Subgrade construction quality multi-index integrated in-situ detection system and method

Abstract

The present application relates to the field of civil engineering and traffic geotechnical engineering, solves the problems of easy breakage of probe, non-dense testing and point cloud distortion of bumpy road, provides a roadbed construction quality multi-index integrated in-situ detection system and method, which comprises a robot platform and a compaction quality detection module, a geometric appearance index detection module and a main control and data fusion center carried thereon; the main control and data fusion center is used for: driving the robot platform to travel on the set cruise route and stopping at the preset grid physical detection station, and then activating the lateral translation of the compaction quality detection module, the pre-pilot hole drilling and the probe extrusion penetration action in sequence; during the compaction degree and moisture content measurement, the current relative coordinates are grabbed as the measurement point position label, and are combined into a position-quality label package; the compaction quality and geometric appearance continuous detection report of the whole roadbed is independently output, and the multi-index integrated efficient detection and intelligent perception of roadbed construction quality are realized.

Description

Technical Field

[0001] This invention relates to the fields of civil engineering and transportation geotechnical engineering, and in particular to an integrated in-situ testing system and method for multiple indicators of roadbed construction quality. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] As a critical load-bearing foundation for transportation infrastructure such as highways, railways, and airport runways, the quality of roadbed construction directly determines the overall safety, durability, and life-cycle performance of the project. Currently, the inspection of roadbed construction quality still heavily relies on traditional discrete sampling methods. Given the continuously expanding scale of engineering construction and the shortage of professional and technical personnel, traditional methods are no longer sufficient to support the urgent need for "intelligent construction" of transportation infrastructure.

[0004] To overcome the bottlenecks of traditional testing, the industry has begun exploring intelligent fields such as time-domain reflectometry, 3D laser scanning, and vehicle-mounted mobile measurement in recent years, and has made some progress. However, although these individual technologies have shown certain advantages in some experimental scenarios, they have exposed deep technical defects in complex and harsh real engineering environments: multiple indicators are isolated from each other, lacking comprehensive testing capabilities, and unable to guide subsequent road rollers to perform precise targeted compaction and elevation repair; penetration testing methods cannot adapt to compacted roadbeds, and lack the ability to perform high-precision continuous scanning of roadbed geometry and standardized indicator extraction, and have poor resistance to bumps and positioning robustness. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides an integrated in-situ testing system and method for multiple indicators of roadbed construction quality. This system can directly and accurately extract the geometric morphology and comprehensive compaction quality indicators in complex in-situ environments, providing a reliable scientific basis for high-quality management, targeted compaction, and intelligent decision-making in roadbed engineering.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: The first aspect of the present invention provides an integrated in-situ testing system for multiple indicators of roadbed construction quality.

[0007] In one or more embodiments, an integrated in-situ detection method for multiple indicators of roadbed construction quality is provided, including: a robot platform and a compaction quality detection module, a geometric morphology indicator detection module and a main control and data fusion center mounted thereon; The compaction quality detection module is used for non-destructive and minimally invasive detection of the compaction degree and moisture content of the roadbed and transmits the data to the main control and data fusion center. The geometric shape index detection module is used to continuously acquire high-precision three-dimensional point cloud data of the roadbed surface, and combine it with the real-time posture data of the robot platform to automatically extract the geometric shape feature index of the roadbed and transmit it to the main control and data fusion center. The main control and data fusion center is used to: drive the robot platform to travel along the set cruise route and stop when it reaches the preset gridded physical detection station, and then sequentially activate the lateral translation, pre-drilling, and probe extrusion and penetration actions of the compaction quality detection module; during the measurement of compaction degree and moisture content, capture the current relative coordinates as the position label of the measuring point, and then combine them into a position-quality label package; integrate the position-quality label package with the roadbed geometric shape feature index, and independently output a continuous detection report of compaction quality and geometric shape of the entire roadbed.

[0008] In one embodiment, the robot platform includes a chassis system and a U-shaped support platform, a top-mounted observation base, and a communication and control unit mounted thereon; the U-shaped support platform is provided with a groove structure for fixing and installing a compaction quality detection module; the top-mounted observation base is mounted on the U-shaped support platform, and the geometric morphology index detection module is mounted on the top-mounted observation base; the communication and control unit also communicates with a remote control unit.

[0009] In one embodiment, the compaction quality detection module includes a liftable support frame, and a lateral translation positioning mechanism, a pre-drilling unit, a multi-needle detection unit, a self-resetting split-type bottom multi-hole positioning guide seat, and a compaction degree and moisture content extraction unit installed in the frame. The pre-drilling unit is mounted on the lateral translation positioning mechanism and is used to pre-drill an array of micro-holes in the roadbed. The multi-needle detection unit is mounted in parallel with the pre-drilling component on the lateral translation positioning mechanism and is used to emit electromagnetic pulses into the soil and collect the propagation time and amplitude attenuation of the reflected echo. The self-resetting split-type bottom multi-hole positioning guide seat is installed at the bottom ground end of the bearing frame and is used to provide absolute vertical guidance in the early stage of penetration and automatically avoid obstacles in the late stage of penetration to ensure that the probe penetrates the soil for the entire stroke. The compaction degree and moisture content extraction unit is used to receive the electromagnetic echo signal collected by the multi-needle detection unit and extract the moisture content and compaction degree of the roadbed soil in a non-destructive and real-time manner through signal analysis and physical inversion algorithms.

[0010] In one implementation, the geometric shape index detection module includes a rigid mounting bracket, a point cloud collector, and a point cloud data processing and feature extraction unit; The point cloud collector is fixed on a rigid mounting bracket and is used to emit laser pulses to the road surface when the robot autonomously cruises along the unstructured roadbed, thereby acquiring the original three-dimensional point cloud data in the local coordinate system in real time; the point cloud data processing and feature extraction unit is used to extract the roadbed geometric shape indicators from the original three-dimensional point cloud data.

[0011] As one implementation method, in the main control and data fusion center, if the compaction degree or geometric shape of a certain measuring point does not meet the standard, a clear location defect warning is directly generated in the local engineering coordinate system, providing the construction party with intuitive guidance for targeted compaction or repair.

[0012] A second aspect of the present invention provides an integrated in-situ testing method for multiple indicators of roadbed construction quality.

[0013] In one or more embodiments, a multi-index integrated in-situ testing method for roadbed construction quality includes: On the set cruise route, the robot platform is driven to move and stop when it reaches the preset gridded physical inspection station. Then, the lateral translation, pre-hole drilling and probe extrusion of the compaction quality inspection module are activated in sequence. During the compaction degree and moisture content measurement, the current relative coordinates are captured as the position label of the measuring point, and then combined into a position-quality label package. During the movement of the driving robot platform, high-precision three-dimensional point cloud data of the roadbed surface is acquired synchronously and continuously. Combined with the real-time posture data of the robot platform, the geometric shape feature index of the roadbed is automatically extracted. The position-quality tag package is integrated with the geometric shape feature index of the roadbed, and a continuous inspection report on the compaction quality and geometric shape of the entire roadbed is independently output.

[0014] As one implementation method, before driving the robot platform to move, it also includes: The robot platform is moved to a known reference starting point at the roadbed construction site. The main control and data fusion center completes the self-test of each hardware module and establishes a local engineering coordinate system with this starting point as the origin, and initializes the chassis odometer and the laser SLAM (Simultaneous Localization and Mapping) mapping algorithm.

[0015] As one implementation method, the geometric morphological characteristics of the roadbed include: roadbed smoothness, cross slope, width, longitudinal elevation, and centerline deviation.

[0016] As one implementation method, when the probe in the compaction quality testing module is tightly bonded to the soil without air gaps, the main control and data fusion center issues a command to emit electromagnetic pulses to the probe for data acquisition and feature inversion, including: The electromagnetic pulse received signal was extracted and decoupled. The electromagnetic wave bidirectional propagation time difference Δt, which characterizes the soil moisture polarization response, and the apparent reflection coefficient S, which characterizes the soil porosity and skeleton conductivity, were separated and extracted from the received signal waveform. Based on the soil type pre-input for the current construction section, the system automatically calls up the "dry density-moisture content joint inversion model" for this type of soil, which has been calibrated in the laboratory and is pre-set, and simultaneously retrieves the standard maximum dry density corresponding to this soil type. The electromagnetic wave bidirectional propagation time difference Δt and apparent reflection coefficient S obtained by decoupling are input into the dry density-moisture content joint inversion model to obtain the dry density and mass moisture content of the current detection point. Then, the actual dry density calculated is compared with the standard maximum dry density called to obtain the actual compaction degree. During the probe penetration into the soil to perform the test, the coordinate space of the probe in the soil is recorded. This coordinate is used as a spatial location label and combined with the actual compaction degree and mass moisture content obtained by inversion to form a "location-mass" data package. After the test is completed, the probe is pulled out and reset.

