Road surface automatic inspection robot control method and equipment

Abstract

The invention relates to the technical field of automatic inspection, in particular to an automatic road surface inspection robot control method and equipment, and the method comprises the steps: collecting environment data and robot motion state data through integrating a plurality of sensors, carrying out the preprocessing, carrying out the data fusion through employing a Kalman filtering algorithm, and generating a comprehensive road surface state map; a disease area is accurately identified and positioned through a convolutional neural network, dynamic path planning and polling strategy optimization are performed through a reinforcement learning algorithm, adaptive control of a complex road surface environment is realized through a fuzzy control algorithm, and meanwhile, a targeted road surface maintenance suggestion and an emergency processing scheme are generated according to real-time polling data, so that the real-time polling of a road surface is realized. According to the method, the scientificity and timeliness of road surface maintenance and traffic management are effectively improved, the road surface diseases can be accurately recognized and the inspection path can be optimized in a complex environment, and the road surface inspection efficiency and the intelligent level of road maintenance are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of automated inspection, and in particular to a control method and equipment for an automated road inspection robot. Background Art

[0002] As a key part of transportation, the timely detection and repair of road defects are of great significance to ensuring traffic safety, extending the service life of roads, and saving maintenance costs. Traditional road inspection methods usually rely on manual inspections or handheld devices for detection. This method is not only inefficient, but also has large subjective errors, and is prone to missing or misdetecting road defects, which cannot meet the growing inspection needs. In order to improve the recognition accuracy and inspection efficiency of road defects, automated and intelligent road inspection technology has become a research hotspot in recent years.

[0003] The existing automatic road inspection technology has improved the inspection efficiency to a certain extent, but there are still some technical bottlenecks. First, most of the existing inspection equipment relies on a single type of sensor or data source, resulting in insufficient comprehensiveness and accuracy of data acquisition in complex road environments. For example, a single camera or ultrasonic sensor often has limitations in detecting different types of diseases, and it is difficult to comprehensively analyze the overall state of the road surface. Secondly, the existing path planning and inspection strategies are mostly based on fixed paths or simple rules, and lack intelligent dynamic optimization capabilities. This makes it difficult for inspection robots to make adaptive adjustments in real time according to environmental changes, which may result in low inspection efficiency or missed inspections of diseases. Furthermore, many existing devices do not have integrated intelligent decision support and cannot generate scientific maintenance recommendations and emergency response plans based on inspection data, thus affecting the timeliness and accuracy of road maintenance. Summary of the invention

[0004] The invention provides a control method for an automatic road surface inspection robot.

[0005] A method for controlling an automatic road inspection robot comprises the following steps:

[0006] S1, data collection and preprocessing: using laser radar, ultrasonic sensor and camera to collect multi-dimensional data, the multi-dimensional data includes robot motion state data and environment data, wherein;

[0007] The robot motion state data includes the robot's current position and speed data;

[0008] The environmental data includes obstacles, road deformation, cracks, settlement and slope;

[0009] And preprocess the collected multi-dimensional data, including denoising, error correction and data filling;

[0010] S2, data fusion and analysis: Based on the pre-processed environmental data, the Kalman filter algorithm is used for data fusion to merge the data from different sensors to generate a comprehensive road condition map, and the convolutional neural network model is used to classify, identify and locate the diseased areas in the map;

[0011] S3, path planning and inspection strategy optimization: Combined with machine learning algorithms, based on real-time environmental data and known disease types, reinforcement learning algorithms are used to dynamically optimize the inspection robot's path to avoid missing diseased areas and improve inspection efficiency;

[0012] S4, real-time feedback and control: Based on the real-time robot motion state data and environmental data, the fuzzy control algorithm is used to dynamically adjust the robot's travel speed and steering angle to achieve adaptive inspection control of complex road environments;

[0013] S5, Decision Support: Provide decision support to generate targeted pavement maintenance recommendations and emergency response plans based on real-time environmental data.

