Vehicle-mounted and airborne cooperation-based intelligent inspection system and method for whole road area of road

Abstract

The invention relates to the technical field of highway intelligent inspection, in particular to a highway full-area intelligent inspection system and method based on vehicle-mounted and airborne cooperation, and aims to efficiently and comprehensively complete highway inspection. The invention discloses a highway full-area intelligent inspection system which comprises a sensing layer, an algorithm layer, a network layer and an application layer. Wherein the sensing layer comprises an inspection vehicle and at least one unmanned aerial vehicle, the inspection vehicle carries a first sensor group and a vehicle-mounted unmanned aerial vehicle field which are used for collecting road surface, road area environment and tunnel electromechanical facility data, and the unmanned aerial vehicle carries a second sensor group used for collecting bridge and side slope data. A complete intelligent closed loop is constructed through a sensing-network-algorithm-application four-layer architecture. Automatic collaborative scheduling based on an optimization algorithm, autonomous starting of a GNSS signal weak area SLAM technology, cloud multi-source data fusion and deep analysis of a special AI model are carried out, and finally result output is carried out through a visual platform.

Description

Technical Field

[0001] This invention relates to the field of intelligent highway inspection technology, and in particular to an intelligent highway inspection system and method based on vehicle-mounted and airborne collaboration. Background Technology

[0002] As a core infrastructure of the national comprehensive transportation system, the health of highways—including their pavement, roadside environment, bridge and tunnel structures, and electromechanical facilities—directly affects public travel safety and socio-economic efficiency. To ensure the long-term stable service of highways, regular inspections and precise maintenance have become core tasks for traffic management departments. Expressways, as the key component of the highway system with the highest technical standards and heaviest traffic, present far greater challenges and demands on the complexity and importance of their inspections compared to ordinary highways, requiring even more stringent standards for efficiency, accuracy, safety, and comprehensive coverage.

[0003] However, current mainstream highway inspection methods still have significant shortcomings and are difficult to meet the needs of modern and refined management. For example, the highway rapid inspection and maintenance system disclosed in patent CN115767005A uses an inspection vehicle equipped with onboard camera components and an onboard computing platform to replace manual inspection and recording. Although this alleviates the problem of heavy manual workload to some extent, it is limited by the fixed viewing angle of the vehicle and cannot reach dangerous or hidden areas such as the bottom of bridges, the upper part of high slopes, and the top of tunnels. It is completely unable to detect structural cracks, spalling, and other defects in such areas, resulting in a high rate of missed detection.

[0004] As a key component of the highway system with the highest technical standards and the greatest traffic intensity, the complexity and importance of highway inspection work far exceed that of ordinary roads. On the one hand, highways have high traffic volume and speed, and maintenance personnel face extremely high personal safety risks when entering the driving lanes for inspection. Traditional manual inspection and single vehicle-mounted inspection solutions cannot completely avoid this risk. On the other hand, highways have a dense network of bridges and tunnels and a wide road area. Multiple types of facilities, such as road surfaces, bridges, slopes, and tunnel electromechanical systems, need to be inspected simultaneously. Existing single-equipment or single-platform inspection methods are no longer sufficient to meet the needs of refined and intelligent management. Summary of the Invention

[0005] The purpose of this invention is to provide a highway full-area intelligent inspection system and method based on vehicle-mounted and airborne collaboration, which aims to efficiently and comprehensively complete the inspection of highways.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides a highway full-area intelligent inspection system based on vehicle-mounted and airborne collaboration, comprising: a perception layer, including an inspection vehicle and at least one drone; the inspection vehicle is equipped with a first sensor group for collecting data on road surface, road environment, and tunnel electromechanical facilities, and a vehicle-mounted drone airport; the drone is equipped with a second sensor group for collecting data on bridges and slopes; the first sensor group includes a vehicle-mounted forward-looking camera and a vehicle-mounted side camera deployed on the inspection vehicle; the vehicle-mounted drone airport includes a vehicle-mounted rear-opening retractable drone airport deployed in the cargo box of the inspection vehicle and a commercially integrated airport deployed on the top of the inspection vehicle; the second sensor group includes a lidar, a zoom camera, and a vision sensor deployed on the drone; and an algorithm layer. The system is used to process and analyze the data collected by the perception layer, including a multi-device collaborative scheduling module, a multi-source data fusion positioning module, and a multi-target disease identification module. The multi-device collaborative scheduling module is configured to generate a collaborative inspection plan between inspection vehicles and drones based on planned inspection tasks. The multi-source data fusion positioning module is configured to assign a unified spatiotemporal reference to the acquired data. The multi-target disease identification module is configured to perform disease analysis and quantification on the data with the unified spatiotemporal reference. The network layer is used to realize data transmission between the perception layer and the application layer, and data transmission between the algorithm layer and the application layer. The application layer is used to visualize, analyze, and generate reports on the disease data processed by the algorithm layer.

[0007] The vehicle-mounted, rear-opening, retractable unmanned aerial vehicle (UAV) airport includes a sliding platform base assembly, a primary sliding platform, a secondary sliding platform, and a centering platform. The sliding platform base assembly is installed at the bottom of the vehicle's cargo box. Parallel base rails are installed on the inner walls of both sides of the sliding platform base assembly. The primary sliding platform includes a primary rail, a first connecting plate, and a second connecting plate. There are two primary rails, each connected to the base rail via a first pulley group to form a first sliding pair. The first and second connecting plates are both located between the two primary rails. The secondary sliding platform includes a secondary rail, a third connecting plate, and a fourth connecting plate. There are two secondary rails, each connected to the primary rail on the primary sliding platform via corresponding second pulley groups to form a second sliding pair. The third and fourth connecting plates are both located between the two secondary rails. A centering platform is fixed above the secondary sliding platform to support and position the landing drone; a drive motor is fixed on the inner wall of the sliding platform base assembly; the sliding platform base assembly, the primary sliding platform and the secondary sliding platform are linked by a synchronous belt transmission mechanism and driven by the drive motor to realize the telescopic movement of the secondary sliding platform relative to the primary sliding platform, as well as the synchronous telescopic movement of the primary sliding platform relative to the sliding platform base assembly.