[0017] As one implementation method, the geometric features of the roadbed, compaction degree and moisture content are compared one by one with the pre-imported design values ​​and preset allowable deviation thresholds. The system automatically determines whether the test indicators of each acceptance station and grid node are qualified, and completes the automated engineering verification of the data.

[0018] Compared with the prior art, the beneficial effects of the present invention are: This invention utilizes the synergistic effect of a compaction quality detection module, a geometric morphology index detection module, and a main control and data fusion center. The compaction quality detection module performs non-destructive, minimally invasive testing of the roadbed's compaction degree and moisture content. The geometric morphology index detection module continuously acquires high-precision three-dimensional point cloud data of the roadbed surface and, combined with real-time attitude data from a robot platform, automatically extracts the roadbed's geometric morphology features. During compaction degree and moisture content measurements, the main control and data fusion center captures the current relative coordinates as the location label for each measurement point, combining these into a location-quality label package and integrating it with the roadbed's geometric morphology features, independently outputting the entire section. The continuous monitoring report on the compaction quality and geometric morphology of the subgrade effectively solves the problems of low efficiency, discrete data, and heavy reliance on manual labor in traditional testing methods, as well as the problems of single equipment, single index testing, and fragmented information on internal compaction quality and surface geometric morphology in existing intelligent testing technologies. At the same time, it overcomes the technical pain points such as easy breakage of on-site probes in compacted subgrades, poor adhesion between probes and soil, and point cloud distortion under bumpy driving conditions. It realizes in-situ collaborative acquisition, spatial coordinate binding, automated verification and intelligent perception of subgrade compaction quality and geometric morphology indicators, providing integrated data support for subgrade construction quality acceptance, defect location and targeted compaction repair. Attached Figure Description

[0019] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0020] Figure 1 This is a three-dimensional structural schematic diagram of the integrated in-situ detection system for multiple indicators of roadbed construction quality according to an embodiment of the present invention; Figure 2 This is the supporting platform for the integrated in-situ testing system for multiple indicators of roadbed construction quality according to embodiments of the present invention; Figure 3 This refers to the lifting and bearing frame in this embodiment of the invention; Figure 4 This is a schematic diagram of the internal results of the in-situ detection system according to an embodiment of the present invention; Figure 5 This is the compaction quality detection module in the embodiments of the present invention; Figure 6 This is a self-resetting split-type bottom multi-hole positioning guide seat in the embodiment of the present invention; Figure 7 This is the geometric topography scanning module of this invention embodiment; Figure 8 This is the signal received by the time-domain reflection of electromagnetic waves in the embodiments of the present invention; Figure 9 This is the fixed calibration probe structure in the embodiments of the present invention; Figure 10 These are the calibration fitting curves of parameters A and B in the embodiments of the present invention; Figure 11 These are the calibration fitting curves of parameters C and D in the embodiments of the present invention; Figure 12 This is the analysis process of the detection system in an embodiment of the present invention.

[0021] The components include: 1. Compaction quality detection module; 2. Geometric morphology detection module; 3. Mobile intelligent robot platform; 4. Communication and control center; 5. High-mobility tracks; 6. Observation base; 7. U-shaped support platform; 1-1. First vertical servo actuator; 1-2. Probe connecting column; 1-3. Support frame slide rail; 1-4. Probe fixer; 1-5. Electromagnetic slider; 1-6. Second transverse slide rail; 1-7. First vertical slide rail; 1-8. Multi-needle probe; 1-9. Parallel micro-drill bit; 1-10. Second vertical servo actuator; 1-11. First transverse slide rail; 1- 12. Pilot hole drill bit connecting frame; 1-13. Lateral servo actuator; 1-14. Multi-axis synchronous rotary drilling drive box; 1-15. Pilot hole drill bit holder; 1-16. Lower pressure flange base; 1-17. First lateral slide rail; 1-18. Passive flaring guide; 1-19. First lateral slider; 1-20. Half-type split guide block; 1-21. Spring return assembly; 1-22. Passive flaring guide pulley block; 1-23. Guide seat positioning slider; 1-24. Guide seat fixing beam; 2-1. Point cloud scanner; 2-2. Point cloud scanning window; 2-3. Point cloud scanner base. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0024] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0025] If the compaction quality and geometric shape of the roadbed are not strictly controlled during the construction process, the soil is prone to stress relaxation and cumulative deformation under the dual effects of long-term complex traffic loads and natural environmental evolution (such as wet-dry cycles and freeze-thaw cycles). This can lead to uneven settlement of the roadbed, cracking of the pavement, rutting and bulging, and even instability and damage to the overall structure. This will not only significantly shorten the service life of the transportation infrastructure, but also bring serious safety hazards and incalculable maintenance and repair costs.

[0026] Traditional discrete sampling inspection methods often employ manual techniques such as sand cone and ring cutter methods for internal physical indicators like compaction, while surface geometric indicators like flatness, elevation, and deviation rely on conventional instruments like 3m straightedges, levels, and total stations. However, existing methods suffer from significant industry-wide limitations: extremely low inspection efficiency, heavy reliance on manual experience, insufficient intelligence, discrete data, and difficulty in achieving continuous dynamic inspection. For example, for a highway subgrade with a total height of 5m and a compacted layer thickness of 20cm, the "Highway Engineering Quality Inspection and Evaluation Standard" requires at least 250 testing points per kilometer. Using traditional manual inspection not only consumes enormous manpower and resources but also results in "lagging sampling," completely failing to provide real-time feedback and control during the construction process. This is especially true in high-temperature, high-altitude, and oxygen-deficient environments, where traditional manual inspection methods face severe challenges. Given the current expanding scale of engineering construction and the shortage of professional and technical personnel, traditional methods are no longer sufficient to support the urgent need for "intelligent construction" of transportation infrastructure.

[0027] The multiple indicators are fragmented and lack comprehensive testing capabilities, which manifests in the following ways: Most existing equipment operates on a "single equipment, single indicator" model, which cannot comprehensively detect internal physical compaction data and surface three-dimensional geometry. This results in quality defect data becoming "suspended data" lacking coordinates, which cannot guide subsequent road rollers to perform precise targeted compaction and elevation repair.

[0028] Penetration testing methods are unsuitable for compacted roadbeds, as evidenced by the following: After being compacted by heavy rollers, the soil of the roadbed becomes extremely hard. When using electromagnetic probes such as Time Domain Reflectometry (TDR) for in-situ testing, if the probe is directly forced into the soil, the slender probe is prone to bending or even breaking. If the operator assists the insertion by shaking or enlarging the hole, it will inevitably damage the original soil structure, resulting in "air gaps" that are difficult to detect with the naked eye between the probe surface and the surrounding soil. Since the dielectric constant of air (approximately 1) is much lower than that of soil, these tiny air gaps will cause severe distortion of the electromagnetic wave reflection signal, rendering the final moisture content and compaction data completely unusable for engineering reference.

[0029] The system lacks the ability to perform high-precision continuous scanning and standardized index extraction of roadbed geometry, and its resistance to bumps and positioning robustness is poor, as manifested in: In recent years, although some studies have attempted to mount 3D laser scanners on vehicle-mounted platforms for roadbed topography scanning, most of them have only reached the stage of rough collection of raw point cloud data. They lack virtual measurement algorithms that are aligned with national engineering quality inspection and evaluation standards (such as the 3m ruler method and the leveling instrument method), and cannot achieve automated and standardized extraction of key geometric feature indicators such as flatness, cross slope, longitudinal elevation, and centerline deviation.