[0014] Optionally, the S1 includes:

[0015] S11, data collection: Use laser radar, ultrasonic sensors and cameras to collect multi-dimensional data including robot motion state data and environmental data, including:

[0016] S111, robot motion state data collection: real-time monitoring of the robot's dynamic state data through the robot's built-in accelerometer and gyroscope, including real-time position coordinates, speed (including linear speed and angular speed) and acceleration data;

[0017] S112, Environmental Data Collection: Obtain the robot's surrounding environment data through lidar, ultrasonic sensors and cameras, including obstacle detection, road deformation, cracks, settlement and slope;

[0018] S12, data preprocessing: preprocessing the collected multi-dimensional data, including:

[0019] S121, denoising: performing noise suppression processing on the collected data;

[0020] S122, error correction: using a known calibration model to perform error correction on the multi-dimensional data;

[0021] S123, data filling: For the missing values ​​in the collected multi-dimensional data, interpolation method is used to fill them to ensure the integrity of the data.

[0022] Optionally, S2 includes:

[0023] S21, data fusion algorithm selection and design: Based on the preprocessed environmental data, the Kalman filter algorithm is selected to fuse the data from different sensors;

[0024] S22, Kalman filter process: sensor data is fused through prediction and update steps;

[0025] S23, data fusion result output: through the calculation result of the Kalman filter, the data from different sensors are fused into a unified fused road surface state data.

[0026] Optionally, S2 further includes:

[0027] S24, road surface state map generation: inputting the fused road surface state data after Kalman filter fusion into a map generation unit to generate a comprehensive road surface state map;

[0028] S25, classification and identification of diseased areas: Analyze the generated pavement status map through a convolutional neural network model to identify and classify the diseased areas in the map;

[0029] S26, disease area result output: the disease area identified by the convolutional neural network model will be output as a visualization result, presented as a map marked with disease type, location and severity.

[0030] Optionally, the S3 includes:

[0031] S31, path planning algorithm selection: combining real-time environmental data and known disease types, dynamically planning an optimal path for the inspection robot through the path planning algorithm to avoid missing diseased areas and improve inspection efficiency;

[0032] S32, Path planning optimization: Dynamically optimize path planning through reinforcement learning algorithm.

[0033] Optionally, S3 further includes:

[0034] S33, Inspection strategy optimization: Based on path planning, optimize the inspection strategy to improve the overall inspection effect, including:

[0035] S331, setting the priority of the diseased area: setting the inspection priority of each diseased area according to the severity, type and location of the diseased area;

[0036] S332, dynamic inspection strategy adjustment: based on the real-time feedback data during the inspection process, such as the robot's motion state data and real-time environment data during the inspection process, the inspection strategy is dynamically adjusted;

[0037] S333, intelligent inspection scheduling: intelligently schedule inspection tasks based on the priority of the diseased area and the robot’s current inspection status.

[0038] Optionally, the S4 includes:

[0039] S41, fuzzy control input variable setting: setting the input variables of the fuzzy control algorithm according to the fused road surface state data;

[0040] S42, fuzzy control rule base design and optimization: set the fuzzy control rule base according to the real-time road status data;

[0041] S43, fuzzy control output calculation: the travel speed and steering angle are calculated by the fuzzy control algorithm and output to the control unit. The control unit generates a control signal by using the fuzzy reasoning method according to the set control rules and input data.

[0042] Optionally, S4 further includes:

[0043] S44, execution of control signals and adjustment of robot behavior: dynamically adjusting the travel speed and steering angle of the robot according to the generated control signals;

[0044] S45, real-time feedback and adaptive adjustment: According to the real-time feedback, the fuzzy control strategy is adjusted to enable the robot to dynamically adapt to complex environmental changes during the inspection process.

[0045] Optionally, the S5 includes:

[0046] S51, Maintenance Recommendations: Generate targeted pavement maintenance recommendations based on the type, severity and pavement condition of the disease;

[0047] S52, emergency response plan: In special circumstances, generate emergency response plans based on real-time environmental data;

[0048] S53, real-time update and adjustment: As road conditions change, maintenance recommendations and emergency response plans are updated in real time to ensure timely and accurate road management.

[0049] A road automatic inspection robot control device, used to implement the above-mentioned road automatic inspection robot control method, includes the following modules:

[0050] Data acquisition module: uses lidar, ultrasonic sensors and cameras to collect multi-dimensional data, and pre-processes the collected multi-dimensional data, including denoising, error correction and data filling;

[0051] Data fusion and analysis module: Based on the pre-processed environmental data, the Kalman filter algorithm is used for data fusion, the data from different sensors are merged to generate a comprehensive road condition map, and the convolutional neural network model is used to classify, identify and locate the diseased areas in the map;

[0052] Path planning and inspection strategy optimization module: Combined with machine learning algorithms, based on real-time environmental data and known disease types, the inspection robot's path is dynamically optimized using reinforcement learning algorithms to avoid missing diseased areas and improve inspection efficiency;

[0053] Real-time feedback and control module: Based on the real-time robot motion state data and environmental data, the fuzzy control algorithm is used to dynamically adjust the robot's travel speed and steering angle to achieve adaptive inspection control of complex road environments;

[0054] Decision support module: Provides decision support and generates targeted pavement maintenance recommendations and emergency response plans based on real-time environmental data.