[0008] The synchronous belt drive mechanism includes: a first drive wheel, a first driven wheel, and a base synchronous belt mounted on the sliding platform base assembly; the first drive wheel is connected to the output shaft of a drive motor, and the first drive wheel and the first driven wheel are connected by a base synchronous belt drive; a rack is provided on the bottom inner side of the base guide rail; a base synchronous belt pressure plate fixedly connected to the base synchronous belt is provided on the upper side of the first connecting plate, and a first drive shaft is fixedly connected to the lower side of the first connecting plate, with both ends of the first drive shaft meshing with the rack via gears; a second drive wheel is provided on the first drive shaft, and a second driven wheel is provided on the bottom of the second connecting plate, with the second drive wheel and the second driven wheel connected by a first-stage synchronous belt drive; a first-stage sliding platform synchronous belt pressure plate fixedly connected to the first-stage synchronous belt is provided on the lower side of the third connecting plate.

[0009] The centering platform includes a platform plate, a first centering lever pair, and a second centering lever pair. Four linear guide rails and four support frames are fixed to the bottom of the platform plate in a U-shape. The four support frames are located near the ends of adjacent linear guide rails. A worm geared motor is also fixed to the bottom of the platform plate near the support frames. Starting from the first support frame near the worm geared motor, the remaining three support frames, distributed clockwise, are the second, third, and fourth support frames. The first support frame includes a first synchronous pulley, a second synchronous pulley, and a third synchronous pulley, which are fixed longitudinally on the second drive shaft. The second support frame includes a fixed... A fourth and fifth synchronous pulley are longitudinally distributed sequentially on the third drive shaft; a third support frame includes a sixth and seventh synchronous pulley, which are fixed to the fourth drive shaft and are longitudinally distributed sequentially; a fourth support frame includes an eighth and ninth synchronous pulley, which are fixed to the fifth drive shaft and are longitudinally distributed sequentially; the first synchronous pulley is connected to the output shaft of the worm geared motor; the second and fourth synchronous pulleys are connected by a synchronous belt drive on the first return platform; the fifth and seventh synchronous pulleys are connected by a synchronous belt drive on the second return platform; the sixth and eighth synchronous pulleys are connected by a synchronous belt drive on the third return platform; and the ninth synchronous pulley is connected to the third synchronous pulley. The first and second centering lever pairs are connected by a synchronous belt drive via the fourth centering platform. Each pair includes two parallel centering levers, with the first pair perpendicular to the second pair. The first pair includes a first centering lever and a second centering lever, while the second pair includes a third centering lever and a fourth centering lever. One end of the first centering lever is fixed to the outer side of the second centering platform synchronous belt via the first base, and the other end is fixed to the inner side of the fourth centering platform synchronous belt via the second base. One end of the second centering lever is fixed to the inner side of the second centering platform synchronous belt via the third base. The other end is fixed to the outside of the fourth return platform synchronous belt via the fourth base; one end of the third return lever is fixed to the inside of the first return platform synchronous belt via the fifth base, and the other end of the third return lever is fixed to the outside of the third return platform synchronous belt via the sixth base; one end of the fourth return lever is fixed to the outside of the first return platform synchronous belt via the seventh base, and the other end of the fourth return lever is fixed to the inside of the third return platform synchronous belt via the eighth base; the first base, second base, third base, fourth base, fifth base, sixth base, seventh base and eighth base are respectively slidably connected to the corresponding linear guide rails via sliders.

[0010] A buffer block is provided at the end of the secondary sliding platform, and a contact switch is provided on the inner wall of the sliding platform base assembly. The contact switch is connected to the drive motor. When the secondary sliding platform is retracted to the limit position, the buffer block triggers the contact switch, and the drive motor stops working. A limit switch base is also fixed at the bottom of the platform plate. A movable limit switch is provided on the limit switch base, and the limit switch is connected to the worm gear motor. When any return lever moves to the limit switch, the worm gear motor stops working.

[0011] This invention also provides a method for intelligent inspection of the entire highway network, based on the aforementioned intelligent inspection system for the entire highway network. The method includes: S1. Based on the planned inspection task, a collaborative inspection plan for inspection vehicles and drones is generated through a multi-device collaborative scheduling module; S2. According to the collaborative inspection plan, the first type of data collected by the inspection vehicle while traveling along the highway, and the second type of data collected by the drone upon takeoff are acquired; when the drone enters a weak signal area of ​​the Global Navigation Satellite System, a simultaneous positioning and mapping algorithm is used for positioning and obstacle avoidance; S3. A unified spatiotemporal reference is assigned to the acquired data through a multi-source data fusion positioning module, and a multi-target defect identification module is used for defect analysis and quantification; S4. The analyzed and quantified defect data is displayed through a two-dimensional or three-dimensional visualization platform, and an inspection report is generated.

[0012] In step S2, after the UAV completes the collection of the second type of data, the UAV is controlled to fly to the preset landing airspace and the onboard vision sensor is activated; the visual guidance target set on the vehicle-mounted UAV field is identified and the relative position and attitude of the UAV and the target are calculated; based on the relative position and attitude, the flight state of the UAV is dynamically adjusted and the UAV is controlled to land on the return platform.

[0013] If the visually guided target is not successfully identified, the drone is controlled to hover at its current position and report the identification anomaly information to the cloud or vehicle control center, waiting for remote instructions or executing the preset go-around procedure.

[0014] In step S1, the multi-device collaborative scheduling module uses a population-based optimization algorithm to dynamically plan the inspection paths and take-off and landing scheduling of multiple UAVs. The constraints include at least one of the following: task priority, UAV range, and geographical location of the take-off and landing point.

[0015] The multi-target defect identification module in step S3 includes: a first deep learning model for quickly detecting road surface cracks, potholes and water seepage. The first deep learning model introduces a dynamic convolution mechanism in the feature fusion stage to improve the robustness of multi-scale defect detection; and a second deep learning model for identifying cracks, spalling and corrosion of bridge and slope structures. The second deep learning model adopts an enhancement strategy for small-target defects during training to improve feature diversity.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention fundamentally overcomes the physical limitations of a single platform by enabling collaborative operations between inspection vehicles and drones. The inspection vehicle is responsible for collecting data on the "ground layer," including road surface, road environment, and tunnel electromechanical facilities, while the drone platform is responsible for collecting data on "aerial layer" structures such as bridges and slopes. This integrated air-ground collaborative inspection mode achieves full road coverage from the road surface to the air, eliminating blind spots inherent in manual inspections and single vehicle-mounted systems. Furthermore, the parallel operation of the dual platforms significantly improves the efficiency and scope of data collection compared to traditional sequential manual inspections or single-equipment inspections.

