Ground penetrating radar equipment based on all-wheel-drive unmanned aerial vehicle and three-dimensional geological map construction method

Abstract

The invention discloses ground penetrating radar equipment based on an all-wheel-drive unmanned aerial vehicle and an underground geological three-dimensional map construction method, and relates to the technical field of three-dimensional map construction. The invention aims to solve the problems of terrain information perception distortion, difficulty in constructing a three-dimensional geologic model which is fused with terrain information and has space authenticity and the like caused by vehicle bumping and inclination in the prior art of integrating a ground penetrating radar into an intelligent unmanned vehicle platform for underground detection. According to the integrated structure design of the all-wheel-drive unmanned aerial vehicle integrated ground penetrating radar, the attitude of the whole unmanned aerial vehicle is kept stable when the unmanned aerial vehicle carries the radar for flight operation, data loss and noise introduction caused by attitude jitter of the unmanned aerial vehicle in the radar acquisition process are reduced, and a data foundation is laid for subsequent accurate three-dimensional geological map construction. According to the multi-sensor fusion method, the absolute precision of RTK, the high resolution of laser ranging and the dynamic response characteristic of an accelerometer are fully utilized, and the centimeter-level absolute position of the unmanned aerial vehicle, the height of the unmanned aerial vehicle to the ground and the absolute altitude of the ground surface below a scanning point are obtained. According to the invention, decoupling control of six degrees of freedom corresponding to the full-drive unmanned aerial vehicle is realized, so that the unmanned aerial vehicle can fly close to the rugged topography, and automatic and full-coverage scanning path flight is carried out with a constant ground clearance and a stable attitude.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional map construction technology, specifically to a ground-penetrating radar equipment based on an all-wheel-drive unmanned aerial vehicle and a method for constructing three-dimensional geological maps. Background Technology

[0002] The research background of 3D mapping of underground geology stems from the urgent need for high-precision and high-efficiency geological exploration in fields such as engineering construction, geological disaster early warning, and resource exploration. Most traditional geophysical methods (surface GPR, resistivity method, seismic reflection / refraction), geochemical exploration, geological mapping, drilling, and pitting all require a large amount of manpower for surface operations, wiring, equipment transportation, and point measurement. This is time-consuming, labor-intensive, and costly in complex or large-scale areas, and the exploration depth is limited. The "Action Plan for the Development of Key Areas of Safety and Emergency Equipment (2023-2025)" clearly states that radar detection equipment needs to be developed to enhance risk perception capabilities for complex underground spaces such as mines and tunnels. Against this backdrop, combining unmanned equipment equipped with ground-penetrating radar for underground geological exploration has become a current research hotspot. Its core objective is to develop a high-efficiency, high-precision, and low-cost unmanned geological exploration solution through automation and intelligent technologies.

[0003] In recent years, the development of ground-penetrating radar (GPR) technology has significantly improved the accuracy and efficiency of underground detection. Fang Sihan and colleagues at Wanxun Technology (Shenzhen) Co., Ltd. have developed a small underground pipeline detection and patrol vehicle, comprising a vehicle body, a navigation module, and a radar detection module. During operation, the navigation module guides the vehicle body along a planned route, while the radar module detects underground pipes and cables, effectively saving manpower and improving detection efficiency. Fang Tingchen and colleagues at Shanghai Construction Group Co., Ltd. have developed a multifunctional concrete quality inspection robot and inspection method. The robot is equipped with GPR, millimeter-wave radar, and a vision module, enabling visualized, intelligent, and unmanned inspection of concrete quality.

[0004] For example, prior art CN118334260A discloses a ground-penetrating radar (GPR) map reconstruction method, apparatus, computer equipment, and storage medium. This method associates real-time acquired GPR data with corresponding location tag data to obtain a low-quality 3D map. The low-quality 3D map is then segmented into multiple B-scan images according to the number of channels in a multi-channel GPR system. Each B-scan image is further segmented into multiple sub-images to be reconstructed according to a preset size. Discrete wavelet transform is used to decompose each sub-image to obtain low-frequency and high-frequency component information. Multiple iterative reconstruction calculations are performed using the low-frequency component information of each sub-image to obtain the reconstructed sub-image. Finally, a high-quality 3D map is reconstructed based on the multiple reconstructed sub-images. This method can achieve accurate 3D map reconstruction even when GPR data is missing. However, this prior art does not address issues such as terrain information perception distortion.

[0005] Currently, integrating ground-penetrating radar (GPR) into unmanned vehicle (UGV) platforms for underground exploration has become a mainstream research direction in this field. This integration aims to achieve efficient and automated detection and imaging of subsurface targets (such as pipelines, cavities, layered structures, and buried objects). However, ground-based UGVs are limited by their physical structure and driving methods. In unstructured terrain (such as scree slopes, gullies, steep slopes, and vegetated areas), there are issues such as distortion of terrain information perception caused by vehicle bumps and tilting, and radar-terrain coupling problems. Therefore, it is difficult to construct a three-dimensional geological model that integrates terrain information and has spatial realism. In addition, the inherent limitations of the obstacle-crossing capabilities of traditional wheeled or tracked UGVs severely restrict their application scope and reliability in complex unstructured scenarios. Summary of the Invention

[0006] The technical problem to be solved by this invention is:

[0007] To address the problems of existing technologies that integrate ground-penetrating radar into intelligent unmanned vehicle platforms for underground exploration, such as distortion of terrain information perception caused by vehicle bumps and tilts, difficulty in constructing a three-dimensional geological model that integrates terrain information and has spatial realism, and inherent limitations in obstacle-crossing capabilities, which seriously restrict its application scope and reliability in complex unstructured scenarios, this invention proposes a ground-penetrating radar equipment and a method for constructing a three-dimensional geological map (underground geological three-dimensional map) based on an all-wheel drive unmanned aerial vehicle.