[0030] This invention addresses the problems of low efficiency, discrete data, and heavy reliance on manual labor in traditional testing methods, as well as the limitations of existing intelligent testing technologies, such as single-device, single-index testing, and the disconnect between internal compaction quality and surface geometry information. It also overcomes technical challenges such as the easy breakage of on-site probes in compacted roadbeds, poor adhesion between probes and soil, and point cloud distortion under bumpy driving conditions. The invention provides an integrated in-situ testing system and method for multiple roadbed construction quality indicators. This system can be applied to the intelligent and rapid evaluation and acceptance of multiple roadbed construction quality indicators, achieving efficient integrated testing and intelligent perception of multiple roadbed construction quality indicators.

[0031] like Figure 1 As shown, this embodiment of the invention provides an integrated in-situ detection method for multiple indicators of roadbed construction quality, including: a robot platform and a compaction quality detection module, a geometric morphology indicator detection module, and a main control and data fusion center mounted on it; The compaction quality testing module is used for non-destructive and minimally invasive testing of the compaction degree and moisture content of the roadbed and transmits the data to the main control and data fusion center. The geometric shape index detection module is used to continuously acquire high-precision three-dimensional point cloud data of the roadbed surface, and combine it with the real-time posture data of the robot platform to automatically extract the geometric shape feature index of the roadbed and transmit it to the main control and data fusion center. The main control and data fusion center is used to: drive the robot platform along the set cruise route and stop when it reaches the preset gridded physical inspection station, and then sequentially activate the lateral translation, pre-drilling, and probe extrusion and penetration actions of the compaction quality inspection module; during the measurement of compaction degree and moisture content, capture the current relative coordinates as the position label of the measuring point, and then combine them into a position-quality label package; integrate the position-quality label package with the roadbed geometric shape feature index, and independently output a continuous inspection report of compaction quality and geometric shape of the entire roadbed.

[0032] In some specific embodiments, the robot platform includes a chassis system and a U-shaped support platform, a top-mounted observation base, and a communication and control unit mounted thereon. The U-shaped support platform has a groove structure for fixing and installing the compaction quality detection module. The top-mounted observation base is mounted on the U-shaped support platform, and the geometric morphology index detection module is mounted on the top-mounted observation base to ensure an open field of view for laser scanning and unobstructed reception of satellite signals. The communication and control unit also communicates with the remote control unit, supporting manual intervention and remote control by operators beyond line of sight, or fully autonomous grid-based cruise detection by the main control and data fusion center.

[0033] Chassis system such as Figure 2As shown: A tracked mobile robot platform capable of traversing unstructured bumpy roads is used. The top of the vehicle body has a horizontally penetrating groove structure and an observation base. The compaction quality minimally invasive detection module is embedded and securely installed within the groove structure, so that the probe is protected circumferentially by the groove structure during lifting and penetration, avoiding damage due to lateral collisions under complex working conditions.

[0034] like Figure 3 As shown, the compaction quality minimally invasive detection module is embedded in the groove structure and is the core execution hardware for non-destructive extraction of roadbed compaction and moisture content indicators. The outermost part of the module is an integral lifting and bearing frame, which is connected to the vehicle body through the surrounding sliding rails. Its core function is to retract the internal precision components as a whole during robot cruising to prevent damage to the chassis caused by unstructured potholes; and to lower it to the ground when it reaches the detection position to provide an absolutely vertical compressive support benchmark. The frame is equipped with a lateral servo translation positioning mechanism, a pre-drilling unit, a multi-needle detection unit, and a self-resetting split multi-hole positioning guide seat. The lateral servo translation positioning mechanism is installed in the middle of the frame. Its function is to provide high-precision horizontal lateral displacement, ensuring that after the drilling operation is completed, the probe can be accurately translated and coaxially aligned with the micro-hole just drilled, realizing the alternation of "drilling" and "needle insertion" in the same physical coordinate system.

[0035] The pre-drilling unit, installed on the side of the translation positioning mechanism and fixed to the lifting frame via slide rails, is designed to pre-cut the compacted, hard subgrade soil using a drill bit to create a micro-hole array, thus opening a complete downward channel for subsequent probe penetration. The pre-drilling unit is used to pre-drill a micro-hole array in hard subgrades. It includes: a second vertical servo actuator: vertically mounted on the lateral translation positioning mechanism, providing constant and controllable downward drilling feed pressure and upward pull-out force for the subsequent multi-needle probe unit; a multi-axis synchronous rotary drilling drive box: installed below the second vertical servo actuator, containing a rotary drilling servo motor. The motor's output shaft transmits rotational power at a constant speed and synchronously to multiple parallel micro-drilling needles below via a precision gear transmission group (or synchronous belt transmission mechanism); and a micro-drilling needle array: its number and spatial distribution spacing are completely consistent with the subsequent probes.

[0036] The diameter of the drill bit (for example, setting the drill diameter to be...) d 1. The probe diameter is d 2, satisfying 0.7 d 2≤ d 1< d 2) It can be determined using the following formula: d 1 = 0.8 d 2, of which: d 1 represents the drill bit diameter. d2 represents the probe diameter. This aperture difference design is key to this detection method: the hard aggregate is pre-damaged to form guiding micropores, while the original soil structure around the pore wall is preserved, so that when a slightly coarser probe is inserted later, the soil around the pore wall can be compacted again, ensuring a dense physical fit between the probe surface and the soil without air gaps.

[0037] A multi-needle probe unit is mounted alongside the pre-drilling assembly on a lateral translation positioning mechanism. It includes: a first vertical servo actuator (or precision linear slide module): used to precisely and vertically press the probe into the pre-drilled micro-hole, comprising a multi-needle probe mounted below the first vertical servo actuator. This probe consists of an inner conductor signal waveguide and an outer conductor shielded waveguide arranged in an array. Its top is connected to a signal processing host mounted on a robot platform via a connector with a built-in impedance converter and a low-loss coaxial cable. This probe is used to emit electromagnetic pulses into the soil and collect the propagation time and amplitude attenuation of the reflected echo.

[0038] A self-resetting, split-type multi-hole positioning guide seat is installed at the bottom ground end of the supporting frame. It provides absolute vertical guidance during the initial penetration stage and automatically avoids obstacles during the final penetration stage to ensure the probe's full-stroke penetration into the soil. Its key feature is that the guide seat employs a passively opening and closing mechanical linkage structure, specifically including: a split-type guide block: composed of two symmetrical semi-assembled metal blocks joined together. When the two halves are in the closed state, the semi-circular grooves at their joint surfaces form a vertical guide through-hole that perfectly matches the spatial arrangement of the micro-drill bit and probe. An elastic reset component: preferably a tension spring laterally positioned on the outside of the left and right guide blocks. In its natural state, the tension spring tightly presses the two guide blocks inward, maintaining the closed guiding state of the guide through-hole. A passively flared guide pulley group: obliquely installed on the top inner edge of the left and right guide blocks. Its mechanical avoidance and reset principle is as follows: During the initial and middle stages of the time-domain reflectometry probe's downward penetration, the closed guide seat provides rigid radial sliding support for the slender probe, preventing it from buckling under pressure; when the penetration is about to reach its deepest point, the wider pressing flange or lower pressing base above the probe contacts the guide pulley group. As the servo actuator pushes further downward, the lower pressing base converts the downward vertical thrust into a lateral component through the pulleys, forcing the left and right guide blocks to overcome the spring tension and slide horizontally to both sides; when the probe is fully penetrated, the multi-hole positioning guide seat is completely opened into an open state. This design completely eliminates the "blind zone" caused by the thickness of the traditional fixed guide seat, ensuring that the effective detection section of the probe is 100% fully inserted and adhered to the subgrade soil, eliminating the edge dielectric effect caused by the probe top being exposed to the air, thus fully guaranteeing the absolute accuracy of electromagnetic wave testing. When the detection is completed, the probe is lifted upward by the servo actuator and detached from the pulley block. Under the pullback stress of the spring, the left and right guide blocks automatically merge and reset to the center, restoring the closed guide state, waiting for the next lateral movement or cross-point operation.

[0039] Compaction and moisture content extraction unit: This unit is embedded in the signal processing host or main control and data fusion center mounted on the robot platform. It is used to receive electromagnetic echo signals collected by the multi-needle detection unit and extract the moisture content and compaction index of the subgrade soil in a non-destructive and real-time manner through signal analysis and physical inversion algorithms.