[0017] 2. This invention constructs a complete intelligent closed loop through a four-layer architecture of perception, network, algorithm, and application. From automated collaborative scheduling based on optimization algorithms to the autonomous activation of SLAM technology in weak GNSS signal areas, to the deep analysis of multi-source data fusion and dedicated AI models in the cloud, and finally to the output of results through a two- or three-dimensional visualization platform, it achieves full automation and intelligence from task planning to result generation. This greatly reduces manual intervention, solves the problem of data processing lag in traditional methods, and provides timely and reliable data support for precise maintenance and scientific decision-making of highways. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of a highway full-area intelligent inspection system based on vehicle-mounted and airborne collaboration provided in an embodiment of this application; Figure 2 This is a schematic diagram of the hardware layout of an in-vehicle device provided in an embodiment of this application; Figure 3 This is a schematic diagram of the deployment of a vehicle-mounted, rear-opening, retractable unmanned aerial vehicle (UAV) provided in an embodiment of this application. Figure 4 This is a partial structural schematic diagram of a vehicle-mounted, rear-opening, telescopic unmanned aerial vehicle (UAV) airport return platform provided in an embodiment of this application; Figure 5 This is a schematic diagram of a partial structure of a vehicle-mounted rear-opening telescopic unmanned airport provided in an embodiment of this application; Figure 6 This is a schematic diagram of a vehicle-mounted, rear-opening, retractable unmanned aerial vehicle (UAV) airport recovery system provided in an embodiment of this application. Figure 7 This is a schematic diagram of a drone return provided in an embodiment of this application; Figure 8 This is a distribution of synchronous pulleys in a first support frame provided in an embodiment of this application.

[0019] Reference numerals: 1 is the sliding platform base assembly; 10 is the vehicle-mounted forward-looking camera; 101 is the base guide rail; 102 is the first connecting plate; 103 is the second connecting plate; 104 is the third connecting plate; 105 is the fourth connecting plate; 106 is the first drive wheel; 107 is the first driven wheel; 108 is the base timing belt; 109 is the rack; 110 is the base timing belt pressure plate; 111 is the first transmission shaft; 112 is the gear; 113 is the second drive wheel; 114 is the second driven wheel; 115 is... 1. Primary synchronous belt; 116 is the primary sliding platform synchronous belt pressure plate; 117 is the contact switch; 2 is the primary sliding platform; 20 is the vehicle-mounted side camera; 201 is the first pulley block; 202 is the primary guide rail; 3 is the secondary sliding platform; 30 is the edge computing industrial control computer; 301 is the secondary guide rail; 302 is the second pulley block; 303 is the buffer block; 4 is the centering platform; 40 is the vehicle-mounted central control unit; 401 is the platform plate; 4021 is the first centering lever; 4022 is the second centering lever; 40 31 is the third centering lever; 4032 is the fourth centering lever; 404 is a linear guide rail; 405 is a worm geared motor; 406 is the first support frame; 407 is the second support frame; 408 is the third support frame; 409 is the fourth support frame; 410 is the limit switch base; 411 is the limit switch; 412 is the second drive shaft; 413 is the first synchronous pulley; 414 is the second synchronous pulley; 415 is the third synchronous pulley; 416 is the first centering platform synchronous belt; 417 is the second centering platform... 418 is the synchronous belt for the third intermediate platform; 419 is the synchronous belt for the fourth intermediate platform; 420 is the slider; 421 is the first base; 422 is the second base; 423 is the third base; 424 is the fourth base; 425 is the fifth base; 426 is the sixth base; 427 is the seventh base; 428 is the eighth base; 5 is the drive motor; 50 is the high-precision positioning and attitude determination system; 60 is the communication module; 70 is the commercial integrated airport; 80 is the vehicle-mounted rear-opening retractable unmanned airport. Detailed Implementation

[0020] 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0021] Reference Figure 1This application provides a highway full-area intelligent inspection system based on vehicle-mounted and airborne collaboration (hereinafter referred to as the intelligent inspection system), including a perception layer, an algorithm layer, a network layer, and an application layer. The perception layer includes an inspection vehicle and at least one drone. The inspection vehicle is equipped with a first sensor group for collecting data on the road surface, road environment, and tunnel electromechanical facilities, and an onboard drone. The drone is equipped with a second sensor group for collecting data on bridges and slopes.

[0022] The algorithm layer is used to process and analyze the data collected by the perception layer, including a multi-device collaborative scheduling module, a multi-source data fusion positioning module, and a multi-target disease identification module.

[0023] The multi-device collaborative scheduling module is configured for planned inspection tasks, generating collaborative inspection schemes between inspection vehicles and drones. As one possible implementation, the module employs a genetic algorithm or particle swarm optimization algorithm as its core optimization logic. Leveraging the optimal solution-solving capabilities of these algorithms under strong constraints, it obtains efficient collaborative schemes under multi-variable disturbances. For example, the constraints of the multi-device collaborative scheduling module include: inspection task priority (e.g., prioritizing the detection of high-risk defects such as bridge cracks and slope collapses); drone endurance, take-off and landing location, and weather and airspace conditions. The module receives real-time equipment status feedback from the perception layer and inspection tasks set by operators at the application layer, dynamically generating collaborative inspection schemes, such as inspection routes for vehicle-mounted equipment traveling along traffic flow, operational zones for multiple drones, and drone take-off and landing scheduling plans.

[0024] The multi-source data fusion positioning module is configured to assign a unified spatiotemporal reference to the acquired data, eliminating data bias. That is, it transforms the image data and corresponding positioning information of road surface, road environment, and tunnel electromechanical facilities collected by vehicle-mounted equipment, as well as the image data and corresponding positioning information of bridge and slope collected by UAVs, into standardized data with unified data format, positioning reference, and timestamp.