[0008] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0009] The ground-penetrating radar equipment based on all-wheel drive UAVs includes an all-wheel drive UAV platform, RTK positioning components, laser ranging components, control unit, ground-penetrating radar, and host computer; the all-wheel drive UAV platform includes a computing platform, which is only used to package the received raw ground-penetrating radar data, latitude, longitude, and altitude information.

[0010] The main body of the all-drive unmanned aerial vehicle platform is a multi-rotor structure. Each degree of freedom can be driven individually by a driver to adjust the position and attitude of the main body of the unmanned aerial vehicle. It can also enable the unmanned aerial vehicle to perform spatial translation while maintaining its attitude.

[0011] The all-drive UAV platform is used to carry RTK positioning components, laser ranging components and ground penetrating radar. The all-drive UAV platform provides control torque for the ground penetrating radar to ensure its stability during hovering and flight, enabling the ground penetrating radar to obtain clear, consistent and low-noise underground reflection signals, and complete the accurate sampling of geological data.

[0012] The ground-penetrating radar mounted on the underside of the UAV is a non-destructive device that uses high-frequency electromagnetic waves to detect underground structures and extract geological features beneath the surface. It utilizes the different degrees of reflection of high-frequency electromagnetic waves at the interface of underground media to invert underground structures through time delay, waveform and amplitude transformation, thereby achieving non-destructive underground detection.

[0013] The RTK positioning component, mounted on the all-drive UAV platform, is used to perform high-precision differential calculation positioning in real time by receiving satellite signals and ground reference station data. It continuously acquires centimeter-level latitude, longitude and altitude information of the all-drive UAV platform in real time while in motion, and then transmits it to the computing platform and control unit of the all-drive UAV platform. The computing platform then transmits it back to the host computer.

[0014] The laser ranging component mounted on the lower part of the drone is a TOF laser ranging sensor, which measures the distance between the drone and the ground (ground altitude) by measuring the laser flight time and transmits the data to the control unit, which then transmits it to the computing platform and host computer.

[0015] The control unit includes a flight control board developed based on the STM32 series chip. The flight control board is used to execute the low-level control program to ensure the stable flight of the UAV system. The computing platform of the all-drive UAV platform is used for path planning, collecting and processing data from various sensors, and communicating with the ground station. The control unit directly obtains information from the laser ranging component and the RTK positioning component, performs altitude information fusion positioning, and accurately and stably obtains the latitude, longitude, altitude, and geological information of a certain point on the detected ground to construct a three-dimensional geological map containing terrain information.

[0016] The control unit is used to achieve high-altitude information fusion positioning, while the host computer is used to process ground-penetrating radar data and stitch together 3D maps.

[0017] A method for constructing a three-dimensional geological map, the method being implemented based on the ground-penetrating radar equipment of the all-drive UAV, includes: all-drive UAV terrain-following flight control and three-dimensional geological map construction;

[0018] The all-drive UAV's terrain-following flight control employs a cascade control structure to achieve precise motion management. The body position controller and body angle controller can receive the desired three-axis position and angle commands from the high-level planner, respectively, and output the desired three-axis velocity and angular velocity. Simultaneously, the RTK positioning component calculates the real-time three-axis position and velocity of the body, and the gyroscope and accelerometer in the control unit calculate the real-time three-axis attitude angle and angular velocity around the three axes, thereby realizing: cascade PID control of the position loop and velocity loop, and cascade PID control of the angle loop and angular velocity loop, respectively. Based on the above all-drive UAV terrain-following flight control, the all-drive UAV platform maintains stable attitude and performs translation during terrain-following flight, thereby ensuring that the UAV carrying the ground-penetrating radar can traverse the area to be explored, provided that the radar data is accurately and stably acquired.

[0019] The construction of the three-dimensional geological map includes high-level information fusion positioning, ground-penetrating radar data processing, and three-dimensional map stitching.

[0020] The altitude information fusion positioning is achieved through data fusion based on complementary filtering. By multi-sensor fusion of RTK data, laser ranging data, and IMU data, the relative altitude of the UAV to the ground, the altitude of the UAV, and the altitude of the ground surface are obtained. First, the motion acceleration is obtained by removing the influence of gravity from the accelerometer data. Then, the relative altitude is obtained from the laser ranging and the absolute altitude is obtained from the RTK. The vertical velocity is calculated by differentiation. When the RTK signal is good, the vertical velocity estimates of RTK and laser are weighted and fused based on its quality factor. The fused velocity and the acceleration integral velocity are fused again using complementary filtering. Finally, the fused position and velocity are complementaryly fused to obtain the accurate altitude to the ground. The absolute altitude of the ground surface is calculated by subtracting the relative altitude from the absolute altitude of RTK.

[0021] The ground-penetrating radar (GPR) data processing includes the following steps: The raw GPR waveform is transformed into a "depth-reflection intensity" profile. First, useless time intervals are removed from each trace by 0–100 ns. Then, a 20 ns moving average is used to remove background noise. Next, a 50–500 MHz bandpass filter is used to retain the effective frequency band. Subsequently, exponential gain is used to amplify the weak deep signals. Finally, the time of each sampling point is converted to the actual depth (formula: depth = time × speed of light / 2). The corresponding amplitude is stored as a "depth-intensity" curve, and all traces are processed to form the entire profile.

