An automated wireless road surface crack detection robot based on ground-penetrating radar and vision

Abstract

This invention proposes an automated wireless road surface crack detection robot based on a combination of ground-penetrating radar (GPR) and vision, comprising a hardware system and a software system. The hardware system includes a GPR and a vehicle body. A camera is mounted on the front of the vehicle body, and a first control motherboard is mounted on the vehicle body. The GPR is separate from the vehicle body but connected wirelessly, guided by the vehicle body to perform cross-scanning detection of cracks, and its stability and responsiveness are balanced according to the crack density. In the software system, a path planning module, based on the analysis results of the crack recognition module, guides the vehicle body to automatically move along the crack trend via the first control motherboard. This invention uses visual analysis of the road crack direction to sample two-dimensional crack information, and automatically plans the movement path in conjunction with the control motherboard. As the vehicle moves along the crack, the GPR, guided by the vehicle body via wireless connection, samples the crack depth information to construct a three-dimensional crack model. This achieves simple and efficient detection of road surface cracks.

Description

Technical Field

[0001] This invention relates to the field of road surface crack detection technology, specifically to an automated wireless road surface crack detection robot based on a combination of ground-penetrating radar and vision. Background Technology

[0002] With the increasing age of roads and the rise in traffic volume, cracks in road surfaces are a common phenomenon. Cracks not only affect the smoothness and safety of driving, but can also allow moisture to seep into the roadbed, leading to more serious road damage such as ruts and potholes. Timely detection and repair of cracks are crucial measures to extend the service life of roads and ensure traffic safety.

[0003] Traditional methods for detecting road cracks mainly rely on manual inspections, where workers drive or walk to inspect and manually record the type, location, size, and severity of cracks. While this method can detect road cracks to some extent, it suffers from drawbacks such as low efficiency, large human error, limited detection coverage, and inconsistent data recording. Automated crack detection equipment can save time and labor, and improve detection efficiency.

[0004] Existing road crack detection devices are mostly rigidly coupled structures that are manually driven or pushed. Due to the rigid coupling of the structure, the detection vehicle needs to perform frequent back-and-forth turning operations when scanning for three-dimensional information reconstruction of cracks, as the ground-penetrating radar needs to scan vertically to the crack. However, manual operation has limited precision and is prone to false scans and misidentifications. Furthermore, the rigid coupling of the crack detection vehicle increases the difficulty of path planning and increases the computational power consumption. Summary of the Invention

[0005] For the reasons mentioned above, this invention provides an automated wireless road surface crack detection robot based on the combination of ground penetrating radar and vision, which solves the problems of high-frequency reciprocating turning operations, high operational difficulty, and low detection accuracy of existing rigid coupling detection vehicles.

[0006] To achieve the above objectives, the present invention adopts the following solution:

[0007] The first aspect of this invention proposes an automated wireless road surface crack detection robot based on a combination of ground-penetrating radar and vision, comprising a hardware system and a software system. The hardware system includes a ground-penetrating radar and a vehicle body. The radar includes a radar antenna, a radar main unit, and ranging wheels. A camera is mounted on the front of the vehicle body, and a first control motherboard, an attitude sensor, and a power supply battery are mounted on the vehicle body. Movable wheel assemblies are mounted on both sides of the vehicle body. The camera, the movable wheel assemblies mounted on the vehicle body, and the attitude sensor are electrically connected to the first control motherboard.

[0008] The ground-penetrating radar is separated from the vehicle body and is equipped with a second control motherboard, attitude sensor and power supply battery. Movable wheel assemblies are installed on both sides. The radar antenna is installed above the movable wheel assemblies, the radar host is installed above the radar antenna, and the ranging wheel is installed behind the radar antenna. The ranging wheel, the attitude sensor installed on the ground-penetrating radar, the power supply battery, the movable wheel assembly and the second control motherboard are electrically connected. The first control motherboard and the second control motherboard communicate wirelessly through the WIFI wireless module. The ground-penetrating radar is guided by the vehicle body to complete the cross-scan detection of cracks.

[0009] The software system is embedded on the first control motherboard and includes:

[0010] The camera image acquisition module acquires image data from the camera.

[0011] The crack recognition module identifies and analyzes cracks in images captured by the camera.

[0012] The path planning module analyzes and calculates the routes of the vehicle body and ground-penetrating radar based on image data.