[0025] The multi-target defect identification module is configured to analyze and quantify defects based on data from a unified spatiotemporal benchmark. As one possible implementation, the module, based on fused standardized data, uses an improved deep learning model to achieve automated identification and quantification of defects across the entire highway network, replacing traditional manual visual inspection and improving identification efficiency and accuracy. For example, a classification modeling strategy is adopted to address the differences in defect characteristics among different inspection targets, resulting in two types of improved models: First, the improved YOLOv12 model targets defects in road surfaces, road environments, and tunnel electromechanical facilities. It introduces a cross-stage partial network structure (CSPNet) to enhance gradient flow and feature reuse capabilities, reducing computational redundancy while maintaining detection accuracy and meeting the real-time processing requirements of onboard edge computing. An embedded coordinate attention module is added before the feature pyramid prediction head, using spatially oriented pooling and non-linear weighting to improve the model's ability to identify defects sensitive to spatial structures such as crack direction and pothole contours. Dynamic convolutional layers are employed in the multi-scale feature fusion stage, where a lightweight attention sub-network adaptively generates convolutional kernel parameters, replacing fixed weights and improving adaptability to complex environments such as rainy roads and dimly lit tunnel environments.

[0026] Second, the improved DEIM model is designed for bridge and slope structural defects. The front-end uses a high-resolution network to preserve the geometric details of the defects through multi-resolution parallel branches and cross-scale fusion. A physically-based dynamic embedding module is introduced, combining structural deformation constraints, such as the mechanical properties of bridge load-bearing structures, with attention modulation to achieve structurally sensitive modeling of crack propagation trends and spalling extents. Training employs a multi-scale Mosaic enhancement and Random Pasting composite strategy to improve the sample density and feature diversity of small-target defects, such as minute cracks at the bottom of bridges, balancing lightweight structures with detection accuracy.

[0027] The network layer is used to implement data transmission between the perception layer and the application layer, and between the algorithm layer and the application layer. Examples include 4G / 5G communication networks, edge preprocessing for breakpoint resumption, and data security transmission protocols.

[0028] The application layer is used to visualize, analyze, and generate reports from the disease data processed by the algorithm layer.

[0029] For example, refer to Figure 2 The first sensor group includes a vehicle-mounted forward-looking camera 10 and a vehicle-mounted side camera 20 deployed on the inspection vehicle. The vehicle-mounted unmanned aerial vehicle (UAV) airport includes a vehicle-mounted rear-opening retractable UAV airport 80 deployed in the cargo box of the inspection vehicle and a commercial integrated airport 70 deployed on the top of the inspection vehicle. The commercial integrated airport 70 and the vehicle-mounted rear-opening retractable UAV airport 80 support the deployment, recovery, and collaborative operation of multiple UAVs. The second sensor group includes a lidar, zoom camera, and vision sensor deployed on the UAVs. Figure 2The inspection vehicle is also equipped with a high-precision positioning and attitude determination system 50. This system provides accurate position and attitude references for the onboard sensor data and is crucial for achieving multi-source data fusion positioning. The edge computing industrial control computer 30 is the physical manifestation of the algorithm layer on the vehicle. The onboard central control unit 40 is the physical manifestation of the application layer on the vehicle. Operators can plan tasks, monitor inspection status, and view preliminary analysis results through the human-machine interface of the onboard central control unit 40, which serves as the entry point for interaction between the intelligent inspection system and the user. The communication module 60, as the physical manifestation of the network layer on the vehicle, is responsible for transmitting the pre-processed edge data back to the cloud in real time and receiving instructions from the cloud.

[0030] The inspection vehicle, serving as the core carrier for ground-based data collection, carries a first sensor group to collect data on the road surface, road environment, and tunnel electromechanical facilities. This data includes image data of these scenarios, such as visual and location information related to road surface cracks, road obstacles, and the status of tunnel lighting. The drone, acting as the aerial data collection carrier, primarily supplements the coverage of high-altitude or complex structures such as bridges and slopes that the inspection vehicle cannot reach. Its second sensor group is specifically designed to collect data on bridges and slopes, including image data of structural defects such as cracks at the bottom of bridges and slope collapses, as well as location information corresponding to the collection location. This fills the data gaps in the high-altitude blind spots of the inspection vehicle, achieving full road area data coverage.

[0031] As one possible approach, to ensure the stability of drone operations, two types of auxiliary equipment are also included: Unmanned aerial vehicle (UAV) take-off and landing (UAV) auxiliary equipment is used to enable UAVs to land precisely at locations other than their original take-off points. For example, after a UAV takes off from point A to perform an inspection mission, it can be precisely landed at a vehicle-mounted UAV airport at point B using this equipment, avoiding the loss of inspection efficiency caused by the UAV returning to its original take-off point.

[0032] The synchronous positioning and mapping module automatically activates when the UAV enters a GNSS signal weak area (such as above a tunnel, on a mountain slope, or other obstructed area). Through synchronous positioning and mapping technology, it enables the UAV to achieve accurate positioning and autonomous obstacle avoidance, ensuring the accuracy of data collection location and the safety of UAV operations.

[0033] For example, refer to Figures 3-8The vehicle-mounted, rear-opening, retractable unmanned aerial vehicle (UAV) airport includes a sliding platform base assembly 1, a primary sliding platform 2, a secondary sliding platform 3, and a centering platform 4. The sliding platform base assembly 1 is bolted to the bottom of the vehicle's cargo box. Parallel base guide rails 101 are bolted to the inner walls of both sides of the sliding platform base assembly 1. The primary sliding platform 2 includes a primary guide rail 202, a first connecting plate 102, and a second connecting plate 103. There are two primary guide rails 202, each connected to the base guide rail 101 on the sliding platform base assembly 1 via a first pulley group 201, forming a first sliding pair. The first connecting plate 102 and the second connecting plate 103 are both located between the two primary guide rails 202. The secondary sliding platform 3 includes a secondary guide rail 301, a third connecting plate 104, and a fourth connecting plate 105. There are two secondary guide rails 301, which are connected to the primary guide rail 202 set on the primary sliding platform 2 via the second pulley group 302 to form a second sliding pair. The third connecting plate 104 and the fourth connecting plate 105 are both located between the two secondary guide rails 301. A centering platform 4 is fixed above the secondary sliding platform 3 to support and position the landing UAV; a drive motor 5 is fixed on the inner wall of the sliding platform base assembly 1.