[0022] The aforementioned 3D map stitching involves piecing together scattered ground-penetrating radar data profiles into a continuous 3D geological map. Specifically, the surface elevation and depth values ​​of each survey line are first converted into absolute elevations, and then the intensity is increased by accumulating data using a uniform voxel grid. Next, adjacent flight strips are aligned using RANSAC-ICP, and overlapping areas are weighted and fused. Then, Kriging interpolation is used to fill the gaps, and finally, a moving cube is used to extract the geological interface, outputting a 3D model that can be directly visualized.

[0023] The present invention has the following beneficial technical effects:

[0024] This invention addresses the gap in the research field of unmanned exploration using UAV-borne ground-penetrating radar. It innovatively proposes a device (a prototype has been manufactured) that utilizes a fully-driven UAV equipped with ground-penetrating radar for geological exploration, and outlines methods for terrain-following control, data fusion, and 3D map construction for the fully-driven UAV. Specifically, the beneficial effects of this invention are:

[0025] 1. This invention proposes an integrated structural design that integrates ground-penetrating radar using an all-drive UAV. Compared with unmanned vehicle-mounted ground-penetrating radar, it has stronger environmental adaptability, overcomes terrain limitations, and facilitates large-scale operations. Compared with traditional UAV-mounted ground-penetrating radar solutions, the all-drive UAV used in this invention can maintain its attitude during flight, and the radar antenna is always parallel to the ground. It is significantly better than traditional UAVs in terms of radar data acquisition consistency and noise level, and has the advantages of good stability and strong anti-interference ability.

[0026] The integrated structural design of the all-wheel drive UAV with ground-penetrating radar proposed in this invention keeps the attitude of the UAV stable during flight operations carrying the radar, reduces data loss and noise introduction caused by UAV attitude jitter during radar acquisition, and makes the acquired raw radar data clearer, more consistent and with a higher signal-to-noise ratio, laying a data foundation for the subsequent construction of accurate three-dimensional geological maps.

[0027] 2. This invention proposes a multi-sensor fusion method that fully utilizes the absolute accuracy of RTK, the high resolution of laser ranging, and the dynamic response characteristics of accelerometers to obtain centimeter-level absolute position of the UAV, the UAV's altitude above the ground, and the absolute elevation of the ground surface below the scanning point.

[0028] 3. The precise terrain-following flight control proposed in this invention adopts a cascaded PID structure to decouple the six degrees of freedom of the fully driven UAV, enabling the UAV to fly close to undulating terrain, maintain a constant ground clearance, and perform automated, full-coverage scanning path flight with stable attitude.

[0029] 4. The standardized processing flow for radar data proposed in this invention includes time window truncation, background removal, bandpass filtering, gain compensation, and time-depth conversion, transforming the original waveform into an interpretable "depth-reflection intensity" profile. The flow is clear, highly operable, and ensures data consistency and comparability.

[0030] 5. This invention proposes a complete three-dimensional geological map construction scheme, including spatial mapping, flight strip stitching, data fusion, spatial interpolation, and interface extraction. It automatically and intelligently integrates discrete, single radar profiles into a continuous, seamless, and realistic three-dimensional geological model that reflects the underground structure, greatly reducing the workload and subjective errors of manual stitching and interpretation.

[0031] 6. The geological map generated by geological exploration based on this invention is not only a traditional underground structure map, but also integrates a topographic-geological integrated three-dimensional model with accurate surface topographic information, which is more in line with the actual geological scene and has higher application value. Attached Figure Description

[0032] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0033] Figure 1 Design configuration diagram of an integrated geological exploration equipment using a fixed tilt-rotor all-wheel drive UAV equipped with ground-penetrating radar (Design configuration diagram of a fixed tilt-rotor all-wheel drive UAV equipped with ground-penetrating radar).

[0034] Figure 2 Design configuration diagram of an integrated geological exploration equipment using a variable tilt rotor all-wheel drive UAV equipped with ground penetrating radar (Design configuration diagram of a variable tilt rotor all-wheel drive UAV equipped with ground penetrating radar).

[0035] Figure 3 This is a block diagram of the terrain-following flight control method for all-drive unmanned aerial vehicles (block diagram of terrain-following control algorithm for all-drive unmanned aerial vehicles).

[0036] Figure 4 This is a rendering of ground-penetrating radar information obtained from a geological profile along the flight path of a UAV, after time-depth conversion. Figure 5 This is a two-dimensional geological profile obtained after converting ground-penetrating radar data.

[0037] Figure 6 This is a schematic diagram of a geological map constructed based on terrain data, using a 3D geological map construction method (a schematic diagram of the effect of data alignment and stitching based on terrain data).

[0038] Figure 7This is a three-dimensional geological map containing topographic information generated for geological exploration using this ground-penetrating radar equipment. Figure 8 A flowchart (illustrative diagram) of the method for constructing 3D maps. Detailed Implementation

[0039] The following will refer to the appendices in the embodiments of the present invention. Figure 1-8 The technical solutions in the embodiments / exemplifications of this invention are clearly and completely described herein. Obviously, the described embodiments / exemplifications are only a part of this invention, and not all of them. Based on the embodiments / exemplifications of this invention, all other embodiments / exemplifications obtained by those skilled in the art without creative effort are within the scope of protection of this invention. It should be noted that, unless otherwise specified, the embodiments / exemplifications and their features in this invention can be combined with each other.

[0040] The ground-penetrating radar equipment based on all-drive UAVs includes: an all-drive UAV platform, RTK positioning components, laser ranging components, a control unit, ground-penetrating radar, and a host computer; (the computing platform is included in the all-drive UAV platform. The computing platform only packages the data from each sensor. Of the three subsequent algorithms, only the first data fusion runs in the control unit, while the rest are transmitted back to the host computer for calculation).