[0040] The geometric shape index detection module includes a rigid mounting bracket, a point cloud collector, and a point cloud data processing and feature extraction unit. The point cloud collector is fixed on the rigid mounting bracket and is used to emit laser pulses to the road surface when the robot autonomously cruises along the unstructured roadbed to acquire the original three-dimensional point cloud data in the local coordinate system in real time. The point cloud data processing and feature extraction unit is used to extract the roadbed geometric shape index from the original three-dimensional point cloud data.

[0041] Specifically, the geometric shape index detection module is stably mounted on the top observation base of the mobile robot platform, employing a rigid mounting structure. It continuously acquires high-precision 3D point cloud data of the roadbed surface and, combined with the robot platform's real-time attitude data, automatically extracts roadbed geometric shape features such as smoothness, cross slope, width, longitudinal elevation, and centerline deviation through algorithms. This module includes a rigid mounting bracket, a point cloud collector, and a point cloud data processing and feature extraction unit. The rigid mounting bracket is made of high-strength, lightweight alloy material, rigidly fixed at its bottom to the robot's U-shaped support platform, and equipped with a quick-release clamp at its top. The point cloud collector is fixed within the quick-release clamp at the top of the rigid mounting bracket. Due to the rigid connection, the scanner's local coordinate system remains fixed and relatively stationary with respect to the robot platform's body coordinate system. As the robot autonomously cruises along the unstructured roadbed, the scanner emits laser pulses towards the road surface, acquiring the original 3D point cloud data in the local coordinate system in real time. The point cloud data processing and feature extraction unit based on attitude compensation addresses the spatial distortion of rigidly mounted scanning data caused by the robot's attitude changes during movement due to the soft and uneven surface of the unstructured roadbed. Embedded in the main control and data fusion center, this unit extracts morphological indicators, including but not limited to roadbed smoothness, cross slope, centerline deviation, width, and longitudinal elevation, through software leveling and spatial geometric algorithms.

[0042] like Figure 1 and Figure 2 As shown, the system consists of a compaction quality detection module 1, a geometric shape scanning module 2, a tracked mobile robot platform 3, a communication and control center 4, high-mobility tracks 5, an observation base 6, and a U-shaped support platform 7. Figure 3 , Figure 4 and Figure 5 As shown, the compaction quality inspection model 1 is embedded in the U-shaped support platform 7 via an electromagnetic slider 1-5. The compaction quality inspection model 1 mainly includes: a first vertical servo actuator 1-1, a probe connecting column 1-2, a support frame slide rail 1-3, a probe holder 1-4, an electromagnetic slider 1-5, a second horizontal slide rail 1-6, a first vertical slide rail 1-7, a multi-needle probe 1-8, a parallel micro-drill bit 1-9, a second vertical servo actuator 1-10, a first horizontal slide rail 1-11, a drill bit connecting column 1-12, a horizontal servo actuator 1-13, a multi-axis synchronous rotary drilling drive box 1-14, a drill bit holder 1-15, a lower pressure base 1-16, a second vertical slide rail 1-17, a passive flared guide pulley group 1-18, a second horizontal slider 1-19, a split guide block 1-20, and a spring return assembly 1-21.

[0043] Figure 3This is a schematic diagram of a lifting frame, which is installed within the U-shaped groove of the U-shaped support platform 7 via electromagnetic sliders 1-5 and support frame slide rails 1-3. Figure 4 and Figure 5 As shown, the multi-needle detection unit adopts the following structure: a first vertical servo actuator 1-1 is installed on the upper part of the probe connecting column 1-2. The probe connecting column 1-2 is connected to the probe holder 1-4. The function of the probe holder 1-4 is to fix and connect the multi-needle probe 1-8. The probe connecting column 1-2 is fixedly installed on the lifting frame through the first vertical slide rail 1-7. The multi-axis pre-hole drilling unit adopts the following structure: a second vertical servo actuator 1-10 is installed on the upper part of the multi-hole drill bit connecting column 1-12. The multi-axis synchronous rotary drilling drive box 1-14 is installed on the lower part of the drill bit connecting column 1-12. The lower part of the drive box 1-14 is connected to the pre-hole drill bit holder 1-15. The multi-hole drill bit connecting column 1-12 is installed on the lifting frame through the second vertical slide rail 1-17. The pilot hole holder 1-15 is used to fix and drive the parallel micro pilot hole drill 1-9 to rotate, while the second vertical servo actuator 1-10 provides a continuous downward thrust to the parallel micro pilot hole drill 1-9. The lateral servo translation and positioning mechanism mainly includes: a lateral servo actuator 1-13, a first lateral slide rail 1-11, and a second lateral slider 1-19. The first lateral slide rail 1-11 and the second lateral slider 1-19 are horizontally installed inside the lifting frame; the probe connecting post 1-2 and the pilot hole connecting post 1-12 are installed on the first lateral slide rail 1-11 and the second lateral slider 1-19. The lateral servo actuator 1-13 provides lateral thrust to the probe connecting post 1-2 and the pilot hole connecting post 1-12 to control the horizontal movement and positioning of the multi-needle detection unit and the multi-axis pre-drilling unit. Figure 6 This is a self-resetting, split-type bottom multi-hole positioning guide unit. The unit consists of a passive flaring guide 1-18, two split guide blocks 1-20, a spring return assembly 1-21, a passive flaring guide pulley assembly 1-22, a guide seat positioning slider 1-23, and a guide seat fixing beam 1-24. The self-resetting, split-type bottom multi-hole positioning guide unit is fixed to the bottom of the lifting frame via the guide seat fixing beam 1-24. The passive flaring guide pulley assembly 1-22 and the lowering base 1-16 are aligned on the same vertical axis. In the inspection state, the spring return assembly 1-21 provides continuous pressure to the two split guide blocks 1-20, keeping them in a closed state. Figure 7 The module is for scanning the geometric dimensions of the roadbed. It consists of a point cloud scanner 2-1, a point cloud scanning window 2-2, and a point cloud scanner base 2-3.

[0044] Main Control and Data Fusion Center: Serving as the control center of the entire detection system, this center is housed within the robot platform's computing host. It coordinates the timing of actions across various hardware modules and enables spatiotemporal registration and deep fusion of heterogeneous detection data. Specifically, this center includes a multi-module collaborative control unit, a spatiotemporal synchronization and data binding unit, and a visualization output unit. Multi-module collaborative control unit: Responsible for the collaborative operation control of "continuous scanning and fixed-point detection". Along the set navigation route, the main control and data fusion center drive the robot platform, controlling the point cloud collector to perform continuous non-contact topographic scanning during movement. When the robot reaches the preset gridded physical inspection station, the main control and data fusion center issues a command to stop the chassis, then sequentially activates the lateral translation, pre-drilling, and probe insertion actions of the compaction quality inspection module. After the compaction degree and moisture content data of the current measuring point are collected and the probe is safely withdrawn and reset, the main control and data fusion center issues another movement command, driving the robot to the next measuring point, achieving fully unmanned collaborative operation.

[0045] Measurement point relative positioning and data recording unit: After initialization at the starting point, it integrates chassis odometer, inertial navigation (IMU), and laser SLAM algorithms to calculate the robot's current coordinates (i.e., equivalent station number and offset) in the local engineering coordinate system in real time. During compaction and moisture content measurements, the current relative coordinates are captured as the location label for that measurement point.

[0046] Data decoupling and parallel output unit: Geometric topography indicators (area data) and internal physical indicators (point data) are processed in parallel. The acquired compaction degree, moisture content, and captured relative coordinate points are combined into a location-quality tag package, and an independent continuous inspection report on the compaction quality and geometric topography of the entire roadbed is output. If the compaction degree or geometric topography of a certain measuring point does not meet the standard, a clear location defect warning is directly generated in the local engineering coordinate system, providing the construction party with intuitive guidance for targeted compaction or repair.