[0034] The first connecting plate 102 and the second connecting plate 103 are connected to two primary guide rails 202, forming an integral structure that constitutes the main frame of the primary sliding platform 2. The third connecting plate 104 and the fourth connecting plate 105 are connected to two secondary guide rails 301, forming an integral structure that constitutes the main frame of the secondary sliding platform 3. The sliding platform base assembly 1, the primary sliding platform 2, and the secondary sliding platform 3 are linked by a synchronous belt transmission mechanism and driven by a drive motor 5 to realize the telescopic movement of the secondary sliding platform 3 relative to the primary sliding platform 2, and the synchronous telescopic movement of the primary sliding platform 2 relative to the sliding platform base assembly 1.

[0035] As one possible implementation, the synchronous belt drive mechanism includes: a first drive wheel 106, a first driven wheel 107, and a base synchronous belt 108 mounted on the sliding platform base assembly 1; the first drive wheel 106 is connected to the output shaft of the drive motor 5, and the first drive wheel 106 and the first driven wheel 107 are connected by the base synchronous belt 108; a rack 109 is bolted to the bottom inner side of the base guide rail 101; and a base synchronous belt pressure plate fixedly connected to the base synchronous belt 108 is provided on the upper side of the first connecting plate 102. Plate 110, a first drive shaft 111 is fixedly connected to the lower side of the first connecting plate 102, and the two ends of the first drive shaft 111 mesh with the rack 109 through gears 112; a second drive wheel 113 is provided on the first drive shaft 111, a second driven wheel 114 is provided at the bottom of the second connecting plate 103, and the second drive wheel 113 and the second driven wheel 114 are connected by a primary synchronous belt 115; a primary sliding platform synchronous belt pressure plate 116 is provided at the bottom of the third connecting plate 104 and is fixedly connected to the primary synchronous belt 115.

[0036] As one possible implementation, a buffer block 303 is provided at the end of the secondary sliding platform 3, and a contact switch 117 is provided on the inner wall of the sliding platform base assembly 1. The contact switch 117 is connected to the drive motor 5. When the secondary sliding platform 3 is retracted to the limit position, the buffer block 303 triggers the contact switch 117, and the drive motor 5 stops working.

[0037] In the initial stage, the sliding platform base assembly 1 is fixed to the bottom of the inspection vehicle's cargo box. The primary sliding platform 2, secondary sliding platform 3, and centering platform 4 are all retracted inside the cargo box to avoid occupying external space or being affected by external interference when the vehicle is in motion. When the drone needs to take off or land, the drive motor 5 is started. After the drive motor 5 starts, it drives the first drive wheel 106 connected to its output shaft to rotate. The first drive wheel 106 drives the first driven wheel 107 through the base synchronous belt 108, forming the first transmission circuit. The first connecting plate 102 of the primary sliding platform 2 is provided with a base synchronous belt pressure plate 110. The base synchronous belt pressure plate 110 clamps and fixes a section of the base synchronous belt 108. Therefore, when the base synchronous belt 108 rotates cyclically, it drags the entire primary sliding platform 2 outward along the base guide rail 101 through the base synchronous belt pressure plate 110. When the primary sliding platform 2 is dragged, the first transmission shaft 111 fixed at its bottom moves accordingly, and the gears 112 at both ends of the first transmission shaft 111 mesh with the rack 109. Since the second drive wheel 113 is mounted on the first drive shaft 111, the rotation of the first drive shaft 111 directly drives the rotation of the second drive wheel 113. The rotation of the second drive wheel 113 is transmitted to the second driven wheel 114 mounted on the second connecting plate 103 via the first-stage synchronous belt 115. The first-stage sliding platform synchronous belt pressure plate 116 at the bottom of the third connecting plate 104 on the second-stage sliding platform 3 clamps and fixes a section of the first-stage synchronous belt 115. Therefore, when the first-stage synchronous belt 115 moves, it directly drags the second-stage sliding platform 3 to extend along the first-stage guide rail 202.

[0038] The recovery process is the reverse of the above. When the platform is fully retracted, the buffer block 306 at the end of the secondary sliding platform 3 will trigger the contact switch 117 on the inner wall of the sliding platform base assembly 1, and the drive motor will stop running. This design not only serves as a physical limit, but also provides a "zeroing" signal for the intelligent inspection system, which can reset the system coordinates and eliminate the cumulative errors that may occur during long-term operation.

[0039] It should be understood that Figure 4 To more clearly demonstrate the structure of the return platform, one of the four linear guide rails has been omitted.

[0040] As one possible implementation, the centering platform includes a platform plate 401, a first centering lever pair, and a second centering lever pair. Four linear guide rails 404 arranged in a U-shape and four support frames are fixed to the bottom of the platform plate 401. The four support frames are located near the ends of adjacent linear guide rails 404. A worm geared motor 405 is also fixed to the bottom of the platform plate 401 near the support frames.

[0041] Starting from the first support frame 406 near the worm geared motor 405, the remaining three support frames distributed clockwise are the second support frame 407, the third support frame 408, and the fourth support frame 409. The first support frame 406 includes a first synchronous pulley 413, a second synchronous pulley 414, and a third synchronous pulley 415, which are longitudinally distributed in sequence on the second drive shaft 412. The second support frame 407 includes a fourth synchronous pulley and a fifth synchronous pulley, which are longitudinally distributed in sequence on the third drive shaft. The third support frame 408 includes a sixth synchronous pulley and a seventh synchronous pulley, which are longitudinally distributed in sequence on the fourth drive shaft. The fourth support frame 409... The frame 409 includes an eighth and a ninth synchronous pulley, which are longitudinally distributed and fixed on the fifth drive shaft. The first synchronous pulley is connected to the output shaft of the worm gear motor 405. The second and fourth synchronous pulleys are connected by a first return platform synchronous belt 416. The fifth and seventh synchronous pulleys are connected by a second return platform synchronous belt 417. The sixth and eighth synchronous pulleys are connected by a third return platform synchronous belt 418. The ninth synchronous pulley is connected to the third synchronous pulley by a fourth return platform synchronous belt 419. Each base is also connected to a corresponding linear guide rail via a slider 420.