[0041] The structure of the all-wheel-drive UAV platform can be divided into fixed-tilt-rotor all-wheel-drive UAVs and variable-tilt-rotor all-wheel-drive UAVs. Unlike traditional coplanar-configured rotor UAVs, all-wheel-drive UAVs allow each of the six degrees of freedom related to its position and attitude to be driven individually by a actuator. This enables the UAV to perform spatial translation while maintaining a constant attitude and attitude control while maintaining a constant position. Ground-penetrating radar antennas need to be kept as horizontal, stable, and at a constant distance from the ground as possible during operation. The all-wheel-drive design provides stronger control torque, resulting in extremely high stability during hovering and flight. This allows for the acquisition of clear, consistent, and low-noise underground reflection signals, enabling accurate sampling of geological data.

[0042] The RTK positioning component is a positioning technology that receives satellite signals and ground reference station data and performs high-precision differential calculations in real time, which can ensure that the system obtains real-time, continuous, centimeter-level latitude, longitude and altitude information when in motion.

[0043] The laser ranging component is a TOF laser ranging sensor, which measures the distance between the UAV and the ground by measuring the laser flight time, and calculates the terrain information by combining the altitude information obtained by the RTK module.

[0044] The control unit includes a flight control board and a computing platform developed based on the STM32 series chips. The flight control board is used to execute the underlying control program to ensure the stable flight of the UAV system, while the computing platform is used for path planning, collecting and processing data from various sensors, and communicating with the ground station.

[0045] The ground-penetrating radar is a non-destructive device that uses high-frequency electromagnetic waves to detect underground structures and extract geological features beneath the surface.

[0046] Based on the above structure and design, the ground-penetrating radar equipment based on all-wheel drive UAVs can accurately and stably acquire the latitude, longitude, altitude and geological information of a certain point on the ground, and can be used to construct a three-dimensional geological map containing terrain information.

[0047] The method for constructing a three-dimensional geological map based on the above equipment includes the following parts: a method for controlling the terrain-following flight of an all-drive UAV and a method for constructing a three-dimensional geological map.

[0048] The all-wheel-drive UAV's terrain-following flight control employs a cascade control structure for precise motion management. The aircraft position controller and angle controller receive the desired three-axis position and angle commands from the higher-level planner, respectively, and output the desired three-axis velocity and angular velocity. Simultaneously, the system uses an RTK system to calculate the real-time three-axis position and velocity, and utilizes onboard gyroscopes and accelerometers to calculate the real-time three-axis attitude angles and angular velocities around the three axes. This enables cascade PID control of the position and velocity loops, and cascade PID control of the angle and angular velocity loops, respectively. Based on this all-wheel-drive UAV terrain-following flight control method, the system can maintain stable attitude and perform translational motion during terrain-following flight. This ensures that, with accurate and stable acquisition of radar data, the UAV, carrying ground-penetrating radar, can traverse the area to be explored.

[0049] The method for constructing a three-dimensional geological map includes a height information fusion positioning method, a ground-penetrating radar data processing method, and a three-dimensional map stitching method.

[0050] The altitude information fusion positioning method is a data fusion method based on complementary filtering. It uses multi-sensor fusion of RTK data, laser ranging data, and IMU data to obtain the UAV's relative altitude to the ground, the UAV's altitude, and the ground surface altitude. First, the motion acceleration is obtained by removing the influence of gravity from accelerometer data. Then, the relative altitude is obtained from laser ranging and the absolute altitude from RTK, and the vertical velocity is calculated using differentiation. When the RTK signal is good, the vertical velocity estimates from RTK and laser are weighted and fused based on their quality factor. Complementary filtering is then used to fuse the fused velocity with the integral velocity of the acceleration. Finally, the fused position and velocity are complementaryly fused to obtain the accurate altitude above the ground. The absolute ground surface altitude is then calculated by subtracting this relative altitude from the absolute RTK altitude. The specific process is as follows:

[0051] Step 1: Obtain the Z-axis acceleration value from the accelerometer and compensate for gravitational acceleration by determining the local gravitational acceleration.

[0052]

[0053] in, These are the raw Z-axis acceleration data measured by the accelerometer, where g is the local gravitational acceleration. This refers to the Z-axis acceleration after eliminating gravity.

[0054] Step 2: Acquire the ground altitude data measured by TOF and differentiate it to obtain the system Z-axis velocity measured by RTK:

[0055]

[0056] in, Here, dt represents the ground altitude data acquired by the TOF laser ranging system, and dt represents the sampling interval of the TOF laser ranging system. The Z-axis velocity of the system is obtained via TOF.

[0057] Step 3: Obtain the absolute altitude data measured by RTK, and differentiate the altitude data obtained by RTK to obtain the system Z-axis velocity measured by RTK:

[0058]

[0059] in, dt represents the absolute altitude information acquired by RTK, where dt is the sampling interval of the RTK positioning system. The Z-axis velocity of the system is obtained through RTK.

[0060] Step 4: Using the quality factor in the RTK data as a confidence level, the velocities obtained from TOF and RTK are fused.

[0061]

[0062] in, The quality factor representing the precision of RTK data. The velocity is the result of fusing the z-axis velocity measured by RTK and the z-axis velocity measured by TOF.

[0063] Step 5: The fused velocity and Z-axis acceleration are then subjected to complementary fusion filtering to further improve the real-time performance of the data.

[0064]

[0065] in, For speed fusion coefficient, The Z-axis velocity of the system is obtained by fusing three sensors: RTK, TOF, and accelerometer.

[0066] Step Six: Perform complementary fusion filtering on the velocity obtained from the three sensors and the ground altitude measured by TOF to obtain the system ground altitude that balances real-time performance and accuracy.

[0067]

[0068] in For location fusion coefficients, This represents the ground altitude obtained after fusion.

