnormal event database is used to store data of sudden events that the model has not fully predicted. For example: immediately initiate the dynamic window method for local path replanning, calculate a new local trajectory that bypasses the temporary obstacle; simultaneously, record the time, location, obstacle type, and actual power consumption of this event in the abnormal event database; in this embodiment, the parameters required by the dynamic path replanning algorithm include kinematic and dynamic constraint parameters based on the physical characteristics of the robot platform and algorithm control parameters for trajectory optimization. These parameters are predetermined through experimental calibration or engineering calculations based on the physical characteristics of the specific robot model, the safety specifications of the work scenario, and the requirements for path smoothness, and are stored in the system configuration file and loaded and called during algorithm initialization; S430. Input the real-time collected operational data into the health early warning matrix to assess the risk status. When the assessment result reaches the second preset threshold, trigger the graded early warning mechanism. For example: the system inputs a real-time operational vector containing emergency stop behavior data and 250W power consumption data into the health warning matrix; the risk assessment function F in the health warning matrix... hazard Calculate the current health risk score H current =0.65; S440. Based on the warning level corresponding to the graded warning mechanism, generate and output targeted robot operation parameter adjustment instructions or system component maintenance suggestions; the specific content of the instructions or suggestions is directly related to the triggered warning level. For example: Second preset threshold [Note: 0.5, Warning: 0.7, Critical: 0.9]; H current =0.65 reached the attention level; a graded warning was triggered, and a suggestion was output: Note: The current system load is high. It is recommended to reduce the maximum driving speed by 20% in subsequent road sections; this is only an example and does not impose any restrictions.

[0020] S500. After the mission is completed, based on the operational data collected during the mission, the environmental risk map and health warning matrix are updated. The update involves incorporating new obstacle events observed during the mission into the map to correct the risk probability, and feeding back the correlation between the operational status and health status during the mission to the matrix to optimize the evaluation rules. Based on the consistency evaluation results of path execution, the path confidence weight and resource allocation strategy of each region are dynamically adjusted. The path confidence is a quantitative evaluation of the model's predictive reliability in the region based on the degree of consistency between the historical actual path and the planned path. Specifically, step S500 includes: S510. After the task is completed, extract all the runtime data collected during the execution of this task; S520. Based on the obstacle avoidance event data in the operational data, update the obstacle risk level information of the corresponding spatiotemporal region in the environmental risk map. The basis for the update is the impact of this event on the historical obstacle statistical characteristics of the region. For example: update the obstacle count C of the corresponding grid in the environmental risk map with the new obstacle event recorded in the abnormal event database. ij In the middle, recalculate the R of the grid. space (i,j) increases the probability of future risks in this area; S530. Based on all operational data from this task, update the correlation between behavioral data, energy consumption data, and fault data in the health warning matrix; For example, the combination of emergency stops and high power consumption observed in this task, along with the result of no motor failure occurring after the task, can be used as new training data to fine-tune the risk assessment function F in the health warning matrix. hazard This makes future assessments of similar situations slightly more conservative, meaning the score H might be slightly lower, since no failure occurred this time; this is just an example and not a limitation. S540. Calculate the consistency assessment results between the actual execution path and the planned path for this task, and dynamically adjust the path confidence weights of each region in the environmental risk map based on the consistency assessment results. The path confidence weights are adjusted in reverse based on the consistency assessment results. For example: Calculate the consistency between the actual trajectory and the planned path in each grid cell; identify grid cells where abnormal events have occurred, indicating a significant deviation between the actual and planned paths, and calculate the confidence weight based on the grid cell's path confidence formula. new (i,j)=Confidence old (i,j)×(1-θ×Inconsistency), lowers the path confidence weight Confidence(i,j) for this grid; where Inconsistency represents the consistency evaluation score calculated based on the current path deviation, such as normalized to the [0,1] interval, the larger the value, the higher the inconsistency, and θ represents the preset learning rate coefficient and 0<θ<1; this is only an example and is not a limitation. S550. Based on the updated environmental risk map and combined with path confidence weights, update the perception system resource allocation strategy for the corresponding area; in this embodiment, the environmental risk map uses an exponentially decaying moving average to update the obstacle count C. ij The formula is: C ij-new =λ×C ij-old +(1-λ)×I, where λ is the forgetting factor, I() is the indicator function (1 if the event occurs, 0 otherwise), and I is the value of the current event; F in the health warning matrix hazardThe function is fine-tuned using an online learning algorithm to achieve incremental learning.