[0055] Beneficial effects of the present invention:

[0056] The present invention integrates laser radar, ultrasonic sensor and camera sensor equipment. The automatic road inspection robot can comprehensively and real-time collect road state data, including robot motion state data and environmental data. Through denoising, error correction and data filling, the accuracy and quality of the collected data are ensured. The Kalman filter algorithm effectively integrates data from different sensors to generate a comprehensive road state map, and then uses a convolutional neural network to accurately identify and locate the diseased area. The whole process improves the comprehensiveness and efficiency of inspections, significantly reduces the errors and omissions of manual inspections, and improves the accuracy of road disease identification.

[0057] The present invention realizes dynamic optimization of the inspection path in path planning through reinforcement learning algorithm, and can automatically adjust the inspection route according to real-time environmental data and known disease types to avoid missing disease areas and ensure the comprehensiveness and efficiency of the inspection. In addition, according to the real-time motion state data and environmental data of the robot, the fuzzy control algorithm is used to dynamically adjust the robot's travel speed and steering angle to ensure that the robot can be adaptively controlled in complex road environments and maintain a stable inspection effect. It has strong adaptive capabilities and can flexibly respond to different road conditions, including slopes, obstacles, road cracks, etc., to improve the stability and reliability of inspections.

[0058] The present invention combines real-time environmental data, disease identification results and pavement status maps to provide decision makers with intelligent maintenance suggestions and emergency treatment plans. It can not only identify the type and location of the disease, but also evaluate the severity of the disease, generate targeted maintenance priorities, and help relevant departments to rationally plan maintenance resources and optimize pavement maintenance plans. The decision support function not only improves the scientificity and timeliness of pavement maintenance, but also can quickly provide emergency treatment plans when sudden pavement diseases or traffic accidents occur, reduce accident response time, and improve the emergency handling capabilities and safety of transportation facilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0060] Figure 1 A schematic diagram of a method flow of an embodiment of the present invention;

[0061] Figure 2 The figure is a schematic diagram of the equipment flow of an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0063] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0064] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0065] like Figure 1 As shown, a control method for an automatic road inspection robot comprises the following steps:

[0066] S1, data collection and preprocessing: Use laser radar, ultrasonic sensor and camera to collect multi-dimensional data, which includes robot motion state data and environmental data;

[0067] The robot motion state data includes the robot's current position and speed data;

[0068] Environmental data includes obstacles, road deformation, cracks, settlement and slope;

[0069] And preprocess the collected multi-dimensional data, including denoising, error correction and data filling;

[0070] S2, data fusion and analysis: Based on the pre-processed environmental data, the Kalman filter algorithm is used for data fusion to merge the data from different sensors to generate a comprehensive road condition map, and the convolutional neural network model is used to classify, identify and locate the diseased areas in the map;

[0071] S3, path planning and inspection strategy optimization: Combined with machine learning algorithms, based on real-time environmental data and known disease types, reinforcement learning algorithms are used to dynamically optimize the inspection robot's path to avoid missing diseased areas and improve inspection efficiency;

[0072] S4, real-time feedback and control: Based on the real-time robot motion state data and environmental data, the fuzzy control algorithm is used to dynamically adjust the robot's travel speed and steering angle to achieve adaptive inspection control of complex road environments;

[0073] S5, Decision Support: Provide decision support to generate targeted pavement maintenance recommendations and emergency response plans based on real-time environmental data.