[0042] For example, the worm geared motor 405 is also fixed on the support frame, which is bolted to the bottom of the platform plate 401. Figure 8 The distribution of the synchronous pulleys in the first support frame 406 is shown; the distribution of the synchronous pulleys in the remaining support frames is the same. According to... Figure 8 It can be seen that the vertical sequential distribution refers to the arrangement from top to bottom.

[0043] Both the first and second centering lever pairs include two parallel centering levers, with the first pair perpendicular to the second pair. The first pair includes a first centering lever 4021 and a second centering lever 4022, while the second pair includes a third centering lever 4031 and a fourth centering lever 4032. The first centering lever 4021 is connected at both ends to a first base 421 and a second base 422, respectively. The first base 421 and the second base 422 are fixed to the outer side of the timing belt 417 of the second centering platform and the inner side of the timing belt 419 of the fourth centering platform, respectively, by pressure plates. The second centering lever is connected at both ends to a third base 423 and a fourth base 424, respectively. The third base 423 and the fourth base 424 are respectively fixed to the inner side of the second return platform synchronous belt 417 and the outer side of the fourth return platform synchronous belt 419 by pressure plates; the two ends of the third return lever are respectively connected to the fifth base 425 and the sixth base 426, which are respectively fixed to the inner side of the first return platform synchronous belt 416 and the outer side of the third return platform synchronous belt 418 by pressure plates; the two ends of the fourth return lever are respectively connected to the seventh base 427 and the eighth base 428, which are respectively fixed to the outer side of the first return platform synchronous belt 416 and the inner side of the third return platform synchronous belt 418 by pressure plates.

[0044] As one possible implementation, a limit switch base 410 is also fixed at the bottom of the platform plate 401. A movable limit switch 411 is provided on the limit switch base 410. The limit switch 411 is connected to the worm gear motor 405. When any return lever moves to the limit switch 411, the worm gear motor 405 stops working.

[0045] The core function of the centering platform is to center the drone that has landed on platform 401. Using a single motor, two sets of vertically arranged centering levers move synchronously and precisely from both sides towards the center, thus pushing and stabilizing the drone in the center position of platform 401. The specific process is as follows: When the drone lands on platform 401, the intelligent inspection system receives a centering command. The turbine reduction electrode 405 activates, and its output shaft directly drives the first synchronous pulley 413. The first synchronous pulley 413 transmits power to the fourth synchronous pulley via the first centering platform synchronous belt 416. The fourth synchronous pulley transmits power to the seventh synchronous pulley via the second centering platform synchronous belt 417. The seventh synchronous pulley transmits power to the eighth synchronous pulley via the third centering platform synchronous belt 418. The eighth synchronous pulley transmits power to the third synchronous pulley 415 via the fourth centering platform synchronous belt 419. In addition to the three synchronous pulleys on the first support frame 406, the second support frame 407, third support frame 408, and fourth support frame 409 each include two synchronous pulleys. Closed-loop synchronous belts connect two synchronous pulleys at the same height on different support frames. All four centering platform synchronous belts are closed-loop structures, with the inner and outer sides moving in opposite directions. For example, when the centering platform synchronous belt rotates clockwise, the inner side moves to the left, and the outer side moves to the right. The centering lever utilizes this characteristic to achieve synchronized and precise movement from both sides towards the center.

[0046] Taking the first centering lever pair as an example, the two end bases of the first centering lever 4021 are fixed to the outer side of the second centering platform synchronous belt 417 and the inner side of the fourth centering platform synchronous belt 419 respectively by pressure plates. The two end bases of the second centering lever 4022 are fixed to the inner side of the second centering platform synchronous belt 417 and the outer side of the fourth centering platform synchronous belt 419 respectively by pressure plates. When the synchronous belts rotate, the outer side of the second centering platform synchronous belt 417 and the inner side of the fourth centering platform synchronous belt 419 move towards the center of the platform plate 401, and the inner side of the second centering platform synchronous belt 417 and the outer side of the fourth centering platform synchronous belt 419 also move towards the center of the platform plate 401. Therefore, the first centering lever 4021 and the second centering lever 4022 move towards the center of the platform plate 401 synchronously from both sides.

[0047] All the bases at both ends of the centering levers are mounted on the corresponding linear guides 404 via sliders 420, restricting the bases to move only along the linear direction specified by the linear guides 404. This prevents the centering levers from deviating during movement. Furthermore, the sliders 420 reduce the friction between the bases and the linear guides 404, ensuring the centering levers move smoothly and stably under the drive of the synchronous belt. When any centering lever touches the limit switch 411, the limit switch 411 sends a signal to the intelligent inspection system. The intelligent inspection system then controls the worm geared motor 405 to stop working, the centering levers stop moving, and the UAV is precisely positioned at the center of the platform plate. When the worm geared motor 405 rotates forward, the centering levers move from both sides towards the center; when the worm geared motor 405 rotates in reverse, the centering levers move from the center towards both sides.

[0048] One possible implementation is to use two limit switches, one to control the movement of the centering lever from both sides towards the center, and the other to stop the movement of the centering lever from the center outwards. By adjusting the position of the movable limit switches, the distance the centering lever moves can be changed, thus adapting to different sizes of drones.

[0049] This application also provides a method for intelligent inspection of the entire highway network, based on the aforementioned intelligent inspection system for the entire highway network. The method includes: S1. Based on the planned inspection tasks, a collaborative inspection plan for inspection vehicles and drones is generated through a multi-device collaborative scheduling module.

[0050] For example, in step S1, the multi-device collaborative scheduling module uses a population-based optimization algorithm to dynamically plan the inspection paths and take-off and landing scheduling of multiple UAVs. The constraints include at least one of the following: task priority, UAV range, and geographical location of the take-off and landing point.

[0051] S2. According to the collaborative inspection plan, acquire the first type of data collected by the inspection vehicle traveling along the highway, and the second type of data collected by the UAV taking off; when the UAV enters the weak signal area of ​​the global navigation satellite system, activate the synchronous positioning and map building algorithm for positioning and obstacle avoidance.