[0047] The system in this embodiment of the invention includes a mobile robot platform, a compaction quality detection module embedded in the platform's U-shaped groove, a top-mounted, rigidly mounted geometric morphology index detection module, and a main control and data processing center. The compaction quality detection module innovatively adopts a collaborative mechanism of "drilling micro-holes for guidance, followed by coaxial extrusion penetration," coupled with a passive avoidance guide seat, eliminating blind spots and air gaps in the soil, and accurately inverting compaction degree and moisture content. The geometric morphology module combines high-precision POS (Position and Orientation System) data to dynamically compensate the attitude of the three-dimensional point cloud, and has a built-in "virtual measurement tool" to automatically extract five geometric indicators, including flatness and cross slope, conforming to national acceptance standards. The main control and data fusion center integrates laser positioning technology to determine spatial coordinates, decoupling and marking physical quality data with spatial coordinates. This invention effectively solves the problems of low efficiency, discrete data, and strong reliance on manual labor in traditional detection methods, as well as the problems of single equipment, single index detection, and the separation of internal compaction quality and surface geometric morphology information in existing intelligent detection technologies. At the same time, it overcomes the technical pain points such as easy breakage of probes in compacted roadbeds, poor adhesion between probes and soil, and point cloud distortion under bumpy driving conditions.

[0048] The integrated in-situ intelligent detection system for multiple roadbed indicators in this embodiment operates as follows in routine open roadbed engineering: The operator starts the mobile robot platform via a handheld wireless remote control terminal, manually controlling its movement along the roadbed centerline at an appropriate operating speed. During remote control operation, the compaction quality micro-invasive detection module 1, controlled by the overall lifting support frame, is in a "high-level retracted" mechanical protection state to prevent damage to precision components caused by unstructured potholes. Simultaneously, the geometric shape scanning module begins continuous operation, collecting real-time raw three-dimensional point cloud data of the roadbed surface and synchronously recording the high-precision position and attitude information of the equipment. When the operator observes or the system terminal indicates that the robot has been remotely driven to the preset compaction degree sampling grid node, the operator issues a command via the remote control, and the robot stops. The main control and data fusion center simultaneously receive the command and control the overall lifting support frame to move downwards along the vertical slide rail until the bottom "self-resetting split multi-hole positioning guide seat" is approximately 0.5cm close to the roadbed surface. This step utilizes the robot's own weight to establish an absolutely vertical and stable mechanical compressive reference surface for subsequent micro-drilling and pin insertion actions. Once the system is stable, the lateral servo translation positioning mechanism precisely aligns the "multi-axis pre-drilling unit" with the array of holes on the "self-resetting split-type multi-hole positioning guide seat." Simultaneously, the second vertical servo actuator 1-10 presses down while the multi-axis synchronous rotary drilling drive box 1-14 synchronously drives the array of micro-drill bits with a diameter of d1 to rotate at high speed, cutting downwards into the hard subgrade soil. After drilling to a preset depth (e.g., a compacted layer thickness of 20cm), it is pulled out in the opposite direction, leaving neatly arranged guide micro-holes in the undisturbed soil. Subsequently, the lateral servo actuator 1-13 drives the "multi-needle detection unit" to precisely move laterally by a fixed offset distance, ensuring that the probe axis of the "multi-needle detection unit" is precisely coaxially aligned with the drilled micro-holes. The first vertical servo actuator 1-1 drives the probe with a diameter of d2 downwards. During the first two-thirds of the penetration stroke, the positioning guide seat remains closed under the pressure of the spring return assembly 1-21. Its internal rigid through-hole provides sufficient radial support for the slender probe, preventing deviation or breakage during penetration. As the probe approaches complete penetration, the downward-pressing flange base 1-16 above the probe contacts the passively flared guide pulley assembly 1-22 at the top of the positioning guide seat. Through the inclined plane force, it forcibly overcomes the spring pressure, horizontally pushing the two halves of the split guide blocks 1-20 open to the sides. As the probe gradually penetrates, the guide seat gradually opens until the probe is fully embedded in the soil. Because the probe diameter is larger than the micro-hole diameter, the thicker probe, during the forced downward pressure, further compacts the undisturbed soil around the hole wall, completely eliminating air gaps and achieving an absolutely dense physical fit. Once the probe has fully penetrated the soil, manual operation or a command issued by the main control and data fusion center will begin the acquisition of compaction quality data and spatial coordinates at that point. After the data acquisition and calibration are completed, the first vertical servo actuator 1-1 will pull the probe upward.Once the probe flange disengages from the pulley block, the left and right guide blocks of the guide seat automatically close and reset instantly under the high-strength pull of the springs. Immediately afterwards, the overall lifting support frame retracts the compaction quality minimally invasive detection module back into the vehicle body. The mobile robot restarts, continues performing continuous geometric scanning of the next segment, and moves towards the next grid measurement point. The entire process can be remotely controlled or automated closed-loop detection.

[0049] In one or more embodiments, the integrated in-situ testing method for multiple indicators of roadbed construction quality includes: On the set cruise route, the robot platform is driven to move and stop when it reaches the preset gridded physical inspection station. Then, the lateral translation, pre-hole drilling and probe extrusion of the compaction quality inspection module are activated in sequence. During the compaction degree and moisture content measurement, the current relative coordinates are captured as the position label of the measuring point, and then combined into a position-quality label package. During the movement of the driving robot platform, high-precision three-dimensional point cloud data of the roadbed surface is acquired synchronously and continuously. Combined with the real-time posture data of the robot platform, the geometric shape feature indicators of the roadbed are automatically extracted, and a continuous geometric shape inspection report of the entire roadbed section is independently output. Among them, the geometric shape feature indicators of the roadbed include, but are not limited to: roadbed smoothness, cross slope, width, longitudinal elevation, and centerline deviation.

[0050] Before the robot platform moves forward, it also includes: The robot platform is moved to a known reference starting point at the roadbed construction site. The main control and data fusion center completes the self-test of each hardware module and establishes a local engineering coordinate system with this starting point as the origin, and initializes the chassis odometer and laser SLAM mapping algorithm.

[0051] The main control and data fusion center flexibly schedules the robot platform's movement mode based on on-site working conditions and testing requirements, and simultaneously completes non-contact 3D mapping of the roadbed surface. Specifically, this includes the following steps: (1) Dual-mode maneuvering and continuous point cloud acquisition: The operator can choose the fully autonomous cruise mode (the main control and data fusion center will autonomously follow the grid based on the preset route and chassis SLAM mapping) or the unlimited distance remote control mode (the system will receive the control instructions from the remote control center beyond line of sight through the 5G / 4G private network communication gateway to allow manual intervention). During the robot's movement on the unstructured roadbed, the point cloud acquisition device fixed on the top rigid support will continuously emit laser pulses to the roadbed to acquire the original three-dimensional point cloud data in the local coordinate system in real time. P local ( x , y , z ).

[0052] (2) Dynamic attitude compensation and mapping based on POS data: At the instant the original point cloud is acquired, the system simultaneously extracts the real-time pose (POS) data from the high-precision satellite positioning and attitude measurement module rigidly bound to the point cloud collector. Using the real-time acquired heading, pitch, and roll angles, a rotation matrix is ​​constructed. R With translation vector T The local point cloud is mapped to the absolute geographic coordinate system in high frequency and in real time. The calculation formula is as follows:

[0053] This step eliminates spatial distortion caused by pitch and roll of the chassis on soft, potholed roads. The system then uses voxel filtering to remove outlier noise such as construction dust.

[0054] (3) Automated extraction of geometric morphology indicators Point cloud data based on attitude compensation and absolute geographic coordinate system P glocal The point cloud data processing and feature extraction unit synchronously executes spatial geometry algorithms in the background, outputting the following five metrics: a. Roadbed smoothness index: Adhering to the "3m straightedge method" in the standard. The algorithm automatically projects a virtual straight line benchmark of 3m length onto the point cloud surface along the longitudinal direction of the roadbed at predetermined acceptance intervals (e.g., 2 points every 200m). The elevation values ​​of all point cloud points directly below the benchmark are calculated. z i Extract the maximum gap value from the vertical distance to the baseline. h max This value is directly used as the flatness test value of the measuring point and compared with the allowable deviation required by the specification.

[0055] b. Centerline offset extraction: The algorithm first automatically extracts the edge lines of the shoulders on both sides at the junction of the roadbed surface and the slope based on the abrupt changes in surface curvature and normal vector; then it calculates the spatial geometric center coordinates of the edge points on both sides of the cross section at the current station number. X c , Y c ) is used as the measured centerline point. Its coordinates are compared with the designed centerline coordinates for that station. X 0, Y 0) Perform distance calculation and output the lateral centerline offset error. At curves, the algorithm automatically encrypts the offset extraction of HY (gradual round point) and YH (round-gradual point).