[0074] S1 includes:

[0075] S11, data collection: Use laser radar, ultrasonic sensors and cameras to collect multi-dimensional data including robot motion state data and environmental data, including:

[0076] S111, robot motion state data collection: real-time monitoring of the robot's dynamic state data through the robot's built-in accelerometer and gyroscope, including real-time position coordinates, speed (including linear speed and angular speed) and acceleration data;

[0077] S112, Environmental data collection: Obtain the robot's surrounding environment data through laser radar, ultrasonic sensor and camera, including obstacle detection, road deformation, cracks, settlement and slope. The laser radar is responsible for detecting obstacles and road deformation, the ultrasonic sensor is used to detect road settlement and cracks, and the camera is used to collect road image data and detect surface crack defects;

[0078] S12, data preprocessing: preprocessing the collected multi-dimensional data, including:

[0079] S121, denoising: noise suppression is performed on the collected data, which is expressed as:

[0080]

[0081] Among them, y(i) is the smoothed data, x(j) is the original data, and N is the window size;

[0082] S122, error correction: using a known calibration model to correct the error of the multi-dimensional data, and correcting the error generated by the sensor by combining the calibration information of the robot during initial positioning and the relative position relationship between the robot and the sensor (such as the sensor angle, position, etc.);

[0083] For example, for LiDAR data, an adaptive error correction algorithm (such as linear regression correction) is used to adjust the deviation of the laser point cloud data to reduce the measurement error caused by installation angle deviation and other reasons;

[0084] S123, data filling: For the missing values ​​in the collected multi-dimensional data, the interpolation method is used to fill in the missing values ​​to ensure the integrity of the data. The linear interpolation formula is expressed as:

[0085]

[0086] Among them, y(x) is the interpolated data value, x1, x2 are the horizontal coordinates of the known points, and y1, y2 are the corresponding vertical coordinate values;

[0087] The collected robot motion state data and environmental data are denoised, error corrected and filled to ensure data integrity and accuracy, providing a reliable data foundation for subsequent data analysis, model training and path planning.

[0088] S2 includes:

[0089] S21, data fusion algorithm selection and design: Based on the preprocessed environmental data, select the Kalman filter algorithm to fuse the data from different sensors. The Kalman filter algorithm is used to combine the measurement data of multiple sensors into a unified state estimate. The specific steps are as follows:

[0090] S211, Kalman filter model construction: Set the system state equation and observation equation. The Kalman filter algorithm will estimate the state at each moment based on these equations. Let the state vector be x k , the observation vector is z k , then the basic equation of Kalman filtering is as follows:

[0091] x k =A·x k-1 +B·u k +w k ;

[0092] Among them, A is the state transfer matrix, B is the control input matrix, and u k is the control input, w k The observation equation for process noise is:

[0093] z k =H·x k +v k ;

[0094] S22, Kalman filter process: sensor data is fused through prediction and update steps, including:

[0095] S221, prediction step: predict the current state according to the system state transition model, the formula is:

[0096]

[0097] in, is the predicted state value;

[0098] S222, update step: use the current observation value to correct the predicted value and calculate the Kalman gain K k , and update the state estimate by the following formula:

[0099]

[0100] in, is the predicted covariance matrix, R is the covariance matrix of the observed noise, and the update step formula is:

[0101]

[0102] in, is the updated state estimate;

[0103] S23, data fusion result output: through the calculation result of Kalman filter, the data from different sensors are fused into a unified fused road surface state data, which includes road surface deformation, cracks, settlement, obstacles and slope information, providing a basis for subsequent map generation and disease analysis;

[0104] The Kalman filter algorithm is used to fuse data from different sensors, eliminating the measurement errors of each sensor and generating a unified and accurate road condition estimate, providing an accurate data basis for subsequent map generation and disease analysis.

[0105] S2 also includes:

[0106] S24, road surface state map generation: inputting the fused road surface state data after Kalman filter fusion into the map generation unit to generate a comprehensive road surface state map. The specific steps are as follows:

[0107] S241, atlas construction: based on the fused pavement condition data (including pavement deformation, cracks, settlement and slope), a pavement condition atlas is constructed. The atlas is a two-dimensional or three-dimensional graphical representation, in which each pixel or area represents the pavement health status, for example, the degree of cracks in different areas, the severity of settlement, etc. are represented by color or grayscale values;

[0108] S242, map optimization and calibration: Based on the generated map, the resolution and accuracy of the map are optimized by comparing and calibrating historical data to ensure that the map can truly reflect the road conditions. This step may include the application of image enhancement technology (such as contrast adjustment, smoothing filtering, etc.) to improve the detail accuracy of the map;