[0013] Preferably, the ground-penetrating radar is guided by the vehicle body to complete the cross-scan detection of cracks, and the cross-scan direction is perpendicular to the front direction of the vehicle body.

[0014] Preferably, the path planning module, based on the analysis results of the crack recognition module, uses the first control motherboard to automatically move the vehicle body along the crack trend. The crack recognition module frames the crack in the image with a rectangle and constructs the x-axis zero point with the center line of the image. The path planning module analyzes the x-axis coordinates of the center point of the rectangle in the image. If the coordinates are positive, it determines that the crack trend is to the right, and the vehicle body turns right until the center point of the crack frame is located at the x-axis zero point. If the coordinates are negative, it determines that the crack trend is to the left, and the vehicle body turns left until the center point of the crack frame is located at the x-axis zero point.

[0015] Based on the aforementioned wireless road crack detection robot, a second aspect of this invention proposes a road crack detection method, comprising the following steps:

[0016] S1: The camera collects road surface image information;

[0017] S2: Road surface image information is transmitted to the crack recognition module;

[0018] S3: When a crack is detected, the crack detection module obtains the length and width information of the crack and analyzes and plans the vehicle's travel route through the path planning module.

[0019] S4: When the vehicle is moving, the ground-penetrating radar continuously scans the cracks in a cross direction perpendicular to the direction of the vehicle's front.

[0020] S5: When the vehicle encounters an inflection point during its journey, and the inflection point disappears from the camera image, the ground-penetrating radar fixes the memory of the vehicle's front orientation at that moment, continues to scan the crack along that direction, and the vehicle's front turns according to the inflection point direction and begins to record the distance traveled.

[0021] S6: When the travel distance reaches the distance L between the camera and the ground-penetrating radar, the ground-penetrating radar continuously updates the vehicle's front-end orientation until the vehicle's front end reaches the turning point again.

[0022] S7: If the next inflection point has disappeared from the camera image before traveling to a distance L, it is determined that the inflection point area is too small and the ground penetrating radar can directly scan and cover it, simplifying it into a straight crack, and repeating step S4.

[0023] S8: Integrate the three-dimensional information of the crack to form a three-dimensional model of the corresponding crack.

[0024] Preferably, in step S6, the cross-scanning of the ground penetrating radar adopts cascade PID control combined with road surface crack information. The road surface crack information is collected by the camera, and crack image information is obtained through image processing. The crack density coefficient b is calculated and mapped to the dynamic adjustment factor k of the control parameter. It is then substituted into each link of the dual-loop PID for adjustment. When b increases, k decreases, reducing the cross-travel amplitude, slowing down the response, and reducing the error rate of misidentifying adjacent cracks. When b decreases, k increases, allowing for a larger cross-travel amplitude and a faster response.

[0025] Preferably, the crack density coefficient b = s -6 Where s is the ratio of crack skeleton pixels to total image pixels, and is a dynamic adjustment factor. Where t is the adjustment coefficient of the dynamic adjustment factor, which is selected according to the order of magnitude of each parameter of the actual PID, and is used to adjust the order of magnitude of the dynamic adjustment factor k.

[0026] The obtained k is substituted into the calculations of the inner and outer loops of the PID controller for adjustment. The basic single-loop PID algorithm is as follows:

[0027]

[0028] in:

[0029] u(k): The result of the PID algorithm, i.e., the control quantity; K p K: The adjustment coefficient of the proportional term, used to adjust the performance of the PID proportional element; i K: The adjustment coefficient for the integral term, used to adjust the performance of the PID integral element; d : The adjustment coefficient of the derivative term, which adjusts the performance of the PID derivative element; e(k) is the error, i.e., the target value minus the current state value of the controlled object; [e(k)-e(k-1)] is the current error minus the previous error;

[0030] Preferably, the obtained k is substituted into the calculations of the inner and outer loops of the PID controller for adjustment, wherein:

[0031] The outer loop adjustment process is as follows:

[0032] 1) Generate the baseline lateral displacement based on the path planning, and multiply it by k to dynamically adjust the target:

[0033] target1 = target × k, where target1 is the target value of the outer loop PID, and target is the baseline value of the target value of the outer loop PID;

[0034] 2) Add dynamic limiting to the outer loop PID output to adjust the target speed of the speed loop PID:

[0035] output 1∈[-V max×k,V max×k];

[0036] Where output is the outer loop PID output value, which serves as the target value for the inner loop PID; Vmax is the target speed limit value for the inner loop speed PID, used to limit the range of the inner loop PID target speed.