[0056] c. Longitudinal elevation extraction: Longitudinal sampling is performed along the extracted measured centerline at preset step lengths. The median elevation of the surface point cloud in the local neighborhood is extracted to generate the measured longitudinal elevation curve. The difference between the curve and the design elevation curve is calculated, and the elevation deviation of the corresponding station number is automatically output.

[0057] d. Width extraction: At the specified acceptance station section, the algorithm directly calculates the horizontal straight-line distance between the extracted left shoulder edge point and the right shoulder edge point to obtain the measured roadbed width W, and determines whether it meets the design requirements.

[0058] e. Cross slope extraction: A dynamic longitudinal profile slicing method is used. Cross profile slices with fixed spacing are generated along the roadbed direction. In the transverse coordinate... y with elevation coordinates z In the two-dimensional profile, the least squares method is used to perform linear fitting on the top surface point cloud. z = ay + b ), calculate slope a Determine the actual cross slope angle α = arctan( a The cross slope deviation is generated by comparing it with the design value.

[0059] When the robot reaches the preset physical detection grid node through SLAM mileage estimation, the main control and data fusion center issues a braking command. The compaction quality detection module executes automated actions sequentially: First, the vertical servo actuator drives the drilling unit downward, drilling an array of micro guide holes on the roadbed surface before pulling it out. Then, the lateral servo actuator moves horizontally, aligning the multi-needle probe unit with the micro-holes. Supported and guided by the self-resetting guide seat, the probe precisely penetrates the micro-hole, achieving a tight, air-gap-free fit between the probe and the soil. When the probe in the compaction quality detection module achieves a tight, air-gap-free fit with the soil, the main control and data fusion center sends a command to emit electromagnetic pulses to the probe for data acquisition and feature inversion, including: The electromagnetic pulse received signal was extracted and decoupled. The electromagnetic wave bidirectional propagation time difference Δt, which characterizes the soil moisture polarization response, and the apparent reflection coefficient S, which characterizes the soil porosity and skeleton conductivity, were separated and extracted from the received signal waveform. Based on the soil type pre-input for the current construction section, the system automatically calls up the "dry density-moisture content joint inversion model" for this type of soil, which has been calibrated in the laboratory and is pre-set, and simultaneously retrieves the standard maximum dry density corresponding to this soil type. The electromagnetic wave bidirectional propagation time difference Δt and apparent reflection coefficient S obtained by decoupling are input into the dry density-moisture content joint inversion model to obtain the dry density and mass moisture content of the current detection point. Then, the actual dry density calculated is compared with the standard maximum dry density called to obtain the actual compaction degree. During the probe penetration into the soil to perform the test, the coordinate space of the probe in the soil is recorded. This coordinate is used as a spatial location label and combined with the actual compaction degree and mass moisture content obtained by inversion to form a "location-mass" data package. After the test is completed, the probe is pulled out and reset.

[0060] like Figure 12 As shown, the main focus is on how the main control and data fusion center accurately calculates the compaction degree, moisture content and geometric morphology indicators required for the project through underlying signal processing and spatial algorithms after the multi-needle detection unit completes the absolutely seamless bonding and the geometric morphology scanning module acquires continuous scanning data.

[0061] When testing the compaction degree and moisture content of the subgrade, the probe is fully penetrated into the soil. The multi-probe detection unit emits picosecond-level electromagnetic signals into the soil along the probe and receives the reflected signals. Simultaneously, the main control and data fusion center processes and filters the echo signals of the electromagnetic waves, interpreting and obtaining the propagation time Δ of the electromagnetic waves within the soil. t and apparent reflectance coefficient S Propagation time Δ t and apparent reflectance coefficient S The methods for obtaining it are as follows: The main control and data fusion center received, such as Figure 8 The electromagnetic wave time-domain reflection signal shown is Figure 8 The horizontal axis represents the time when the main control and data fusion center receives the electromagnetic wave signal, and the vertical axis represents the ratio of the amplitude of the currently received electromagnetic wave signal to the initial amplitude, i.e., the apparent reflection coefficient. S Generally, time-domain reflection signals have three reflection peaks, the first peak point ( t 1, S 1) is the reflected signal transmitted back from the impedance change at the top of the probe, the second peak point and the first peak point ( t 2, S 2) is the reflected signal transmitted back from the impedance change at the bottom of the probe, the third peak point ( t 3, S 3) The electromagnetic wave reflected from the bottom of the probe is reflected back into the soil when it reaches the top of the probe, and then propagates again within the soil. Therefore, the bidirectional travel time Δ of the electromagnetic wave propagating along the probe within the soil is... t = t 2- t 1= t 3- t 2. The compaction degree and moisture content of the roadbed can be obtained by the inversion model of formulas (1) to (6). At the moment the physical index is obtained, the main control and data fusion center simultaneously capture the three-dimensional coordinates of the point. X , Y (Z), which is used as a spatial location label and bound to the compaction quality data to generate a quality digital node with spatiotemporal coordinates.

[0062] (1) (2) (3) (4) (5) (6) in: S 1, S 2 and S 3 represents the amplitudes corresponding to the three reflection peaks in the reflected signal; L This refers to the probe length; c The speed of light in a vacuum can be taken as 299,792,458 m / s; K ins The actual apparent dielectric constant is directly measured by the test system; K T The apparent dielectric constant is corrected to account for the effects of temperature and probe size; ρ d This is the measured dry density of the roadbed; ρ dmax This represents the maximum dry density of the roadbed. ρ w The density of water; w The measured moisture content of the roadbed; T The current ambient temperature; A , B , C and D For calibration parameters; M This is the probe size correction factor, calibrated at the probe factory. β These are parameters related to the dielectric loss of the soil.

[0063] Probe size correction factor M The following steps can be used to determine this: S1. Prepare 3-5 groups of soil samples with different dry densities and moisture contents indoors, and place all samples in a constant temperature and humidity chamber at 25 degrees Celsius for no less than 24 hours. The total number of sample groups is counted as follows: N ; S2. Remove the sample from the constant temperature and humidity chamber and conduct indoor testing using a fixed calibration probe. In this embodiment, the fixed calibration probe is as follows: Figure 9 As shown, the fixed calibration probes are arranged in a three-pronged configuration, each 20cm long, 4mm in diameter, and 25cm apart. The fixed calibration probes are used to test each soil sample, and the dielectric constant of each soil sample measured by the fixed calibration probes is recorded as follows: k i , i The sample group number; S3. Using a probe that needs to be calibrated, perform indoor tests again on each group of soil samples to obtain the dielectric constant of each group of soil samples measured by the probe. K i , i The sample group number; S4. After the above tests, the corresponding calibrated probe M The value can be determined according to formulas (7) to (9). If the probe size used in the field is the same as the fixed calibration probe size, M You can directly take 1 without factory calibration.

[0064] (7) (8) (9) in, and They are respectively K i and k i The average value.

[0065] The calibration parameters required for formulas (1) to (6) A , B , C and D Determine using the following steps: S1. Collect 3 to 5 sets of subgrade soil samples on-site using the same sampling tube and compact them. The mass of the sampling tube is recorded as follows: m j The mass of the collection tube after being filled with soil is recorded as follows: , j The sample group number; S2. Using the probe mounted on the instrument, the soil inside the collection tube was tested sequentially to obtain the dielectric constant of each soil sample. K j And the soil samples of each group are calculated using formula (1). β j ; S3. Moisture content of the soil sampled by each collection tube. w j It can be obtained quickly by burning alcohol.

[0066] S4. Obtain all the above steps. w j , m j , , K j as well as β jThe data is input into the main control and data fusion center.

[0067] S5. The main control and data fusion center automatically calls the internal calculation program to calculate the dry density of each soil sample according to formula (10) and formula (11). ρ dj and the correction of apparent dielectric constant K Tj ,in h The height of the collection tube, K j The actual apparent dielectric constant is the one directly measured for each soil sample.

[0068] (10) (11) S6. The integration program within the main control and data fusion center automatically fits the following: Figure 10 Linear curve, calibration parameters A This is the intercept of the curve. B Let be the slope of the curve.

[0069] S7. The integration program within the main control and data fusion center automatically fits the following: Figure 11 Linear curve, calibration parameters C Let be the slope of the curve. D This is the intercept of the curve.