[0109] S25, classification and identification of diseased areas: Analyze the generated pavement status map through the convolutional neural network model, identify and classify the diseased areas in the map. The specific steps are as follows:

[0110] S251, convolutional neural network model training: using the images of the marked diseased areas in the historical environmental data, the convolutional neural network model is trained. The training data includes images of different types of road diseases, with various types of diseases (such as cracks, potholes, settlement, etc.) and their severity annotated. The network input is the road state map, and the output is the disease classification result;

[0111] S252, Graph feature extraction: The convolutional neural network extracts features from the graph through multiple layers of convolution and pooling layers, including edge, texture, morphology and other information. These features help identify different types of pavement damage. The convolution layer extracts features from the graph image through filters to obtain detailed features related to the damage.

[0112] S253, classification of damaged areas: Based on the extracted features, the convolutional neural network classifies each area in the atlas into different types of damaged areas (such as cracks, potholes, settlement, etc.). The classification results and damage severity of each area will be marked and assigned a probability value for subsequent processing and decision-making;

[0113] S254, disease area positioning: The convolutional neural network model further outputs the location coordinate information of each disease area, marking the specific location of the disease area in the map, so as to provide an accurate target area for the robot's path planning and subsequent inspections;

[0114] S26, disease area result output: the disease area identified by the convolutional neural network model will be output as a visualization result, presented as a map with disease type, location and severity, which will be used as input for subsequent path planning and decision support;

[0115] By using convolutional neural networks to accurately classify and locate the diseased areas in the map, road diseases can be quickly identified and located. This information provides the inspection robot with accurate disease data, optimizes subsequent path planning and maintenance decisions, and improves inspection efficiency and maintenance accuracy.

[0116] S3 includes:

[0117] S31, path planning algorithm selection: Combine real-time environmental data and known disease types, and use the path planning algorithm to dynamically plan an optimal path for the inspection robot to avoid missing diseased areas and improve inspection efficiency. The specific steps are as follows:

[0118] S311, input of environmental data and disease information: the fused road surface state data and the disease area identification information output by the convolutional neural network model will be used as input data for path planning. The road surface state data includes the overall state information of the road surface (such as road surface deformation, slope, etc.), and the disease information includes the location, type and severity of the identified disease area;

[0119] S312, path planning objective definition: Based on the input data, set the path planning objectives, including minimizing the inspection time, covering all diseased areas, avoiding obstacles, and considering the feasibility of the inspection path. The objective function is expressed in the following form:

[0120]

[0121] Among them, p i Represents a point in the path, distance(p i ,p i+1 ) represents the distance between two adjacent points, penalty(p i ) is a penalty factor related to the severity of the disease, reflecting the importance of diseased areas in path planning;

[0122] S313, select reinforcement learning to optimize path planning: Select the reinforcement learning algorithm to optimize path planning. The goal of reinforcement learning is to gradually learn how to choose the optimal path through the interaction between the agent and the environment. The agent will eventually obtain an optimal path by continuously exploring and evaluating different paths. The specific reinforcement learning framework is:

[0123]

[0124] Among them, Q(p,a) represents the expected return of taking action a in state p, R(p) is the immediate reward of the current state, and γ is the discount factor, which represents the impact of future rewards;

[0125] S32, Path Planning Optimization: Dynamically optimize the path planning through reinforcement learning algorithm to ensure that the inspection robot covers all diseased areas and maximizes the inspection efficiency. The specific optimization process includes:

[0126] S321, Path selection and adjustment: Dynamically adjust the robot's route based on the location and severity of the diseased area to ensure that all diseased areas are covered and invalid inspections are minimized;

[0127] S322, Feedback and Update: After each inspection, the path planning strategy is updated based on the actual inspection results to optimize the selection of subsequent paths. Real-time feedback is provided based on the data collected during the robot inspection process, and the feedback information is input into the reinforcement learning model to further optimize the path planning.