[0021] This invention provides another technical solution: a control system for an intelligent wire-laying robot based on perception and obstacle avoidance. The system includes: a model building module, a dynamic programming module, a real-time comparison module, an event response module, and an update and optimization module. The model building module is used to collect historical operational data of the robot and build an environmental optimization model based on the historical operational data. The environmental optimization model includes an environmental risk map and a health warning matrix. The model building module includes a data acquisition unit, a map construction unit, and a matrix construction unit. The data acquisition unit is used to collect historical and real-time operational data of the robot over multiple work cycles. The operational data includes obstacle avoidance event data, path deviation data, energy consumption data, and fault data. The map construction unit is used to perform spatiotemporal distribution analysis on obstacle avoidance event data to generate and update the environmental risk map. The matrix construction unit is used to perform correlation analysis on obstacle avoidance event data, path deviation data, energy consumption data, and fault data in the operational data to build a health warning matrix. The dynamic programming module is used to generate and evaluate multiple alternative paths based on the environmental risk map and the current operation time before the task begins, and dynamically allocate the computing and storage resources of the sensing system according to the risk level of the selected path. The real-time comparison module is used to collect running data in real time during task execution and compare it with the prediction data provided by the environmental optimization model to generate real-time deviation analysis results. The event response module is used to trigger dynamic path replanning and anomaly recording based on real-time deviation analysis results, and to trigger graded early warnings and maintenance recommendations based on the risk assessment results of the operational data in the health warning matrix. The update and optimization module is used to update the environmental risk map and health warning matrix based on the data from this task after the task is completed, and to adjust the path confidence weight and resource allocation strategy according to the path execution consistency assessment results.

[0022] Specifically, the dynamic planning module includes a path planning unit, a path evaluation unit, and a resource allocation unit. The path planning unit is used to extract regional risk information from the environmental risk map based on the current operation time and generate at least two alternative paths with differentiated risk characteristics. The path evaluation unit is used to assign a weighted score to each alternative path based on the path length, the risk level of the areas traversed, and the environmental stability of the areas. The resource allocation unit is used to dynamically configure the sensing data acquisition frequency and processing resource allocation priority of the corresponding areas in the sensing system according to the risk level of each area traversed by the selected path.

[0023] Specifically, the event response module includes an anomaly handling unit, a health warning unit, and an instruction output unit. The anomaly handling unit is used to determine an abnormal operation event when the real-time deviation analysis result exceeds a first preset threshold, trigger dynamic path replanning, and record the event data to the anomaly event database. The health warning unit is used to input the real-time collected operation data into the health warning matrix for risk assessment, and trigger a graded warning mechanism when the assessment result reaches a second preset threshold. The instruction output unit is used to generate and output corresponding robot operation parameter adjustment instructions or system component maintenance suggestions according to the triggered warning level.

[0024] Specifically, the update and optimization module includes a model update unit, a confidence adjustment unit, and a strategy optimization unit. The model update unit is used to update the obstacle risk level information in the environmental risk map and the correlation in the health warning matrix based on the operational data collected in this task. The confidence adjustment unit is used to dynamically adjust the path confidence weight of each region in the environmental risk map according to the consistency assessment results between the actual execution path and the planned path of this task. The strategy optimization unit is used to update the perception system resource configuration strategy for the corresponding region according to the updated environmental risk map and path confidence weight.