[0070] While performing high-precision single-point detection and spatial coordinate binding of the aforementioned internal physical indicators of the roadbed, the system simultaneously processes the continuously acquired planar geometric topography data to obtain geometric topography indicators such as roadbed smoothness, width, centerline deviation, longitudinal elevation, and cross slope. Because the surface of unstructured roadbeds is soft and uneven, the posture changes generated by the robot during movement can cause severe spatial distortion in the data acquired by the rigidly mounted 3D laser scanner. This invention abandons the traditional easily damaged mechanical shock-absorbing gimbal and adopts a "software leveling" algorithm based on high-precision POS (Position and Orientation System) data for dynamic posture compensation. The specific steps are as follows: S1. Obtaining the original 3D point cloud coordinate data P local ( x , y , z Simultaneously, the system's main control and data fusion center, through a hardware triggering mechanism, synchronously extract transient pose data rigidly bound to the point cloud collector, including: three-dimensional translation coordinates ( X t , Y t , Zt ), and the roll angle of the vehicle body (denoted as ...). φ ), pitch angle (denoted as θ ) and yaw angle (denoted as ψ ); S2. The system backend frequently constructs translation vector matrices. H = [ X t , Y t , Z t ] T And the three-dimensional rotation matrix from the scanner's local coordinate system to the absolute engineering geographic coordinate system. R .matrix R It is obtained by multiplying the rotation submatrices around the X, Y, and Z coordinate axes, and its specific expansion formula is as follows: (12) The rotation submatrices for each coordinate can be obtained using the following formula: (13) (14) (15) S3. The system performs a spatial affine transformation on each scanned point in the local original point cloud, mapping it to an absolute geographic coordinate system. The calculation formula is as follows: (16) Using this formula, the system performs "reverse rotation and translation compensation" on the distorted point cloud. Subsequently, the system uses an internal filtering algorithm to remove dust and multipath reflection noise from the construction site, outputting a high-precision absolute geographic coordinate system point cloud dataset. P global .

[0071] S4. Based on high-precision point cloud datasets P global The main control and data fusion center simultaneously implement the built-in spatial geometry virtual measurement algorithm to automatically extract key geometric morphology indicators of the roadbed project.

[0072] S5. Subgrade Smoothness Extraction. This extraction method conforms to the subgrade specification method, automatically projecting a perfectly horizontal line of length [value missing] onto the 3D point cloud surface. l A virtual straight baseline of 3000mm. Assume the absolute elevation of this baseline is... Z ref The algorithm traverses all point cloud sets within the local search radius directly below the baseline. Psub {( X i , Y i , Z i )}, calculate the vertical clearance from each point to the baseline. Extract the largest spacing deviation value and use this deviation value as the flatness test value for that point. H max : (17) S6. Roadbed Width Extraction. This extraction method is aligned with roadbed specifications. The algorithm traverses the cross-sectional point cloud and calculates the normal vector for each point. n With Gaussian curvature K g By setting a curvature abrupt change threshold, the system automatically extracts the spatial polylines of the shoulder edges on both sides at the junction of the roadbed top surface and the slope. After accurately identifying the spatial polyline points of the shoulder boundaries, the system automatically extracts the left edge points on the same cross-section. P L ( X L , Y L , Z L and the right edge point P R ( X R , Y R , Z R The distance between the two points on the horizontal projection plane is calculated to obtain the measured roadbed width. W : (18) S7. Centerline offset extraction. This extraction method is aligned with the roadbed specifications. First, the coordinates of the midpoint of the line connecting two points on the roadbed edge are calculated according to formula (19). X C , Y C ), and compare it with the design centerline coordinates input in advance by the system ( X C0 , Y C0 By comparing the two points, the plane normal geometric distance between them is calculated according to formula (20), and the measured centerline deviation error Δ is output. D (19) (20) S8. Longitudinal Elevation Extraction. This extraction method is aligned with roadbed specifications. Longitudinal sampling is performed along the extracted measured centerline at preset step lengths (e.g., every 10m). A system with a radius of [missing information] is constructed centered on the sampling points. r For a local circular neighborhood, calculate the average or median of the point cloud elevations within that neighborhood, and use this median as the measured elevation for that station. Z m Combine it with the design elevation Z d The difference is calculated to output the longitudinal elevation deviation Δ. Z .

[0073] S9. Roadbed cross slope extraction. A dynamic longitudinal profile slicing fitting method is used to extract a set of transverse point cloud data perpendicular to the road orientation from a specified acceptance section. (In the transverse coordinate...) y with elevation coordinates z In the constructed two-dimensional profile, the least squares method is used to linearly fit the point cloud on the top surface, and the objective function is to minimize the sum of squared residuals: (twenty one) The slope of the fitted curve is obtained by differentiating the objective function of formula (21). a The cross slope angle is obtained using formulas (22) and (23). α and the percentage of cross slope i slope Cross slope deviation Δ i Calculated according to formula (24), where i 0 represents the design cross slope of this section.

[0074] (twenty two) (twenty three) (twenty three) After completing the measurement of geometric morphology indicators, compaction degree, and moisture content, the system performs unified spatiotemporal registration and fusion of the extracted five planar geometric morphology indicators (smoothness, longitudinal elevation, centerline deviation, width, and cross slope) and point physical indicators (measured compaction degree and mass moisture content). The system then compares and verifies all the measured data with the pre-imported design values ​​and the allowable deviation thresholds in the "Highway Engineering Quality Inspection and Evaluation Standard." The system automatically determines whether each inspection indicator for each acceptance station and grid node is qualified, and directly maps these continuous physical attributes to the surface of the three-dimensional point cloud solid model of the geometric morphology, rendering a three-dimensional comprehensive quality color heat map that combines "morphological undulation" and "internal compaction density." For areas with a verification result of "unqualified (False)," the diagnostic algorithm automatically tracks the boundary of the defect polygon based on the abrupt changes in the contour lines of the heat map, accurately extracting the absolute spatial coordinate set {(X err , Y err , Z err Finally, the system automatically formats the original test data, automated verification results, and extracted abnormal coordinate points into a "Comprehensive Report on In-situ Testing of Multiple Indicators for Roadbed Construction Quality" according to engineering acceptance standards and archiving formats. This report not only includes traditional data reports but also intuitively outputs geometric comparison curves for each station section and a three-dimensional digital twin model of the entire road section's quality, completely replacing the extremely time-consuming manual data processing and chart drawing.

[0075] Taking the construction quality acceptance of the roadbed section (200m long) from K2+000 to K2+200 of a newly built expressway as an example, this paper fully demonstrates the data collection, index verification and targeted report generation process of this system in a real engineering environment.

[0076] The subgrade fill material for this section is silty clay, with a designed half-width of 11.25m and a designed cross slope of 2.0%. According to the "Highway Engineering Quality Inspection and Evaluation Standards," the acceptance criteria for this upper subgrade layer (0-30cm below the top surface) are set as follows: compaction degree ≥96%, maximum gap in smoothness ≤15mm, cross slope deviation ±0.3%, and width not less than the design value. Before testing, operators input the above standard thresholds and the maximum dry density (1.85 g / cm³) from the laboratory compaction test into the system's main control and data fusion center.

[0077] The mobile robot starts from K2+000 and cruises towards K2+200 at a speed of 4 km / h. The system continuously acquires roadbed point clouds and completes attitude compensation (software leveling) of the absolute coordinate system. At the K2+150 section, the system outputs specific geometric shape measurement data according to the virtual measurement algorithm. Evenness measurement: The maximum gap deviation extracted along the longitudinal direction using a virtual 3m ruler is 18mm (judgment: exceeding the limit, unqualified). Width and centerline deviation measurement: The roadbed width calculated from the left and right edge points is 11.30m (judgment: meets ≥11.25m, qualified); the centerline deviation error is 25mm (judgment: within the allowable range of ±50mm, qualified). Cross slope measurement: The actual cross slope angle calculated from the point cloud fitting of this section is converted to 1.85% (judgment: the deviation from the design value of 2.0% is within ±0.3%, qualified). When the robot reaches the designated inspection node (station K2+150, 3.5m to the left of the centerline), it automatically engages the parking brake. The compaction quality minimally invasive inspection module sequentially completes the coaxial extrusion penetration of the drilled micro-hole and probe. The probe emits a high-frequency electromagnetic pulse, and the system captures the echo and decouples the calculations to obtain the bidirectional propagation time difference Δ of the electromagnetic wave. t =1.25ns, calculated from the reflected signal β= 0.55. The main control and data fusion center calls the built-in "dry density-moisture content joint inversion model" and calculates the actual dry density of the subgrade soil at the measuring point as 1.78 g / cm³ and the mass moisture content as 14.2% according to the method in implementation method 2. The system calculates the ratio of the actual dry density to the maximum dry density (1.85 g / cm³) and obtains a compaction degree of 96.21% (judgment: ≥96%, qualified). The system simultaneously captures the three-dimensional coordinates of the point ( X 150 , Y 150 , Z 150 Pack the data.