[0128] S3 also includes:

[0129] S33, Inspection strategy optimization: Based on path planning, optimize the inspection strategy to improve the overall inspection effect, including:

[0130] S331, Priority setting of damaged areas: According to the severity, type and location of the damaged area, set the inspection priority of each damaged area. For example, areas with deeper cracks or larger settlements are given higher priority to ensure that these areas are inspected and repaired in time. The priority can be calculated as follows:

[0131] P i=w1·severity(i)+w2·size(i)+w3·proximity(i);

[0132] Among them, severity(i) is the severity of the disease, size(i) is the area of ​​the disease area, and proximity(i) is the distance between the disease and the robot;

[0133] S332, dynamic inspection strategy adjustment: based on the real-time feedback data during the inspection process, such as the robot's motion state data and real-time environmental data during the inspection process, the inspection strategy is dynamically adjusted. The strategy includes re-evaluating the inspection priority of the diseased area, updating the parameters of the path planning algorithm, and adjusting the inspection speed, steering angle and other parameters according to the new inspection data;

[0134] S333, intelligent inspection scheduling: intelligently schedule inspection tasks according to the priority of the diseased area and the current inspection status of the robot to ensure the rationality of task allocation and maximize inspection efficiency. For example, if the robot detects a new diseased area during the inspection, the inspection path and task allocation will be automatically adjusted according to the priority of the area.

[0135] S4 includes:

[0136] S41, fuzzy control input variable setting: according to the fused road state data, the input variables of the fuzzy control algorithm are set, and the input variables include the robot motion state (such as current position, speed, acceleration) and real-time environmental data (such as road deformation, slope, obstacle detection results, etc.);

[0137] S42, design and optimization of fuzzy control rule base: set the fuzzy control rule base according to the real-time road state data. The rule base defines the relationship between input variables and output variables (such as travel speed and steering angle). The rules are based on the experience accumulation and simulation of complex road environments. The following control logic is designed:

[0138] If the road slope is steep and there are obstacles, the output reduces speed and adjusts the steering angle;

[0139] If the road is in good condition and clear of obstacles, increase your speed;

[0140] If there is a diseased area ahead, adjust the steering angle to bypass the area;

[0141] S43, fuzzy control output calculation: The travel speed and steering angle are calculated by the fuzzy control algorithm and output to the control unit. The control unit generates a control signal by using the fuzzy reasoning method according to the set control rules and input data. The process includes the following steps:

[0142] S431, fuzzy reasoning: converting the input real-time data into fuzzy sets, and calculating the fuzzy values ​​of each output variable (speed, steering angle) through fuzzy reasoning;

[0143] S432, defuzzification processing: defuzzify the fuzzy output to obtain specific control values, namely the travel speed and steering angle of the robot.

[0144] S4 also includes:

[0145] S44, execution of control signals and adjustment of robot behavior: According to the generated control signals, the robot's travel speed and steering angle are dynamically adjusted to adapt to the real-time changing road environment. The specific steps are as follows:

[0146] S441, speed adjustment: adjust the robot's travel speed according to the output of fuzzy control. The specific method is as follows:

[0147] When the road surface is flat and without obstacles, increase the travel speed and improve the inspection efficiency;

[0148] When there are obstacles, diseased areas or large slopes, the robot slows down to ensure that it can complete the inspection task safely and accurately;

[0149] S442, steering angle adjustment: adjusting the steering angle of the robot according to real-time environmental information;

[0150] When obstacles or diseased areas are detected, the robot's direction of travel is adjusted through a fuzzy control algorithm to avoid collisions and ensure comprehensive inspections;

[0151] By dynamically adjusting the steering angle, the robot can flexibly respond to changes in road conditions;

[0152] S45, real-time feedback and adaptive adjustment: According to the real-time feedback, adjust the fuzzy control strategy so that the robot can dynamically adapt to complex environmental changes during the inspection process. The specific steps are as follows:

[0153] S451, feedback information update: During the inspection process, the robot motion state data and environmental data are continuously collected, and these data are fed back to the control unit in real time;

[0154] S452, optimize control strategy: based on feedback information, optimize the fuzzy control rule base and control algorithm to achieve adaptive adjustment of the robot's travel speed and steering angle.

[0155] S5 includes:

[0156] S51, Maintenance Recommendations: Generate targeted pavement maintenance recommendations based on the type and severity of the damage and the pavement condition. The recommendations will take into account the following factors:

[0157] S511, Maintenance priority determination: Determine the priority of maintenance work based on the severity of the disease, the area where it occurs, the importance of the road section, etc. For example, for areas with severe cracks and severe settlement, it is recommended to give priority to repair;

[0158] S512, maintenance plan design: provide corresponding maintenance plans according to the damage type and road condition. For example, crack treatment may include crack filling, pothole repair may include local reinforcement, etc.