[0037] The inner loop adjustment process is as follows:

[0038] The aggressiveness of the response is adjusted according to the degree of crack density.

[0039] K p _speed=k p ×k;

[0040] K i _speed=k i ×k;

[0041] Where: Kp_speed is the coefficient of the proportional control term of the inner loop PID, kp is the reference value of the coefficient of the proportional control term of the inner loop PID, Ki_speed is the coefficient of the integral control term of the inner loop PID, and ki is the reference value of the coefficient of the integral control term of the inner loop PID.

[0042] Preferably, in step S6, the cross-scanning of the ground-penetrating radar is performed by segmenting the crack at the crack inflection point, and then performing a delayed cross-scanning based on the distance L between the inflection point and the distance L from the front-facing camera to the ground-penetrating radar. Specifically:

[0043] When a crack in the camera image shows an inflection point with a curvature greater than a threshold l, it is identified and recorded. As the vehicle approaches the inflection point and disappears from the camera image, the ground-penetrating radar (GPR) fixes the vehicle's front-end orientation at that moment and continues to scan the crack along the perpendicular direction of that direction. The vehicle's front-end turns according to the inflection point direction and begins recording the distance traveled. When the traveled distance reaches the distance L between the camera and the GPR, the GPR continuously updates and memorizes the vehicle's front-end orientation until the vehicle's front-end reaches the inflection point again. If the next inflection point disappears from the camera image before traveling a distance L, it is determined that the inflection point area is too small, and the GPR can directly scan and cover it, simplifying it to a straight crack. The GPR immediately and continuously updates the vehicle's front-end orientation and scans the next segment of the crack along the perpendicular direction of the vehicle's front-end orientation.

[0044] Beneficial effects: Compared with existing technologies, this invention innovatively adopts "dynamic path decoupling control technology". Through the electromechanical separation architecture of the vehicle navigation unit and the ground penetrating radar module, an intelligent collaborative operation system is constructed, which completely revolutionizes the traditional road crack recognition mode and reconstructs the spatial-temporal dimension of road crack detection. Through spatial decoupling control, the vehicle's travel direction and the crack scanning direction are decoupled and free. The vehicle only needs to travel automatically along the crack direction and guide the ground penetrating radar to the correct route, which simplifies the travel route and improves the detection efficiency. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of an automated wireless road surface crack detection robot based on the combination of ground penetrating radar and vision according to the present invention.

[0046] Figure 2 This is a schematic diagram of the control process of the automatic road crack wireless detection robot of the present invention;

[0047] Figure 3 This is a flowchart of an automatic wireless road surface crack detection robot method based on the combination of ground penetrating radar and vision, according to the present invention.

[0048] Figure 4 In this embodiment of the invention, the ratio of the pixel value of the sparser crack after image processing to the total pixel value of the image is given.

[0049] Figure 5 In this embodiment of the invention, the ratio of the pixel value of the denser crack after image processing to the total pixel value of the image is given.

[0050] Figure 6 This is a longitudinal profile image transmitted in real time by ground-penetrating radar during camera capture.

[0051] The image is labeled as follows:

[0052] 1. Ground penetrating radar; 1a. Radar antenna; 1b. Radar main unit; 1c. Ranging wheel; 2. Camera; 3. Moving wheel assembly; 4. Vehicle body; 5a. First control motherboard; 5b. Second control motherboard; 6. Power supply battery; 7. WIFI wireless module; 8. Attitude sensor; 9. Camera image acquisition module; 10. Crack recognition module; 11. Path planning module. Detailed Implementation

[0053] 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.

[0054] like Figure 1 , Figure 2 As shown, the present invention proposes an automated wireless road surface crack detection robot based on the combination of ground penetrating radar and vision, including a hardware system and a software system. The hardware system includes a ground penetrating radar 1 and a vehicle body 4. A camera 2 is installed at the front end of the vehicle body 4. A first control motherboard 5a, an attitude sensor 8, and a power supply battery 6 are mounted on the vehicle body 4. Moving wheel assemblies 3 are installed on both sides of the vehicle body 4. The camera 2, the moving wheel assemblies 3 installed on the vehicle body 4, and the attitude sensor 8 are electrically connected to the first control motherboard 5a.