[0078] The system integrates the continuous planar geometric point cloud with discrete point compaction coordinate data, and uses a spatial interpolation algorithm to render a "comprehensive color heat map of geometric shape and compaction" for the 200m road section on a 3D digital model. By comparing the above calculation results, the system automatically identifies a local area at K2+150 with an unevenness exceeding the limit (18mm), and generates a "Comprehensive Report on In-situ Testing of Subgrade Construction Quality Multiple Indicators from K2+000 to K2+200" for archiving.

[0079] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including functions for executing... Figure 12 The program code for the method shown. In such an embodiment, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by the central processing unit, it performs the various functions defined in the apparatus of this application.

[0080] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0081] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-index integrated in-situ testing system for roadbed construction quality, characterized in that, include: The robot platform, along with the compaction quality detection module, geometric morphology index detection module, and main control and data fusion center mounted on it; The compaction quality detection module is used for non-destructive and minimally invasive detection of the compaction degree and moisture content of the roadbed and transmits the data to the main control and data fusion center. The geometric shape index detection module is used to continuously acquire high-precision three-dimensional point cloud data of the roadbed surface, and combine it with the real-time posture data of the robot platform to automatically extract the geometric shape feature index of the roadbed and transmit it to the main control and data fusion center. The main control and data fusion center is used to: drive the robot platform to travel along the set cruise route and stop when it reaches the preset gridded physical detection station, and then sequentially activate the lateral translation, pre-drilling, and probe extrusion and penetration actions of the compaction quality detection module; during the measurement of compaction degree and moisture content, capture the current relative coordinates as the position label of the measuring point, and then combine them into a position-quality label package; integrate the position-quality label package with the roadbed geometric shape feature index, and independently output a continuous detection report of compaction quality and geometric shape of the entire roadbed.

2. The integrated in-situ testing system for multiple indicators of roadbed construction quality as described in claim 1, characterized in that, The robot platform includes a chassis system and a U-shaped support platform, a top-mounted observation base, and a communication and control unit mounted on it. The U-shaped support platform has a groove structure for fixing and installing the compaction quality detection module. The top-mounted observation base is mounted on the U-shaped support platform, and the geometric morphology index detection module is mounted on the top-mounted observation base. The communication and control unit also communicates with the remote control unit.

3. The integrated in-situ testing system for multiple indicators of roadbed construction quality as described in claim 1, characterized in that, The compaction quality detection module includes a liftable bearing frame, and a lateral translation positioning mechanism, a pre-drilling unit, a multi-needle detection unit, a self-resetting split-type bottom multi-hole positioning guide seat, and a compaction degree and moisture content extraction unit installed in the frame. The pre-drilling unit is mounted on the lateral translation positioning mechanism and is used to pre-drill an array of micro-holes in the roadbed. The multi-needle detection unit is mounted in parallel with the pre-drilling component on the lateral translation positioning mechanism and is used to emit electromagnetic pulses into the soil and collect the propagation time and amplitude attenuation of the reflected echo. The self-resetting split-type bottom multi-hole positioning guide seat is installed at the bottom ground end of the bearing frame and is used to provide absolute vertical guidance in the early stage of penetration and automatically avoid obstacles in the late stage of penetration to ensure that the probe penetrates the soil for the entire stroke. The compaction degree and moisture content extraction unit is used to receive the electromagnetic echo signal collected by the multi-needle detection unit and extract the moisture content and compaction degree of the roadbed soil in a non-destructive and real-time manner through signal analysis and physical inversion algorithms.

4. The integrated in-situ testing system for multiple indicators of roadbed construction quality as described in claim 1, characterized in that, The geometric shape index detection module includes a rigid mounting bracket, a point cloud collector, and a point cloud data processing and feature extraction unit. The point cloud collector is fixed on a rigid mounting bracket and is used to emit laser pulses to the road surface when the robot autonomously cruises along the unstructured roadbed, thereby acquiring the original three-dimensional point cloud data in the local coordinate system in real time; the point cloud data processing and feature extraction unit is used to extract the roadbed geometric shape indicators from the original three-dimensional point cloud data.

5. The integrated in-situ testing system for multiple indicators of roadbed construction quality as described in claim 1, characterized in that, In the main control and data fusion center, if the compaction degree or geometric shape of a certain measuring point does not meet the standard, a clear location defect warning will be generated directly in the local engineering coordinate system, providing the construction party with intuitive guidance for targeted compaction or repair.

6. A multi-index integrated in-situ testing method for roadbed construction quality, characterized in that, The integrated in-situ testing system for multiple indicators of roadbed construction quality, based on any one of claims 1-5, includes: On the set cruise route, the robot platform is driven to move and stop when it reaches the preset gridded physical inspection station. Then, the lateral translation, pre-hole drilling and probe extrusion of the compaction quality inspection module are activated in sequence. During the compaction degree and moisture content measurement, the current relative coordinates are captured as the position label of the measuring point, and then combined into a position-quality label package. During the movement of the driving robot platform, high-precision three-dimensional point cloud data of the roadbed surface is acquired synchronously and continuously. Combined with the real-time posture data of the robot platform, the geometric shape feature index of the roadbed is automatically extracted. The position-quality tag package is integrated with the geometric shape feature index of the roadbed, and a continuous inspection report on the compaction quality and geometric shape of the entire roadbed is independently output.

7. The integrated in-situ testing method for multiple indicators of roadbed construction quality as described in claim 6, characterized in that, Before the robot platform moves forward, it also includes: The robot platform is moved to a known reference starting point at the roadbed construction site. The main control and data fusion center completes the self-test of each hardware module and establishes a local engineering coordinate system with this starting point as the origin, and initializes the chassis odometer and laser SLAM mapping algorithm.

8. The integrated in-situ testing method for multiple indicators of roadbed construction quality as described in claim 6, characterized in that, The geometric morphological characteristics of the roadbed include: roadbed smoothness, cross slope, width, longitudinal elevation, and centerline deviation.

9. The integrated in-situ testing method for multiple indicators of roadbed construction quality as described in claim 6, characterized in that, When the probe in the compaction quality testing module is in close contact with the soil without air gaps, the main control and data fusion center sends a command to emit electromagnetic pulses to the probe for data acquisition and feature inversion, including: The electromagnetic pulse received signal was extracted and decoupled. The electromagnetic wave bidirectional propagation time difference Δt, which characterizes the soil moisture polarization response, and the apparent reflection coefficient S, which characterizes the soil porosity and skeleton conductivity, were separated and extracted from the received signal waveform. Based on the soil type pre-input for the current construction section, the system automatically calls up the "dry density-moisture content joint inversion model" for this type of soil, which has been calibrated in the laboratory and is pre-set, and simultaneously retrieves the standard maximum dry density corresponding to this soil type. The electromagnetic wave bidirectional propagation time difference Δt and apparent reflection coefficient S obtained by decoupling are input into the dry density-moisture content joint inversion model to obtain the dry density and mass moisture content of the current detection point. Then, the actual dry density calculated is compared with the standard maximum dry density called to obtain the actual compaction degree. During the probe penetration into the soil to perform the test, the coordinate space of the probe in the soil is recorded. This coordinate is used as a spatial location label and combined with the actual compaction degree and mass moisture content obtained by inversion to form a "location-mass" data package. After the test is completed, the probe is pulled out and reset.

10. The integrated in-situ testing method for multiple indicators of roadbed construction quality as described in claim 6, characterized in that, The system compares the roadbed's geometric features, compaction degree, and moisture content with the pre-imported design values ​​and preset allowable deviation thresholds one by one, automatically determining whether the test indicators of each acceptance station and grid node are qualified, and completing the automated engineering verification of the data.

Images