[0025] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A control method for an intelligent wire-laying robot based on perception and obstacle avoidance, characterized in that: The method includes: S100. Collect historical operating data of the robot in multiple work cycles, and construct an environmental optimization model based on the historical operating data. The environmental optimization model includes an environmental risk map reflecting the spatiotemporal distribution characteristics of obstacles, and a health warning matrix showing the correlation between operating data, energy consumption data and fault data. S200. Before the task begins, multiple alternative paths with different risk characteristics are pre-generated based on the current operation time and environmental risk map, and the comprehensive score of each alternative path is calculated; the computing and storage resources of the sensing system are adaptively allocated according to the risk level of the areas traversed by the alternative paths. S300: During task execution, the robot's operating data is collected in real time and compared with the predicted data of the environmental optimization model in real time. S400. When the deviation between the running data and the predicted data exceeds a preset threshold, dynamic path replanning is triggered and abnormal events are recorded; when the running data reaches the risk threshold in the health warning matrix, a graded warning mechanism is triggered and running adjustment or maintenance suggestions are output. S500 After the mission is completed, based on the operational data collected during the mission, update the environmental risk map and health warning matrix, and dynamically adjust the path confidence weight and resource allocation strategy for each region according to the consistency assessment results of path execution.

2. The control method for an intelligent wire-laying robot based on perception and obstacle avoidance according to claim 1, characterized in that: Step S100 includes: S110. Collect historical operating data of the robot in multiple work cycles. The historical operating data includes obstacle avoidance event data, path deviation data, energy consumption data, and fault data. The obstacle avoidance event data includes obstacle occurrence time, geographical coordinates, obstacle type, and robot response behavior. The path deviation data includes quantified data of the deviation between the robot's actual working path and the preset planned path. The energy consumption data includes real-time power consumption data of each execution subsystem of the robot. The fault data includes fault occurrence time, associated execution components, and fault type. S120. Based on the analysis of the obstacle avoidance event data, generate the environmental risk map, wherein the analysis includes statistical processing of the temporal distribution, geographical distribution and type distribution of obstacle occurrences; S130. Based on the correlation analysis between the obstacle avoidance event data, path deviation data, energy consumption data and fault data in the historical operation data, construct the health early warning matrix.

3. The control method for an intelligent wire-laying robot based on perception and obstacle avoidance according to claim 1, characterized in that: Step S200 includes: S210. Based on the current operation time, extract the regional risk information corresponding to the time period from the environmental risk map; based on the regional risk information, generate at least two alternative paths with differentiated risk characteristics; S220. Calculate the comprehensive score of each candidate route, wherein the comprehensive score is calculated by weighting the route length, the risk level of the area traversed, and the stability of the regional environment of each candidate route. S230. Based on the risk level of each area traversed by the selected alternative path, dynamically configure the sensing data acquisition frequency and processing resource allocation priority of the corresponding area in the sensing system.

4. The control method for an intelligent wire-laying robot based on perception and obstacle avoidance according to claim 1, characterized in that: Step S300 includes: S310. During the task execution process, the robot's operation data is collected in real time, including real-time obstacle avoidance event data, real-time path deviation data, real-time energy consumption data, and real-time fault status data. S320. Based on the robot's current working position and current working time, obtain the prediction data for the corresponding working time period and working area from the environmental optimization model. The prediction data includes predicted obstacle information, predicted path deviation range, predicted energy consumption range, and predicted fault risk status. S330. The running data and the predicted data are compared dimension by dimension to generate real-time deviation analysis results.

5. The control method for an intelligent wire-laying robot based on perception and obstacle avoidance according to claim 1, characterized in that: Step S400 includes: S410. Obtain the real-time deviation analysis result, and compare the real-time deviation analysis result with the corresponding first preset threshold. If the deviation of at least one dimension exceeds its first preset threshold, it is determined to be an abnormal operation event. S420. When the abnormal operation event is determined to have occurred, dynamic path replanning based on the current environment information is triggered, and the relevant data of the abnormal operation event is recorded in the abnormal event database. S430. Input the real-time collected operation data into the health early warning matrix to assess the risk status. When the assessment result reaches the second preset threshold, trigger the graded early warning mechanism. S440. Based on the warning level corresponding to the graded warning mechanism, generate and output targeted robot operation parameter adjustment instructions or system component maintenance suggestions.