[0055] The ground-penetrating radar 1 is separated from the vehicle body 4. It is equipped with a second control motherboard 5b, radar antenna 1a, radar host 1b, ranging wheel 1c, attitude sensor 8, and power supply battery 6. Moving wheel assemblies 3 are installed on both sides. The ranging wheel 1c is connected to the radar antenna 1a through a torsion spring device. The radar antenna 1a is connected to the radar host 1b through a coaxial cable. The ranging wheel 1c, the attitude sensor 8 installed on the ground-penetrating radar 1, the power supply battery 6, the moving wheel assembly 3 are electrically connected to the second control motherboard 5b. The first control motherboard 5a and the second control motherboard 5b communicate wirelessly through the WIFI wireless module 7. The ground-penetrating radar 1 is guided by the vehicle body 4 to complete the cross-scan detection of cracks.

[0056] Among them, camera 2 is used to capture images of road surface cracks, moving wheel assembly 3 drives the vehicle body and ground-penetrating radar to move, ranging wheel 1c transmits the vehicle body's moving distance in real time, power supply battery 6 provides power to the device, WIFI wireless module 7 is used to realize wireless communication between the vehicle body and ground-penetrating radar, and attitude sensor 8 is used to determine the orientation of the vehicle body and ground-penetrating radar.

[0057] In this embodiment, the ground-penetrating radar 1 is separated from the vehicle body and wirelessly connected. Guided by the front of the vehicle, it completes the cross-scan detection of cracks and adjusts the stability and responsiveness of the PID according to the density of cracks.

[0058] The software system is embedded on the first control motherboard (5a), including:

[0059] Camera image acquisition module 9 acquires image data from the camera;

[0060] The crack recognition module 10 performs crack recognition and analysis on the images captured by the camera 2.

[0061] The path planning module 11 analyzes and calculates the route of the vehicle body 4 and the ground-penetrating radar 1 from the image data.

[0062] In this embodiment, the path planning module 11 analyzes the crack image collected by the camera image acquisition module 9 based on the crack recognition module 10 and controls the main board to enable the vehicle to automatically travel along the crack trend. The crack recognition module 10 frames the crack in the image with a rectangular frame and constructs the x-axis zero point with the center line of the image. The path planning module 11 analyzes the x-axis coordinate of the center point of the frame in the image. If the coordinate is positive, it is determined that the crack trend is to the right, and the vehicle turns right until the center point of the crack frame is located at the x-axis zero point. If the coordinate is negative, it is determined that the crack trend is to the left, and the vehicle turns left until the center point of the crack frame is located at the x-axis zero point, so that the vehicle can automatically travel along the crack trend.

[0063] For example, when the vehicle reaches the turning point, it obtains the coordinates x1 of the center point x of the next crack frame. If x1>0, the main board controls the vehicle to continuously execute the right turn command, while continuously updating the coordinates of the center point of the overall crack frame in the camera image until x1=0. Then, the main board controls the vehicle to execute the straight-go command, the vehicle stops rotating, and travels in a straight line.

[0064] like Figure 3 As shown, the road surface crack detection method based on the above-mentioned detection robot includes the following steps:

[0065] S1: Camera 2 collects road surface image information;

[0066] S2: Road surface image information is transmitted to crack recognition module 10;

[0067] S3: When a crack is detected, the crack detection module 10 obtains the length and width information of the crack and analyzes and plans the travel route of the vehicle body 4 through the path planning module 11.

[0068] S4: When the vehicle body 4 is moving, the ground-penetrating radar 1 continuously scans the crack in a direction perpendicular to the front direction of the vehicle body 4. That is, the vehicle body 4 moves along the direction of the crack, and the ground-penetrating radar 1 performs a "z"-shaped reciprocating scan along the path of the vehicle body 4.

[0069] S5: When the vehicle encounters an inflection point during its journey, and the inflection point disappears from the image of camera 2, ground-penetrating radar 1 fixes the memory of the direction of the front of the vehicle body 4 at this time, and continues to scan the crack along that direction. The front of the vehicle turns according to the direction of the inflection point and begins to record the distance traveled.

[0070] S6: When the travel distance reaches the distance L between the camera 2 and the ground-penetrating radar 1, the ground-penetrating radar 1 continuously updates the current orientation of the vehicle body 4 until the vehicle head reaches the turning point again.