[0069] The pseudocode corresponding to the height information fusion positioning process is shown in the table below:

[0070]

[0071] The ground-penetrating radar data processing method includes the following steps, the purpose of which is to transform the raw ground-penetrating radar waveform into a "depth-reflection intensity" profile. First, useless time intervals are cut off from each trace in the range of 0–100 ns. Then, a 20 ns moving average is used to remove background noise. Next, a 50–500 MHz bandpass filter is used to retain the effective frequency band. Subsequently, exponential gain is used to amplify the weak deep signals. Finally, the time of each sampling point is converted to the actual depth (formula: depth = time × speed of light / 2). The corresponding amplitude is then stored as a "depth-intensity" curve. Once all traces are processed, they are pieced together to form the entire profile. The specific process is as follows:

[0072] Step 1: Traverse the raw dataset detected by the radar, process each trace in a loop, and remove unimportant data segments from the radar records by time window truncation, such as early direct waves and late deep noise, to retain the effective analysis period and improve processing efficiency and data quality.

[0073] Step 2: By using a digital bandpass filter, low-frequency offset and high-frequency random noise are suppressed, while retaining the effective reflected signal near the radar antenna's main frequency, thereby improving the signal-to-noise ratio of the data.

[0074]

[0075] in, This is the low-pass cutoff frequency. This is the Qualcomm cutoff frequency.

[0076] Step 3: Using exponential gain compensation, the filtered signal is multiplied by a gain function that increases exponentially with time. This compensates for the amplitude attenuation of electromagnetic waves during underground propagation due to medium absorption and geometric diffusion effects, thus enhancing the signal from deep, weakly reflecting layers and making it more clearly visible in the cross-sectional view.

[0077]

[0078] in, It is the gain function. It is a gain coefficient (normal number) selected based on the attenuation characteristics of the medium.

[0079] Step 4: Transform the data from the time domain to the depth domain to generate a depth profile with clear physical spatial meaning, facilitating direct geological interpretation.

[0080]

[0081] Where c is the speed of light. Let t be the dielectric constant of the medium, t be time, and z be the depth information obtained from the conversion.

[0082] The pseudocode corresponding to the ground-penetrating radar data processing procedure is shown in the table below:

[0083]

[0084] The method for constructing a three-dimensional geological map involves piecing together scattered ground-penetrating radar data profiles into a continuous three-dimensional geological map. First, the surface elevation and depth values ​​of each survey line are converted into absolute elevations. Then, the intensity is increased by accumulating data using a uniform voxel grid. Next, adjacent flight strips are aligned using RANSAC-ICP, and overlapping areas are weighted and fused. Then, Kriging interpolation is used to fill the gaps. Finally, a moving cube is used to extract the geological interface, and a three-dimensional model that can be directly visualized is output.

[0085] Step 1: Let the radar record set of the k-th flight strip be . Significant feature points are filtered by reflection intensity threshold:

[0086]

[0087] in, For point The intensity of reflection, This is an empirical threshold.

[0088] Step 2: Find the optimal rigid body transformation Where R is the rotation matrix and t is the translation vector, the alignment of the flight strip is achieved by minimizing the distance between matching point pairs:

[0089]

[0090] in, , To match point pairs, outliers that are too far away (distance greater than 0.05 times the voxel resolution) are eliminated iteratively.

[0091] Step 3: Transform the ground-penetrating radar measurement point p from the measurement coordinate system to the world coordinate system, and calculate... The coordinates of the voxel grid in which it is located:

[0092]

[0093] Among the symbols Indicates rounding down. This is known as voxel resolution.

[0094] Step 4: For voxels species included The fused reflection intensity values ​​of the data points are calculated as follows:

[0095]

[0096] Among them, weight Based on points To the voxel center European distance calculate:

[0097]

[0098] Step 5: Use a spherical model to fit the spatial correlation of the data:

[0099]

[0100] Where h is the distance between sample point pairs. denoted as nugget value, representing variability or measurement error at a small scale; c is the sill value, representing the total variability; and a is the range, representing the influence range of spatial autocorrelation.

[0101] Step Six: For the points to be interpolated A set of weighting coefficients can be solved using the following equation:

[0102]

[0103] in, These are Lagrange multipliers used to constrain the sum of weights to 1. It is a point and The semivariogram values ​​between them. The weights are obtained. Then, the interpolation point The estimated value at the location and its estimated covariance The calculation is as follows:

[0104]

[0105] Step 7: For each voxel, calculate the scalar values ​​of its eight corner points. With a given isosurface threshold The difference:

[0106] The coordinates of the intersection points of the isosurface and the voxel edges are calculated using linear interpolation. For connecting vertices... (value is) ) and (value is) The edges of ) and their intersection points The calculation formula is:

[0107]

[0108] The pseudocode corresponding to the 3D map stitching process is shown in the table below:

[0109]

[0110] Example:

[0111] Combined with appendix Figures 1 to 7 This invention describes the structural composition of a ground-penetrating radar equipment based on an all-wheel-drive unmanned aerial vehicle (UAV) and the related data fusion, processing, and map construction methods. Figure 1 , Figure 2 The figures shown are design configuration diagrams of ground-penetrating radar equipment based on all-wheel drive unmanned aerial vehicles with fixed and variable tilt rotors. Figure 3 This is a schematic diagram of the terrain-following flight control algorithm for an all-wheel-drive unmanned aerial vehicle (UAV). Figure 4 This is a rendering of ground-penetrating radar information obtained from a geological profile along the flight path of a UAV, after time-depth conversion. Figure 5 This is a two-dimensional geological profile obtained after converting ground-penetrating radar data. Figure 6 This is a schematic diagram illustrating the effect of data alignment and stitching based on terrain data. Figure 7This is a three-dimensional geological map containing topographic information generated for geological exploration using this ground-penetrating radar equipment. Figure 8 A schematic diagram illustrating a method for constructing a 3D map.