[0159] S52, Emergency Response Plan: In special circumstances, an emergency response plan is generated based on real-time environmental data. This plan is usually used to deal with sudden major road damage, traffic accidents or sudden weather changes. The specific steps include:

[0160] S521, Emergency Repair Plan: Provide emergency repair suggestions based on the severity of the damaged area, giving priority to ensuring traffic safety and smooth flow;

[0161] S522, Traffic diversion and reminder: For areas with serious road damage, emergency measures such as temporary traffic control, road closures, detours, etc. are proposed to ensure traffic safety;

[0162] S53, real-time update and adjustment: As road conditions change, maintenance recommendations and emergency response plans are updated in real time to ensure timely and accurate road management;

[0163] S531, real-time data feedback: collect new data generated during the inspection process (such as newly discovered diseased areas, road surface changes, etc.), and update maintenance and emergency response plans based on the newly collected data to ensure the dynamic adaptability of the plans.

[0164] like Figure 2 As shown, a road automatic inspection robot control device is used to implement the above-mentioned road automatic inspection robot control method, including the following modules:

[0165] Data acquisition module: uses lidar, ultrasonic sensors and cameras to collect multi-dimensional data, and pre-processes the collected multi-dimensional data, including denoising, error correction and data filling;

[0166] Data fusion and analysis module: Based on the pre-processed environmental data, the Kalman filter algorithm is used for data fusion, the data from different sensors are merged to generate a comprehensive road condition map, and the convolutional neural network model is used to classify, identify and locate the diseased areas in the map;

[0167] Path planning and inspection strategy optimization module: Combined with machine learning algorithms, based on real-time environmental data and known disease types, the inspection robot's path is dynamically optimized using reinforcement learning algorithms to avoid missing diseased areas and improve inspection efficiency;

[0168] Real-time feedback and control module: Based on the real-time robot motion state data and environmental data, the fuzzy control algorithm is used to dynamically adjust the robot's travel speed and steering angle to achieve adaptive inspection control of complex road environments;

[0169] Decision support module: Provides decision support and generates targeted pavement maintenance recommendations and emergency response plans based on real-time environmental data.

[0170] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0171] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A control method for an automatic road inspection robot, characterized in that: The following steps are involved: S1, data collection and preprocessing: using laser radar, ultrasonic sensor and camera to collect multi-dimensional data, the multi-dimensional data includes robot motion state data and environment data, wherein; The robot motion state data includes the robot's current position and speed data; The environmental data includes obstacles, road deformation, cracks, settlement and slope; And preprocess the collected multi-dimensional data, including denoising, error correction and data filling; S2, data fusion and analysis: Based on the pre-processed environmental data, the Kalman filter algorithm is used for data fusion to merge the data from different sensors to generate a comprehensive road condition map, and the convolutional neural network model is used to classify, identify and locate the diseased areas in the map; S3, path planning and inspection strategy optimization: Combined with machine learning algorithms, based on real-time environmental data and known disease types, reinforcement learning algorithms are used to dynamically optimize the inspection robot's path to avoid missing diseased areas and improve inspection efficiency; S4, real-time feedback and control: Based on the real-time robot motion state data and environmental data, the fuzzy control algorithm is used to dynamically adjust the robot's travel speed and steering angle to achieve adaptive inspection control of complex road environments; S5, Decision Support: Provide decision support to generate targeted pavement maintenance recommendations and emergency response plans based on real-time environmental data.

2. A method for controlling an automatic road inspection robot according to claim 1, characterized in that: The S1 includes: S11, data collection: Use laser radar, ultrasonic sensors and cameras to collect multi-dimensional data including robot motion state data and environmental data, including: S111, robot motion state data collection: real-time monitoring of the robot's dynamic state data through the robot's built-in accelerometer and gyroscope, including real-time position coordinates, speed and acceleration data; S112, Environmental Data Collection: Obtain the robot's surrounding environment data through lidar, ultrasonic sensors and cameras, including obstacle detection, road deformation, cracks, settlement and slope; S12, data preprocessing: preprocessing the collected multi-dimensional data, including: S121, denoising: performing noise suppression processing on the collected data; S122, error correction: using a known calibration model to perform error correction on the multi-dimensional data; S123, data filling: fill in the missing values ​​in the collected multi-dimensional data using interpolation method.