[0071] S7: If the next inflection point has disappeared from the image of camera 2 before traveling to a distance L, it is determined that the inflection point area is too small. Ground penetrating radar 1 can directly scan and cover it, simplifying it into a straight crack. Repeat step S4.

[0072] S8: Integrate the three-dimensional information of the crack to form a three-dimensional model of the corresponding crack.

[0073] In this embodiment, the cross-scanning of the ground-penetrating radar 1 in S6 adopts cascade PID control combined with road surface crack information. The camera 2 collects road surface crack information, and image processing yields crack image information. A crack density coefficient b is calculated and mapped to a dynamic adjustment factor k of the control parameters. This k is then substituted into the various stages of the dual-loop PID control for adjustment. When b increases (cracks are dense), k decreases, thereby reducing the cross-scanning amplitude, slowing the response, and reducing the error rate of misidentifying adjacent cracks. When b decreases (cracks are sparse), k increases, allowing for a larger cross-scanning amplitude and a faster response. b is the crack density coefficient, calculated from the crack pixel ratio after image processing, specifically as follows... Figure 4 , Figure 5 As shown, Figure 4 s1 is the ratio of the pixel values ​​of the cracks after image processing to the total pixel values ​​of the image under relatively sparse conditions. Figure 5 s² is the ratio of the pixel values ​​of the cracks after image processing to the total pixel values ​​of the image, where the cracks are relatively dense. Figure 6 The longitudinal profile image transmitted in real time by ground-penetrating radar 1 during the shooting process of camera 2.

[0074] Formula for calculating s:

[0075]

[0076] Substituting s1 and s2 into the following calculation yields the corresponding crack density coefficient b:

[0077] b = s -6

[0078] We can obtain: b1 = 0.68^-6 = 10.11, b2 = 0.68^-6 = 2.15

[0079] Then, b is converted into the dynamic adjustment factor k using the following formula:

[0080]

[0081] in:

[0082] k: Dynamic adjustment parameter, used to adjust the parameters in each PID controller;

[0083] b: Crack density coefficient, which reflects the density of cracks in the image; the larger the coefficient, the denser the cracks.

[0084] t: The adjustment coefficient of the dynamic adjustment factor k, which is selected according to the order of magnitude of the actual PID parameters, and is used to adjust the dynamic adjustment factor k to a suitable order of magnitude.

[0085] The obtained k is then substituted into the calculations of the inner and outer loops of the PID controller for adjustment:

[0086]

[0087] in:

[0088] u(k): The calculation result of the PID algorithm, i.e., the control quantity.

[0089] K p The proportional term adjustment coefficient, used to adjust the performance of the PID proportional component;

[0090] K i The adjustment coefficient for the integral term, used to adjust the performance of the PID integral element;

[0091] K d The adjustment coefficient of the derivative term, which adjusts the performance of the PID derivative component;

[0092] e(k): Error, i.e., target value - current state value of the controlled object, [e(k)-e(k-1)]: current error minus the previous error.

[0093] Outer loop regulation includes:

[0094] 1. Generate a baseline lateral displacement (crossing amplitude) based on path planning, and multiply it by k to dynamically adjust the target:

[0095] target 1 = target × k

[0096] in:

[0097] target1: The target value of the outer loop PID;

[0098] target: The baseline value of the outer loop PID target value, which can be flexibly adjusted according to application requirements;

[0099] k: Dynamic adjustment factor, used to adjust the parameters in each PID controller;

[0100] 2. Add dynamic limiting to the outer loop PID output to adjust the target speed of the speed loop PID:

[0101] output 1∈[-V max×k,V max×k]

[0102] in:

[0103] output: The output value of the outer loop PID, which serves as the target value for the inner loop PID;

[0104] Vmax: The target speed limit value of the inner loop speed PID, which limits the range of the inner loop PID target speed;

[0105] k: Dynamic adjustment factor, used to adjust the parameters in each PID controller;

[0106] Inner loop adjustment:

[0107] The aggressiveness of the response is adjusted according to the degree of crack density:

[0108] K p _speed=k p ×k

[0109] Ki_speed=ki×k

[0110] in:

[0111] Kp_speed: Coefficient of the proportional control term of the inner loop PID controller;

[0112] kp: Reference value of the coefficient of the proportional control term of the inner loop PID;

[0113] Ki_speed: Inner loop PID integral control term coefficient;

[0114] ki: Reference value of the integral control term coefficient of the inner loop PID controller;