[0112] Figure 1 The first one is a fixed tilt six-rotor fully driven UAV platform. Its six rotors with fixed tilt angles can provide tiltable thrust vectors for the UAV to achieve decoupled control of the UAV's attitude.

[0113] Figure 2 The first one is a variable tilt quadcopter fully driven UAV platform. Each pair of rotors on opposite sides is driven by a servo motor to enable the UAV to perform translational movements while maintaining a stable attitude.

[0114] Figure 1 , 2 The component in section 2 is a laser rangefinder mounted on the underside of the drone. It calculates the drone's distance to the ground by emitting laser pulses and measuring the round-trip time difference. The core formula is: Distance = (Speed ​​of light × Time) / 2.

[0115] Figure 1 , 2 The third type is a ground-penetrating radar mounted on the underside of the UAV. It utilizes the different degrees of reflection of high-frequency electromagnetic waves at the interface of underground media to invert underground structures through time delay, waveform and amplitude transformation, thereby achieving non-destructive underground detection.

[0116] Reference Figure 1 , 3 -7. Detailed explanation of this embodiment: The method for three-dimensional reconstruction of underground geology described in this embodiment includes the following steps:

[0117] Step 1: Use a fully-driven UAV to traverse the area to be explored and acquire ground-penetrating radar data, RTK data, TOF laser ranging data, and IMU data.

[0118] Step 2: Differentiate the acquired RTK and TOF height information to obtain the Z-axis velocity, and use the RTK quality factor as the RTK velocity weight to perform velocity fusion, obtaining the velocity after RTK and TOF fusion:

[0119]

[0120] Step 3: The gravity-free Z-axis accelerometer data is fused with the velocity obtained by fusing RTK and TOF data using complementary filtering to obtain the Z-axis velocity fused from the three sensors.

[0121]

[0122] Where K1 is the speed fusion coefficient and dt is the running cycle of the fusion algorithm.

[0123] Step 4: Perform complementary fusion filtering on the velocity obtained from the three sensors and the ground altitude data acquired from the TOF sensor to obtain ground altitude data that combines high static accuracy and fast dynamic response, i.e., ground elevation.

[0124]

[0125] Where K2 is the location fusion coefficient.

[0126] Step 5: Traverse each track obtained by ground penetrating radar, filter the obtained tracks using methods such as time windowing and bandpass filtering, compensate for the energy attenuation of electromagnetic waves propagating underground using exponential gain, and finally convert the time axis of the data into the depth axis to generate a list of depth and amplitude for each track.

[0127] Step 6: Read the raw observation records of the ground penetrating radar, including the surface elevation, plane coordinates and depth-amplitude list, and partition the data according to the flight strip markings and perform voxel addressing.

[0128] Step 7: Convert the reflection amplitude acquired by ground penetrating radar from the time-depth domain to a unified three-dimensional Cartesian grid, and complete the spatial discretization and indexed storage of the original data.

[0129] Step 8: Using significant reflectors as geometric feature primitives, estimate the optimal rigid body transformation parameters between adjacent flight strips to achieve spatial consistency correction between flight strips. For overlapping voxels, an inverse distance weighting strategy is introduced to perform a weighted average of the reflection intensity contributed by multiple flight strips, thereby suppressing sampling bias and improving the signal-to-noise ratio in a statistical sense.

[0130] Step 9: Perform a 3D interpolation method within the bounding box, model spatial correlation through the variogram function, achieve unbiased optimal prediction, and upgrade the discrete sampling to a continuous random field.

[0131] Step 10: Apply the moving cube algorithm to perform topology-preserving isosurface tracing on the interpolated volume data to generate geological interfaces in the form of triangular meshes, thus explicitly representing the three-dimensional geometry of strata or structural surfaces, such as... Figure 4 and 5 As shown.

[0132] Verification has shown that the method proposed in this invention solves the technical problem raised in this invention. Simulation experiments and practical applications have both verified the technical effects and practicality claimed in this invention.

[0133] The three-dimensional geological map construction method (algorithm) proposed in this invention is the underlying technical core of this invention, and various products can be derived based on this algorithm. A three-dimensional geological map construction system is developed using a programming language based on the algorithm (method) proposed in this invention. This system has program modules corresponding to the steps of the three-dimensional geological map construction method, and executes the steps of the aforementioned three-dimensional geological map construction method during runtime. The computer program of the developed system (software) is stored on a computer-readable storage medium, and the computer program is configured to implement the steps of the aforementioned three-dimensional geological map construction method when called by a processor. In other words, this invention is materialized on a carrier, becoming a computer program product. It is used to achieve accurate construction of three-dimensional geological maps using ground-penetrating radar equipment based on all-wheel-drive unmanned aerial vehicles.

[0134] Various implementations of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, application-specific integrated circuits (ASICs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0135] A computational program (also referred to as a program, software, software application, or code) includes machine instructions for a programmable processor and can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or apparatus (e.g., disk, optical disk, memory, programmable logic device PLD) for providing machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0136] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this application can be achieved, they are all within the protection scope of this invention.