[0052] In step S2, after the UAV completes the collection of the second type of data, the UAV is controlled to fly to the preset landing airspace and the onboard vision sensor is activated; the visual guidance target set on the vehicle-mounted UAV field is identified and the relative position and attitude of the UAV and the target are calculated; based on the relative position and attitude, the flight state of the UAV is dynamically adjusted and it is controlled to land on the return platform.

[0053] As one possible implementation, visually guided targets are placed on the surface of the take-off and landing platform of a vehicle-mounted unmanned aerial vehicle (UAV) airport. The targets contain a unique QR code and a high-contrast concentric circle pattern.

[0054] If the visually guided target is not successfully identified, the drone is controlled to hover at its current position and report the identification anomaly information to the cloud or vehicle control center, waiting for remote instructions or executing the preset go-around procedure.

[0055] S3. The multi-source data fusion positioning module assigns a unified spatiotemporal reference to the acquired data, and the multi-target disease identification module is used for disease analysis and quantification.

[0056] For example, the multi-target defect identification module in step S3 includes: a first deep learning model (improved YOLOv12 model) for quickly detecting pavement cracks, potholes and water seepage. The first deep learning model introduces a dynamic convolution mechanism in the feature fusion stage to improve the robustness of multi-scale defect detection; and a second deep learning model for identifying cracks, spalling and corrosion of bridge and slope structures. The second deep learning model (improved DEIM model) adopts an enhancement strategy for small-target defects during training to improve feature diversity.

[0057] S4. Display the analyzed and quantified disease data through a two-dimensional or three-dimensional visualization platform and generate an inspection report.

[0058] In the description of this specification, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0059] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A highway full-area intelligent inspection system based on vehicle-mounted and airborne collaboration, characterized in that, include: The perception layer includes an inspection vehicle and at least one drone. The inspection vehicle is equipped with a first sensor group for collecting data on road surface, road environment and tunnel electromechanical facilities and a vehicle-mounted drone airport. The drone is equipped with a second sensor group for collecting data on bridges and slopes. The first sensor group includes a vehicle-mounted forward-looking camera (10) and a vehicle-mounted side camera (20) deployed on the inspection vehicle. The vehicle-mounted drone airport includes a vehicle-mounted rear-opening telescopic drone airport (80) deployed in the cargo box of the inspection vehicle and a commercial integrated airport (70) deployed on the top of the inspection vehicle. The second sensor group includes a lidar, a zoom camera and a vision sensor deployed on the drone. The algorithm layer is used to process and analyze the data collected by the perception layer, including a multi-device collaborative scheduling module, a multi-source data fusion positioning module, and a multi-target disease identification module; The multi-device collaborative scheduling module is configured to generate a collaborative inspection plan between inspection vehicles and drones based on planned inspection tasks; the multi-source data fusion positioning module is configured to assign a unified spatiotemporal reference to the acquired data; and the multi-target disease identification module is configured to perform disease analysis and quantification on the data with the unified spatiotemporal reference. The network layer is used to implement data transmission between the perception layer and the application layer, and data transmission between the algorithm layer and the application layer. The application layer is used to visualize, analyze, and generate reports on the disease data processed by the algorithm layer.

2. The intelligent highway inspection system based on vehicle-mounted and airborne collaboration as described in claim 1, characterized in that: The vehicle-mounted rear-opening telescopic unmanned airport (80) includes a sliding platform base assembly (1), a primary sliding platform (2), a secondary sliding platform (3), and a centering platform (4); the sliding platform base assembly (1) is installed at the bottom of the vehicle cargo box, and parallel base guide rails (101) are installed on the inner walls of both sides of the sliding platform base assembly (1); the primary sliding platform (2) includes a primary guide rail (202), a first connecting plate (102), and a second connecting plate (103); there are two primary guide rails (202), which are connected to the base guide rails (101) through corresponding first pulley groups (201) to form a first sliding pair; the first connecting plate (102) and the second connecting plate (103) are both located between the two primary guide rails (202); the secondary sliding platform (3) includes a secondary guide rail (301), a third connecting plate (104), and a fourth connecting plate (105 ...01) includes a There are two guide rails (301), which are connected to the first guide rail (202) on the first sliding platform (2) through the corresponding second pulley group (302) to form a second sliding pair. The third connecting plate (104) and the fourth connecting plate (105) are both located between the two second guide rails (301). A centering platform (4) is fixed above the second sliding platform (3) to support and position the landing UAV. A drive motor (5) is fixed on the inner wall of the sliding platform base assembly (1). The sliding platform base assembly (1), the first sliding platform (2) and the second sliding platform (3) are linked by a synchronous belt transmission mechanism and driven by the drive motor (5) to realize the telescopic movement of the second sliding platform (3) relative to the first sliding platform (2) and the synchronous telescopic movement of the first sliding platform (2) relative to the sliding platform base assembly (1).

3. The intelligent highway inspection system based on vehicle-mounted and airborne collaboration according to claim 2, characterized in that, The synchronous belt drive mechanism includes: a first drive wheel (106), a first driven wheel (107), and a base synchronous belt (108) mounted on the sliding platform base assembly (1); the first drive wheel (106) is connected to the output shaft of the drive motor (5), and the first drive wheel (106) and the first driven wheel (107) are connected by the base synchronous belt (108); a rack (109) is provided on the bottom inner side of the base guide rail (101); a base synchronous belt pressure plate (110) is provided on the upper side of the first connecting plate (102) and is fixedly connected to the base synchronous belt (108). A first drive shaft (111) is fixedly connected to the lower side of the plate (102). The two ends of the first drive shaft (111) mesh with the rack (109) through gears (112). A second drive wheel (113) is provided on the first drive shaft (111). A second driven wheel (114) is provided at the bottom of the second connecting plate (103). The second drive wheel (113) and the second driven wheel (114) are connected by a first-stage synchronous belt (115). A first-stage sliding platform synchronous belt pressure plate (116) is fixedly connected to the first-stage synchronous belt (115) on the lower side of the third connecting plate (104).