[0115] The cross-scanning of the ground-penetrating radar 1 in S6 is based on segmenting the crack at its inflection point. A delayed cross-scan is performed using the inflection point location and the distance L between the front-facing camera and the ground-penetrating radar. Specifically:

[0116] When a crack in the image from camera 2 exhibits an inflection point with a curvature exceeding a set threshold limit, it is identified and recorded. As the vehicle approaches the inflection point and it disappears from the image from camera 2, ground-penetrating radar 1 fixes the vehicle's orientation at that moment and continues to cross-scan the crack along the perpendicular direction of that direction. The vehicle turns according to the inflection point direction and begins recording the distance traveled. When the traveled distance reaches a distance L between camera 2 and ground-penetrating radar 1, ground-penetrating radar 1 continuously updates the vehicle's orientation until the vehicle reaches the inflection point again. If the next inflection point disappears from the image from camera 2 before traveling a distance L, it is determined that the inflection point area is too small, and ground-penetrating radar can directly scan and cover it, simplifying it to a straight crack. It immediately and continuously updates the vehicle's orientation and scans the next segment of the crack along the perpendicular direction of the vehicle's orientation. This achieves the time difference between ground-penetrating radar scanning and vehicle identification, compensating for the time difference between camera 2 identifying the crack and ground-penetrating radar 1 reaching the crack location.

[0117] This invention uses visual analysis to sample the two-dimensional information of road surface cracks. Combined with a control board, it automatically plans the vehicle's movement path. As the vehicle travels along the crack, a ground-penetrating radar, guided wirelessly by the vehicle body, samples the crack depth information to construct a three-dimensional model of the crack. This invention simplifies vehicle movement paths and enables the construction of a three-dimensional crack model, achieving convenient and efficient detection of road surface cracks.

[0118] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A method for detecting road cracks using an automated wireless road crack detection robot based on a combination of ground-penetrating radar and vision, characterized in that, This method employs an automated wireless road surface crack detection robot based on the combination of ground penetrating radar and vision, including a hardware system and a software system. The hardware system includes a ground penetrating radar (1) and a vehicle body (4). A camera (2) is installed at the front end of the vehicle body (4). A first control motherboard (5a), an attitude sensor (8), and a power supply battery (6) are mounted on the vehicle body (4). Moving wheel assemblies (3) are installed on both sides of the vehicle body (4). The camera (2), the moving wheel assemblies (3) installed on the vehicle body (4), and the attitude sensor (8) are electrically connected to the first control motherboard (5a). The ground-penetrating radar (1) is separated from the vehicle body (4). It is equipped with a second control motherboard (5b), an attitude sensor (8), and a power supply battery (6). Moving wheel assemblies (3) are installed on both sides. The attitude sensor (8), power supply battery (6), and moving wheel assembly (3) installed on the ground-penetrating radar (1) are electrically connected to the second control motherboard (5b). The first control motherboard (5a) and the second control motherboard (5b) communicate wirelessly through the WIFI wireless module (7). The ground-penetrating radar (1) is guided by the vehicle body (4) to complete the cross-scan detection of cracks. The software system is embedded on the first control motherboard (5a) and includes: The camera image acquisition module (9) acquires image data from the camera; The crack recognition module (10) performs crack recognition and analysis on the images captured by the camera (2); The path planning module (11) analyzes and calculates the route of the vehicle body (4) and the ground-penetrating radar (1) based on the image data; The method includes the following steps: S1: Camera (2) collects road surface image information; S2: Road surface image information is transmitted to the crack recognition module (10); S3: When a crack is detected, the crack detection module (10) obtains the length and width information of the crack and analyzes and plans the travel route of the vehicle body (4) through the path planning module (11); S4: When the vehicle body (4) is moving, the ground-penetrating radar (1) continuously scans the cracks in a direction perpendicular to the front direction of the vehicle body (4); S5: When the vehicle encounters an inflection point during its journey, when the inflection point disappears from the image of the camera (2), the ground-penetrating radar (1) fixes the memory of the direction of the front of the vehicle (4) at this time, and continues to scan the crack along the direction perpendicular to the direction of the front of the vehicle. The front of the vehicle turns according to the direction of the inflection point and begins to record the distance traveled. S6: When the travel distance reaches the distance L between the camera (2) and the ground-penetrating radar (1), the ground-penetrating radar (1) continuously updates the current vehicle head orientation (4) until the vehicle head reaches the turning point again. S7: If the next inflection point has disappeared in the image of the camera (2) before traveling to a distance L, it is determined that the inflection point area is too small, and the ground-penetrating radar (1) can directly scan and cover it, simplifying it into a straight crack, and repeating step S4; S8: Integrate the three-dimensional information of the crack to form a three-dimensional model of the corresponding crack; In step S6, the cross-scanning of the ground penetrating radar (1) adopts cascade PID control combined with road surface crack information. The camera (2) collects road surface crack information, obtains crack image information through image processing, calculates the crack density coefficient b, maps b to the dynamic adjustment factor k of the control parameter, and adjusts it in each link of the double-loop PID. When b increases, k decreases, reducing the cross-travel amplitude, slowing down the response, and reducing the error rate of misidentifying adjacent cracks; when b decreases, k increases, allowing for a larger cross-travel amplitude and a faster response.

