A spiral amphibious mapping system for mine water structure detection

Abstract

This invention discloses a spiral amphibious mapping system for detecting hydraulic structures in mines, belonging to the field of autonomous mobile robots and 3D reconstruction technology. It includes a spiral amphibious robot platform and a remote monitoring platform. The robot platform comprises a movement and drive module, an environmental perception module, and a control and processing module. The environmental perception module simultaneously collects surface point cloud data, underwater point cloud data, and instantaneous attitude data. The control and processing module executes an attitude-anchor point dual-constraint iterative nearest-point algorithm, fusing physical constraints generated from instantaneous attitude data with geometric anchor point constraints extracted from the point cloud to obtain accurate rigid body transformation parameters. The remote monitoring platform receives these parameters and point cloud data, constructs and displays an integrated land-water 3D map. This invention achieves high-precision cross-media registration and coordinate unification of surface and underwater point clouds, solving the problems of fragmented land-water data and difficulty in locating underwater faults, significantly improving the accuracy and efficiency of detecting hydraulic structures in mines.

Description

Technical Field

[0001] This invention belongs to the field of environmental perception and three-dimensional reconstruction technology, and in particular relates to a spiral amphibious mapping system for detecting mine hydraulic structures. Background Technology

[0002] Currently, defect detection in the land-water interface area of ​​mine hydraulic structures mainly relies on divers or remotely operated underwater vehicles (ROVs) for underwater exploration. Existing exploration technologies typically employ devices equipped with inertial measurement units (IMUs) for localization, and combine this with the Iterative Closest Point (ICP) algorithm to register and fuse surface laser point clouds with underwater sonar point clouds in order to construct a complete 3D model. Meanwhile, some amphibious robot platforms also exist on the market, often employing a hybrid drive system combining wheeled / tracked structures with propeller propulsion to adapt to different land and water environments.

[0003] However, existing technologies still face the following technical challenges in practical applications: First, in environments with continuous water flow disturbance, underwater detection vehicles are prone to swaying, leading to severe drift in IMU-integrated positioning methods. Simultaneously, water swaying increases the registration failure rate of ICP algorithms that rely solely on point cloud geometric features, making it difficult to achieve high-precision fusion of surface and underwater point cloud data. Second, because underwater sonar scanning data and surface lidar point cloud data belong to different coordinate systems and are difficult to accurately register, the data is fragmented, preventing the formation of a continuous, integrated 3D model. This makes it difficult to accurately locate underwater detected defects within a coordinate system referenced by these structures. Third, existing composite amphibious robot platforms have complex structures, slow mode-switching responses, and poor maneuverability in loose media such as mud, sand, and gravel on dam slopes, making it difficult to provide a stable mobile platform for high-precision sensing tasks. Summary of the Invention

[0004] This invention aims to overcome the shortcomings of existing technologies and provide a spiral amphibious mapping robot system. Through an innovative heterogeneous sensor fusion algorithm and a reliable dedicated hardware platform, it solves the problem of high-precision registration of point cloud data above and below water in turbulent water environments, and improves the robot's passability and stability in complex water and land terrain. Ultimately, it achieves rapid location, accurate navigation and integrated visualization of underwater fault points.

[0005] To achieve the above objectives, the present invention provides a spiral amphibious mapping system for detecting hydraulic structures in mines, comprising: Spiral amphibious robot platform and remote monitoring platform; The spiral amphibious robot platform includes a mobility and drive module, an environmental perception module, and a control and processing module. The mobility and drive module is used to carry the various modules of the system and drive the robot to move in an amphibious environment. The environmental perception module is used to simultaneously collect surface point cloud data of above-water structures, underwater point cloud data of underwater structures, and instantaneous attitude data of the robot. The control processing module is used to receive the surface point cloud data, underwater point cloud data and instantaneous attitude data, and execute the attitude-anchor point dual-constraint iterative nearest point algorithm. By fusing the physical constraints generated from the instantaneous attitude data and the environmental geometric anchor point constraints extracted from the point cloud data, the rigid body transformation parameters for unifying the surface and underwater coordinate systems are calculated. The remote monitoring platform is communicatively connected to the spiral amphibious robot platform and is used to receive the rigid body transformation parameters and point cloud data to construct and display an integrated three-dimensional map of land and water.

[0006] Preferably, the moving and driving module includes two independent driving units arranged symmetrically on the left and right, each driving unit being a hollow, sealed coaxial double helix; The surface of the double helix is ​​provided with continuous helical blades, and a waterproof motor and reducer are integrated inside; The double helix generates propulsion on land by meshing its blades with the ground, and in water by using its blades as a propeller to generate thrust, thus seamlessly switching between land and water propulsion modes.

[0007] Preferably, the environmental perception module includes: The waterborne three-dimensional perception unit includes a three-dimensional lidar installed on the top of the robot cabin, used to acquire waterborne point cloud data of the waterborne structure; The underwater 3D sensing unit includes a multibeam imaging sonar installed at the bottom of the robot's cabin, used to acquire underwater point cloud data of underwater structures; The attitude sensing unit includes an inertial measurement unit rigidly mounted inside the robot cabin near the center of gravity, used to provide instantaneous attitude data of the carrier.

[0008] Preferably, the control processing module executes the attitude-anchor point dual-constraint iterative nearest point algorithm, including: The attitude constraint generation unit is used to perform spectral analysis on the instantaneous attitude data, estimate the swaying period, and perform integral averaging within the swaying period to filter out swaying noise, thereby calculating the instantaneous attitude rotation matrix used to construct physical constraints. The geometric anchor point extraction and matching unit is used to extract line segment features from the surface point cloud data and the underwater point cloud data respectively as geometric anchor points, and to establish corresponding sets of surface and underwater anchor point pairs through feature matching. The dual-constraint optimization solution unit is used to construct and solve the objective function of the fused standard point cloud registration term, the physical constraint term consisting of the instantaneous attitude rotation matrix, and the geometric constraint term consisting of the anchor point pair set, to obtain the rigid body transformation parameters.