4. The intelligent highway inspection system based on vehicle-mounted and airborne collaboration according to claim 3, characterized in that: The centering platform (4) includes a platform plate (401), a first centering lever pair, and a second centering lever pair. The bottom of the platform plate (401) is fixed with four linear guide rails (404) arranged in a U-shape and four support frames. The four support frames are located near the ends of adjacent linear guide rails (404). A worm geared motor (405) is also fixed near the support frames at the bottom of the platform plate (401). Starting from the first support frame (406) near the worm geared motor (405), the remaining three support frames, arranged clockwise, are the second support frame (407), the third support frame (408), and the fourth support frame (409). The first support frame (406) includes a support fixed to the second drive shaft (412). The first synchronous pulley (413), the second synchronous pulley (414), and the third synchronous pulley (415) are arranged longitudinally in sequence; the second support frame (407) includes the fourth and fifth synchronous pulleys arranged longitudinally in sequence and fixed on the third drive shaft; the third support frame (408) includes the sixth and seventh synchronous pulleys arranged longitudinally in sequence and fixed on the fourth drive shaft; the fourth support frame (409) includes the eighth and ninth synchronous pulleys arranged longitudinally in sequence and fixed on the fifth drive shaft; the first synchronous pulley (413) is connected to the output shaft of the worm geared motor (405); the second synchronous pulley (414) and the fourth synchronous pulley are connected by a first intermediate platform synchronous belt (416); the fifth synchronous pulley and the seventh synchronous pulley are connected by a second intermediate platform synchronous belt (416). The central platform synchronous belt (417) is used for transmission connection. The sixth and eighth synchronous pulleys are connected by the third central platform synchronous belt (418), and the ninth synchronous pulley is connected by the third synchronous pulley (415) through the fourth central platform synchronous belt (419). The first and second central shift lever pairs each include two parallel central shift levers. The first central shift lever pair is perpendicular to the second central shift lever pair. The first central shift lever pair includes a first central shift lever (4021) and a second central shift lever (4022). The second central shift lever pair includes a third central shift lever (4031) and a fourth central shift lever (4032). One end of the first central shift lever (4021) is connected to the second central platform synchronous belt (418) through the first base (421). 7) The outer side is fixed, and the other end of the first return center lever (4021) is fixed to the inner side of the fourth return center platform synchronous belt (419) through the second base (422); one end of the second return center lever (4022) is fixed to the inner side of the second return center platform synchronous belt (417) through the third base (423), and the other end of the second return center lever (4022) is fixed to the outer side of the fourth return center platform synchronous belt (419) through the fourth base (424); one end of the third return center lever (4031) is fixed to the inner side of the first return center platform synchronous belt (416) through the fifth base (425), and the other end of the third return center lever (4031) is fixed to the outer side of the third return center platform synchronous belt (418) through the sixth base (426);One end of the fourth centering lever (4032) is fixed to the outside of the first centering platform synchronous belt (416) via the seventh base (427), and the other end of the fourth centering lever (4032) is fixed to the inside of the third centering platform synchronous belt (418) via the eighth base (428); the first base (421), second base (422), third base (423), fourth base (424), fifth base (425), sixth base (426), seventh base (427), and eighth base (428) are slidably connected to their respective linear guides via sliders (420).

5. The intelligent highway inspection system based on vehicle-mounted and airborne collaboration according to claim 4, characterized in that: The end of the secondary sliding platform (3) is provided with a buffer block (303), and the inner wall of the sliding platform base assembly (1) is provided with a contact switch (117), which is connected to the drive motor (5). When the secondary sliding platform (3) is retracted to the limit position, the buffer block (303) triggers the contact switch (117), and the drive motor (5) stops working. The bottom of the platform plate (401) is also fixed with a limit switch base (410), and a movable limit switch (411) is provided on the limit switch base (410), which is connected to the worm gear motor (405). When any return lever moves to the limit switch (411), the worm gear motor (405) stops working.

6. A method for intelligent inspection of the entire highway network, comprising a highway network intelligent inspection system based on vehicle-mounted and airborne collaboration as described in any one of claims 1-5, characterized in that, Includes the following steps: S1. Based on the planned inspection task, a collaborative inspection plan for inspection vehicles and drones is generated through the multi-device collaborative scheduling module; S2. According to the collaborative inspection scheme, acquire the first type of data collected by the inspection vehicle while traveling along the highway, and the second type of data collected by the UAV after takeoff; when the UAV enters the weak signal area of ​​the Global Navigation Satellite System, activate the synchronous positioning and map building algorithm for positioning and obstacle avoidance; S3. The multi-source data fusion positioning module assigns a unified spatiotemporal reference to the acquired data, and the multi-target disease identification module is used for disease analysis and quantification. S4. Display the analyzed and quantified disease data through a two-dimensional or three-dimensional visualization platform and generate an inspection report.

7. The intelligent inspection method for the entire highway area according to claim 6, characterized in that: In step S2, after the UAV completes the collection of the second type of data, the UAV is controlled to fly to the preset landing airspace and the onboard vision sensor is activated. The system identifies visual guidance targets set up on the vehicle-mounted drone field and calculates the relative position and attitude of the drone and the target; based on the relative position and attitude, it dynamically adjusts the drone's flight state and controls the drone to land on the return platform.

8. The intelligent inspection method for the entire highway area according to claim 7, characterized in that: If the visually guided target is not successfully identified, the drone is controlled to hover at its current position and report the identification anomaly information to the cloud or vehicle control center, waiting for remote instructions or executing a preset go-around procedure.

9. The intelligent inspection method for the entire highway area according to claim 6, characterized in that: In step S1, the multi-device collaborative scheduling module uses a population-based optimization algorithm to dynamically plan the inspection paths and take-off and landing scheduling of multiple UAVs. The constraints include at least one of the following: task priority, UAV range, and geographical location of the take-off and landing point.

10. A method for intelligent inspection of the entire highway area according to claim 6, characterized in that, The multi-target defect identification module in step S3 includes: a first deep learning model for quickly detecting road surface cracks, potholes and water seepage. The first deep learning model introduces a dynamic convolution mechanism in the feature fusion stage to improve the robustness of multi-scale defect detection; and a second deep learning model for identifying cracks, spalling and corrosion of bridge and slope structures. The second deep learning model adopts an enhancement strategy for small-target defects during training to improve feature diversity.

Images