Claims

1. A ground penetrating radar equipment based on an all-wheel unmanned aerial vehicle, characterized in that, It comprises a full-drive unmanned aerial vehicle platform, an RTK positioning component, a laser ranging component, a control unit, a ground penetrating radar and an upper computer; the full-drive unmanned aerial vehicle platform comprises a computing platform, and the computing platform is only used for packaging received ground penetrating radar original data, longitude and latitude, and altitude information; The unmanned aerial vehicle body of the full-drive unmanned aerial vehicle platform is a multi-rotor structure, each degree of freedom can be driven by a driver to adjust the position and attitude of the unmanned aerial vehicle body, and the unmanned aerial vehicle can also move horizontally in space while keeping the attitude unchanged; The full-drive unmanned aerial vehicle platform is used for carrying the RTK positioning component, the laser ranging component and the ground penetrating radar, the full-drive unmanned aerial vehicle platform controls the moment of the ground penetrating radar to make the ground penetrating radar stable in hovering and flying, so that the ground penetrating radar obtains clear, consistent and low-noise underground reflection signals and completes accurate sampling of geological data; The ground penetrating radar assembled on the lower part of the unmanned aerial vehicle is a non-destructive device for detecting underground structures and extracting geological features under the ground surface by using high-frequency electromagnetic waves, and the device realizes non-destructive detection of the underground by using different reflection degrees of high-frequency electromagnetic waves on the interface of underground media, time delay, waveform and amplitude inversion of underground structures; The RTK positioning component assembled on the full-drive unmanned aerial vehicle platform is used for receiving satellite signals and ground reference station data, performing real-time high-precision differential calculation positioning, continuously acquiring centimeter-level longitude and latitude and altitude information of the full-drive unmanned aerial vehicle platform in a moving state, and then transmitting the information to the computing platform and the control unit of the full-drive unmanned aerial vehicle platform, and the computing platform transmits the information back to the upper computer; The laser ranging component assembled on the lower part of the unmanned aerial vehicle is a TOF laser ranging sensor, which measures the distance between the unmanned aerial vehicle and the ground (height above ground) by measuring the flight time of laser and transmits the distance to the control unit, and then transmits the distance to the computing platform and the upper computer through the control unit; The control unit comprises a flight control board developed based on an STM32 series chip, and the flight control board is used for executing a bottom layer control program to ensure stable flight of the unmanned aerial vehicle system; the computing platform of the full-drive unmanned aerial vehicle platform is used for path planning, collecting and processing sensor data and communicating with the ground station; the control unit directly acquires information from the laser ranging component and the RTK positioning component, performs height information fusion positioning, and accurately and stably acquires longitude, latitude, altitude and geological information of a detected point on the ground to construct a three-dimensional geological map containing terrain information; The control unit is used for realizing height information fusion positioning, and the upper computer is used for realizing ground penetrating radar data processing and three-dimensional map splicing.

2. The all-wheel drive drone-based ground penetrating radar equipment of claim 1, wherein, The full-drive unmanned aerial vehicle platform is a fixed tilt-rotor type full-drive unmanned aerial vehicle or a variable tilt-rotor type full-drive unmanned aerial vehicle.

3. A method of constructing a three-dimensional geological map, characterized by, The method is realized based on the ground penetrating radar equipment of the full-drive unmanned aerial vehicle in claim 1 or 2, and comprises full-drive unmanned aerial vehicle terrain following flight control and three-dimensional geological map construction. The full-drive unmanned aerial vehicle ground simulation flight control adopts a cascade control structure to achieve precise motion management. The body position controller and the body angle controller can respectively receive the three-axis expected position and three-axis expected angle instructions of the body from the high-level planner output, and output the three-axis expected speed and three-axis expected angular velocity. Meanwhile, the real-time position and speed of the three axes of the body are measured by the RTK positioning component, and the real-time attitude angle and angular velocity of the three axes of the body are measured by the gyroscope and accelerometer in the control unit, so as to respectively realize the cascade PID control of the position loop and the speed loop and the cascade PID control of the angle loop and the angular velocity loop. Based on the above full-drive unmanned aerial vehicle ground simulation flight control, the full-drive unmanned aerial vehicle platform keeps stable in the process of ground simulation flight, so as to ensure that the unmanned aerial vehicle carrying the ground penetrating radar traverses the flight area to be explored under the premise that the radar data is accurately and stably obtained. The three-dimensional geological map construction includes height information fusion positioning, ground penetrating radar data processing and three-dimensional map splicing. The height information fusion positioning is realized based on data fusion of complementary filtering. The relative height of the unmanned aerial vehicle to the ground, the altitude of the unmanned aerial vehicle and the ground surface altitude are obtained through multi-sensor fusion of RTK data, laser ranging data and IMU data. First, the motion acceleration is obtained by removing the gravity influence through the accelerometer data, then the relative height is obtained from the laser ranging and the absolute altitude is obtained from the RTK, and the vertical velocity is calculated by differentiation. When the RTK signal is good, the vertical velocity estimates of RTK and laser are fused based on the quality factor weighting, the fusion velocity is integrated with the acceleration to obtain the integrated velocity, and then the integrated velocity is fused again by using complementary filtering. Finally, the accurate height to the ground is obtained by complementary fusion of the fused position and velocity, and the absolute altitude information of the ground surface is calculated by subtracting the relative height from the absolute altitude obtained by RTK. The ground penetrating radar data processing includes the following steps: converting the ground penetrating radar original waveform into a "depth-reflectivity intensity" profile, cutting off the useless time period of each trace line by 0-100 ns, removing background noise by 20 ns sliding average, then reserving the effective frequency band by 50-500 MHz band pass filtering, subsequently amplifying the weak signal in the deep part by exponential gain; finally, converting the time of each sampling point into actual depth (formula: depth = time × light speed / 2 ), and storing the corresponding amplitude as a "depth-intensity" curve, and all the traces are processed to form a complete profile. The three-dimensional map splicing is to splice the scattered ground penetrating radar data profiles into a continuous three-dimensional geological map. Specifically, the ground surface elevation and depth value of each survey line are first converted into absolute elevation, and the intensity is accumulated according to the unified voxel grid. Then, the adjacent flight strips are aligned by using RANSAC-ICP, the overlapping areas are fused by weighting, and then the gaps are filled by using Kriging interpolation. Finally, the geological interface is extracted by using the moving cube, and the three-dimensional model which can be directly visualized is output.