2. The method for detecting road surface cracks according to claim 1, characterized in that, Crack density coefficient Where s is the ratio of crack skeleton pixels to total image pixels, and is a dynamic adjustment factor. , where t is the adjustment coefficient of the dynamic adjustment factor, which is selected according to the order of magnitude of each parameter of the actual PID, and is used to adjust the order of magnitude of the dynamic adjustment factor k; The obtained k is substituted into the calculations of the inner and outer loops of the PID controller for adjustment. The basic single-loop PID algorithm is as follows: ； in: u(m): The calculation result of the PID algorithm, i.e., the control quantity; Kp: The adjustment coefficient of the proportional term, which adjusts the performance of the PID proportional link; Ki: The adjustment coefficient of the integral term, which adjusts the performance of the PID integral link; Kd: The adjustment coefficient of the derivative term, which adjusts the performance of the PID derivative link; e(m) is the error, i.e., the target value minus the current state value of the controlled object; [e(m)-e(m-1)] is the current error minus the previous error.

3. The method for detecting road surface cracks according to claim 1, characterized in that, The obtained k is substituted into the calculations of the inner and outer loops of the PID controller for adjustment, where: The outer loop adjustment process is as follows: 1) Generate the baseline lateral displacement based on the path planning, and multiply it by k to dynamically adjust the target: Where, target1 is the target value of the outer loop PID, and target is the baseline value of the target value of the outer loop PID; 2) Add dynamic limiting to the outer loop PID output to adjust the target speed of the speed loop PID: ； in, Vmax is the output value of the outer loop PID, which serves as the target value for the inner loop PID; Vmax is the target speed limit value for the inner loop speed PID, used to limit the range of the target speed of the inner loop PID. The inner loop adjustment process is as follows: The aggressiveness of the response is adjusted according to the degree of crack density. ； ； in: The coefficient of the proportional control term in the inner-loop PID controller. This is the reference value for the proportional control term coefficient of the inner-loop PID controller. The coefficient of the integral control term in the inner-loop PID controller. This is the reference value for the coefficient of the integral adjustment term of the inner loop PID.

4. The method for detecting road surface cracks according to claim 1, characterized in that, In step S6, the cross-scanning of the ground penetrating radar (1) involves segmenting the crack by the crack inflection point, and performing a delayed cross-scanning based on the distance L between the inflection point and the distance L from the ground penetrating radar to the front camera; specifically: When a crack in the image of the camera (2) has an inflection point with a curvature greater than the threshold l, it is identified and recorded. When the vehicle body (4) is close to the inflection point, and the inflection point disappears from the image of the camera (2), the ground penetrating radar (1) fixes the memory of the orientation of the front of the vehicle body (4) at this time, and continues to cross-scan the crack along the vertical direction of the vehicle's travel direction. The front of the vehicle body (4) turns according to the inflection point direction and begins to record the travel distance. When the travel distance reaches the distance L between the camera (2) and the ground penetrating radar (1), the ground penetrating radar (1) continuously updates the memory of the orientation of the front of the vehicle body (4) at this time until the front of the vehicle reaches the inflection point again. When the next inflection point has disappeared from the image of the camera (2) before traveling to the distance L, it is determined that the inflection point area is too small, and the ground penetrating radar can directly scan and cover it, simplifying it into a straight crack. It immediately continues to update the orientation of the front of the vehicle body (4) at this time, and scans the next section of the crack along the vertical direction of the orientation of the front of the vehicle body (4).

Images