6. The control method for an intelligent wire-laying robot based on perception and obstacle avoidance according to claim 1, characterized in that: Step S500 includes: S510. After the task is completed, extract all the runtime data collected during the execution of this task; S520. Based on the obstacle avoidance event data in the operational data, update the obstacle risk level information of the corresponding spatiotemporal region in the environmental risk map; S530. Based on all the operational data of this task, update the correlation between behavioral data, energy consumption data and fault data in the health warning matrix; S540. Calculate the consistency assessment result between the actual execution path and the planned path of this task, and dynamically adjust the path confidence weight of each region in the environmental risk map based on the consistency assessment result. S550. Based on the updated environmental risk map and the path confidence weights, update the perception system resource configuration strategy for the corresponding area.

7. A control system for an intelligent wire-laying robot based on perception and obstacle avoidance, characterized in that: The system includes: a model building module, a dynamic programming module, a real-time comparison module, an event response module, and an update and optimization module. The model building module is used to collect historical operating data of the robot and build an environmental optimization model based on the historical operating data. The environmental optimization model includes an environmental risk map and a health warning matrix. The model building module includes a data acquisition unit, a map building unit, and a matrix building unit. The data acquisition unit is used to collect historical and real-time operating data of the robot in multiple work cycles. The operating data includes obstacle avoidance event data, path deviation data, energy consumption data, and fault data. The map building unit is used to perform spatiotemporal distribution analysis on the obstacle avoidance event data to generate and update the environmental risk map. The matrix building unit is used to perform correlation analysis on the obstacle avoidance event data, path deviation data, energy consumption data, and fault data in the operating data to build the health warning matrix. The dynamic programming module is used to generate and evaluate multiple alternative paths based on the environmental risk map and the current operation time before the task begins, and to dynamically allocate the computing and storage resources of the sensing system according to the risk level of the selected path. The real-time comparison module is used to collect running data in real time during task execution and compare it with the prediction data provided by the environmental optimization model to generate real-time deviation analysis results. The event response module is used to trigger dynamic path replanning and anomaly recording based on the real-time deviation analysis results, and to trigger graded early warning and maintenance suggestion output based on the risk assessment results of the operational data in the health early warning matrix. The update and optimization module is used to update the environmental risk map and health warning matrix based on the task execution data after the task is completed, and to adjust the path confidence weight and resource allocation strategy according to the path execution consistency assessment results.

8. The intelligent wire-laying robot control system based on perception and obstacle avoidance according to claim 7, characterized in that: The dynamic planning module includes a path planning unit, a path evaluation unit, and a resource allocation unit. The path planning unit is used to extract regional risk information from the environmental risk map based on the current operation time, and generate at least two alternative paths with differentiated risk characteristics. The route evaluation unit is used to give a weighted score to each candidate route based on the route length, the risk level of the area it passes through, and the stability of the regional environment. The resource allocation unit is used to dynamically configure the sensing data acquisition frequency and processing resource allocation priority of the corresponding area in the sensing system according to the risk level of each area traversed by the selected path.

9. The intelligent wire-laying robot control system based on perception and obstacle avoidance according to claim 7, characterized in that: The event response module includes an exception handling unit, a health warning unit, and an instruction output unit; The anomaly handling unit is used to determine an abnormal operation event when the real-time deviation analysis result exceeds a first preset threshold, trigger dynamic path replanning, and record the event data to the abnormal event database. The health early warning unit is used to input the real-time collected operational data into the health early warning matrix for risk assessment. When the assessment result reaches the second preset threshold, a graded early warning mechanism is triggered. The instruction output unit is used to generate and output corresponding robot operating parameter adjustment instructions or system component maintenance suggestions based on the triggered warning level.

10. The intelligent wire-laying robot control system based on perception and obstacle avoidance according to claim 7, characterized in that: The update and optimization module includes a model update unit, a confidence adjustment unit, and a strategy optimization unit. The model update unit is used to update the obstacle risk level information in the environmental risk map and the correlation in the health warning matrix based on the operational data collected in this task. The confidence adjustment unit is used to dynamically adjust the path confidence weight of each region in the environmental risk map based on the consistency assessment results between the actual execution path and the planned path of this task. The strategy optimization unit is used to update the perception system resource configuration strategy for the corresponding area based on the updated environmental risk map and path confidence weights.

Images