3. A method for controlling an automatic road inspection robot according to claim 2, characterized in that: The S2 includes: S21, data fusion algorithm selection and design: Based on the preprocessed environmental data, the Kalman filter algorithm is selected to fuse the data from different sensors; S22, Kalman filter process: sensor data is fused through prediction and update steps; S23, data fusion result output: through the calculation result of the Kalman filter, the data from different sensors are fused into a unified fused road surface state data.

4. A method for controlling an automatic road inspection robot according to claim 3, characterized in that: The S2 further includes: S24, road surface state map generation: inputting the fused road surface state data after Kalman filter fusion into a map generation unit to generate a comprehensive road surface state map; S25, classification and identification of diseased areas: Analyze the generated pavement status map through a convolutional neural network model to identify and classify the diseased areas in the map; S26, disease area result output: the disease area identified by the convolutional neural network model will be output as a visualization result, presented as a map marked with disease type, location and severity.

5. A method for controlling an automatic road inspection robot according to claim 4, characterized in that: The S3 includes: S31, path planning algorithm selection: combining real-time environmental data and known disease types, dynamically planning an optimal path for the inspection robot through the path planning algorithm to avoid missing diseased areas and improve inspection efficiency; S32, Path planning optimization: Dynamically optimize path planning through reinforcement learning algorithm.

6. A method for controlling an automatic road inspection robot according to claim 5, characterized in that: The S3 further includes: S33, Inspection strategy optimization: Based on path planning, optimize the inspection strategy to improve the overall inspection effect, including: S331, setting the priority of the diseased area: setting the inspection priority of each diseased area according to the severity, type and location of the diseased area; S332, dynamic inspection strategy adjustment: dynamically adjust the inspection strategy based on real-time feedback data during the inspection process; S333, intelligent inspection scheduling: intelligently schedule inspection tasks based on the priority of the diseased area and the robot’s current inspection status.

7. A method for controlling an automatic road inspection robot according to claim 6, characterized in that: The S4 includes: S41, fuzzy control input variable setting: setting the input variables of the fuzzy control algorithm according to the fused road surface state data; S42, fuzzy control rule base design and optimization: set the fuzzy control rule base according to the real-time road status data; S43, fuzzy control output calculation: the travel speed and steering angle are calculated by the fuzzy control algorithm and output to the control unit. The control unit generates a control signal by using the fuzzy reasoning method according to the set control rules and input data.

8. A method for controlling an automatic road inspection robot according to claim 7, characterized in that: The S4 further comprises: S44, execution of control signals and adjustment of robot behavior: dynamically adjusting the travel speed and steering angle of the robot according to the generated control signals; S45, real-time feedback and adaptive adjustment: According to the real-time feedback, the fuzzy control strategy is adjusted to enable the robot to dynamically adapt to complex environmental changes during the inspection process.

9. A method for controlling an automatic road inspection robot according to claim 8, characterized in that: The S5 includes: S51, Maintenance Recommendations: Generate targeted pavement maintenance recommendations based on the type, severity and pavement condition of the disease; S52, emergency response plan: Generate emergency response plan based on real-time environmental data; S53, real-time update and adjustment: maintenance recommendations and emergency response plans are updated in real time as road conditions change.

10. A control device for an automatic road inspection robot, used to implement a control method for an automatic road inspection robot as claimed in any one of claims 1 to 9, characterized in that: Includes the following modules: Data acquisition module: uses lidar, ultrasonic sensors and cameras to collect multi-dimensional data, and pre-processes the collected multi-dimensional data, including denoising, error correction and data filling; Data fusion and analysis module: Based on the pre-processed environmental data, the Kalman filter algorithm is used for data fusion, the data from different sensors are merged to generate a comprehensive road condition map, and the convolutional neural network model is used to classify, identify and locate the diseased areas in the map; Path planning and inspection strategy optimization module: Combined with machine learning algorithms, based on real-time environmental data and known disease types, the inspection robot's path is dynamically optimized using reinforcement learning algorithms to avoid missing diseased areas and improve inspection efficiency; Real-time feedback and control module: Based on the real-time robot motion state data and environmental data, the fuzzy control algorithm is used to dynamically adjust the robot's travel speed and steering angle to achieve adaptive inspection control of complex road environments; Decision support module: Provides decision support and generates targeted pavement maintenance recommendations and emergency response plans based on real-time environmental data.

Images