4. The three-dimensional geological map construction method according to claim 3, wherein the height information fusion positioning comprises: In the construction of three-dimensional geological maps Step one: obtaining the Z-axis acceleration value from the accelerometer on the control unit, and compensating the gravity acceleration by determining the local gravity acceleration: The specific implementation process is as follows: Step two: obtaining the height to the ground data measured by TOF, and obtaining the system Z-axis velocity measured by RTK by differentiation: wherein, is the Z-axis acceleration raw data measured by the accelerometer, g is the local gravity acceleration, is the Z-axis acceleration after eliminating gravity; Step three: obtaining the absolute altitude data measured by RTK, and obtaining the system Z-axis velocity measured by RTK by differentiation of the altitude data obtained by RTK: wherein, is the TOF acquired height above ground data, dt is the TOF laser ranging system sampling interval, is the system Z-axis velocity obtained by TOF; Step four: fusing the velocities obtained from TOF and RTK respectively by using the quality factor in the RTK data as the confidence level: wherein, is the absolute altitude information obtained for the RTK, dt is the sampling interval of the RTK positioning system, is the system Z-axis velocity obtained by the RTK; Step five: complementarily fusing the fused velocities with the Z-axis acceleration for further improving the real-time performance of the data: wherein, is a quality factor representing the accuracy of the RTK data, is the fused velocity of the z-axis velocity measured by the RTK and the z-axis velocity measured by the TOF. ​ wherein, is a speed fusion coefficient, is a system Z-axis speed obtained by fusing RTK, TOF, and accelerometer three sensors. Step six: The fused speed of three sensors and the height above ground measured by TOF are complementarily fused to obtain the system height above ground which takes into account real-time and accuracy: wherein is a position fusion coefficient, is the height above the ground after fusion.

5. The three-dimensional geological map construction method according to claim 3 or 4, characterized in that, The specific implementation process of ground penetrating radar data processing in three-dimensional geological map construction is as follows: Step one: traverse the original data set detected by the radar, loop process each trace, remove the unimportant data segment (such as early direct wave and late deep noise) in the radar record through time window interception, and retain the effective analysis period, so as to improve the processing efficiency and data quality; Step two: through a digital band-pass filter, low-frequency offset and high-frequency random noise are suppressed, and effective reflection signals near the main frequency of the radar antenna are retained, so as to improve the signal-to-noise ratio of the data: wherein is a low pass cutoff frequency, is a high pass cutoff frequency; Step three: through exponential gain compensation, the filtered signal is multiplied by an exponential gain function that increases with time, to compensate for the amplitude attenuation caused by medium absorption and geometric diffusion effect during the underground propagation of electromagnetic waves, so that the signal of deep weak reflection layer is enhanced, and it is more clear and visible on the profile: wherein, is a gain function, is a gain factor (normal number) selected according to the attenuation characteristics of the medium; Step four: convert the data from time domain to depth domain to generate depth profile with clear physical space meaning, which is convenient for direct geological interpretation: where c is the speed of light, is the dielectric constant of the medium, t is time, and z is the depth information obtained from the conversion.

6. The three-dimensional geological map construction method according to claim 5, characterized in that, The specific implementation process of three-dimensional map splicing in three-dimensional geological map construction is as follows: Step one: Set the radar record set of the kth strip as Significant feature points are screened by the reflection intensity threshold: wherein is the intensity of the reflection of the point , is an empirical threshold value; Step two: finding the optimal rigid body transformation where R is the rotation matrix and t is the translation vector. The alignment of the strips is achieved by minimizing the distance between the pairs of matching points: wherein, , For matching point pairs, outliers with a distance greater than 0.05 times the voxel resolution are iteratively removed. Step three: convert the ground penetrating radar measurement point p from the measurement coordinate system to the world coordinate system and calculate the voxel grid coordinates in which it is located: where the symbol represents the floor function, i.e. the voxel resolution; Step four: For each voxel A method of determining a reflectance intensity value for a data point comprising The fused reflectance intensity value for a data point is calculated as follows: Among them, weight Based on points To the voxel center European distance The calculation is as follows: Step five: use a spherical model to fit the spatial correlation of the data: where h is the distance between pairs of sample points, is the nugget value, representing the variability or measurement error at a very small scale, is the base value, representing the total variability; is the range, representing the influence zone of spatial autocorrelation; Step six: For the point to be interpolated a set of weight coefficients is solved by the following equation: where, is the Lagrange multiplier to constrain the sum of weights to be 1, is the semi-variogram value between points and to obtain the weight After that, the estimated value at the point to be interpolated and its estimated covariance are calculated as follows: Step seven: For each voxel, calculate the scalar value of its 8 corners Difference from the given isosurface threshold : The intersection point coordinates of the isosurface and the voxel edge are calculated by linear interpolation. For the edge connecting the vertices (value ) and (value ), the calculation formula of the intersection point is as follows: 。 7. A three-dimensional geological map building system characterized by comprising: The system has program modules corresponding to the steps of any one of claims 3-6, and when running, it executes the steps of the three-dimensional geological map construction method.

8. A computer-readable storage medium, characterized in that: The computer readable storage medium stores a computer program, which is configured to be called by the processor to realize the steps of the three-dimensional geological map construction method of any one of claims 3-6.

Images