[0009] Preferably, the optimization process of the dual-constraint optimization unit in constructing the objective function simultaneously considers three constraints: aligning the surface point cloud and the underwater point cloud in space, approximating the instantaneous attitude rotation matrix with the solved rotation transformation, and ensuring that the surface anchor point and the corresponding underwater anchor point are precisely coincident after transformation.

[0010] Preferably, the geometric anchor point extraction and matching unit is further used to identify linear structures from the point cloud through local principal component analysis, extract the center point and direction vector of the linear structure as line segment features, and perform anchor point matching based on the positional similarity of the center point and the directional similarity of the direction vector.

[0011] Preferably, the control processing module is further configured to, after obtaining the rigid body transformation parameters, map the position of the fault point detected underwater to a unified three-dimensional coordinate system based on the surface point cloud through coordinate mapping, obtain the surface coordinates of the fault point, and send the fault point and its surface coordinates to the remote monitoring platform.

[0012] Preferably, the remote monitoring platform includes: The 3D map reconstruction module is used to generate an integrated 3D map of land and water in a unified coordinate system based on the received rigid body transformation parameters, surface point cloud data and underwater point cloud data. The path planning and navigation module is used to plan a navigation path and generate corresponding navigation instructions in the integrated land and water 3D map based on the set starting point and the underwater target point located by the system.

[0013] Preferably, the remote monitoring platform further includes: The visualization annotation module is used to fuse and render the above-water point cloud, underwater point cloud, and fault points identified or annotated by the system into the same interactive 3D scene for display.

[0014] Preferably, the spiral amphibious robot platform further includes a waterproof main cabin, and the control processing module and power system are integrated into the waterproof main cabin; The movement and drive module is connected to the waterproof main body via a support frame with shock absorbers.

[0015] Compared with the prior art, the present invention has the following advantages and technical effects: This invention utilizes an attitude-anchor point dual-constraint iterative nearest-point algorithm executed by the control processing module. This algorithm integrates physical constraints generated from instantaneous attitude data with geometric anchor point constraints composed of line segment features, effectively resisting carrier swaying errors caused by water flow disturbances and significantly improving the cross-medium registration accuracy of surface laser point clouds and underwater sonar point clouds. Secondly, the coordinate mapping established based on the precise rigid body transformation parameters can accurately transform underwater detected fault points into a unified coordinate system based on the surface point cloud, fundamentally solving the problem of difficult underwater fault point location caused by the separation of land and water data. Finally, this invention employs a double helix as the sole movement and drive module, achieving seamless switching between land and water propulsion modes. Its helical blades have excellent maneuverability in loose media, providing a stable and reliable movement platform for the environmental perception module and ensuring the quality of data acquisition. Attached Figure Description

[0016] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of the structure of the spiral amphibious robot according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the overall system workflow of an embodiment of the present invention. Figure 3 This is a schematic diagram of the attitude-anchor point dual-constraint iterative nearest point (DC-ICP) algorithm according to an embodiment of the present invention. Detailed Implementation

[0017] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0018] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0019] like Figure 2 As shown, this embodiment provides a spiral amphibious mapping system for detecting mine hydraulic structures. The system is built around three core functions: rapid positioning, precise navigation, and visual annotation. It includes: Spiral amphibious robot platform and remote monitoring platform; The spiral amphibious robot platform includes a mobility and drive module, an environmental perception module, and a control and processing module. The mobility and drive module is used to carry the various modules of the system and drive the robot to move in amphibious environments; The environmental perception module is used to simultaneously collect surface point cloud data of above-water structures, underwater point cloud data of underwater structures, and instantaneous attitude data of the robot. The control and processing module is used to receive surface point cloud data, underwater point cloud data and instantaneous attitude data, and execute the attitude-anchor point dual-constraint iterative nearest point algorithm. By fusing the physical constraints generated from the instantaneous attitude data and the environmental geometric anchor point constraints extracted from the point cloud data, the rigid body transformation parameters for unifying the surface and underwater coordinate systems are calculated. The remote monitoring platform communicates with the spiral amphibious robot platform to receive rigid body transformation parameters and point cloud data, and to construct and display an integrated 3D map of land and water.

[0020] The system in this embodiment consists of a robot platform equipped with heterogeneous sensors and a remote monitoring platform, aiming to achieve integrated high-precision detection of structures in complex aquatic and terrestrial environments. The robot uses a double-helix propulsion system, possessing both land-based and underwater navigation capabilities, with a compact structure and strong terrain adaptability. It integrates a lidar, multi-beam imaging sonar, and an inertial measurement unit (IMU). The attitude-anchor point dual-constraint iterative nearest point (DC-ICP) algorithm combines the instantaneous physical attitude constraints provided by the IMU with the constraints of stable geometric anchor points (preferably line segment features) extracted from the environmental point cloud, jointly guiding and correcting the registration process of the surface and underwater point clouds. This effectively suppresses the carrier swaying error caused by water flow disturbance, achieving high-precision cross-media 3D model fusion.

[0021] Based on this, the system in this embodiment implements three core functions: 1) Rapid positioning: Through precise cross-medium coordinate mapping, the defect points detected underwater are transformed into a known coordinate system on the water in real time, realizing rapid positioning based on the reference objects on the water; 2) Precise navigation: Plan the optimal route on a unified 3D map and generate instructions to guide divers or ROVs to accurately reach the underwater target point; 3) Visual annotation: The surface laser point cloud, underwater sonar point cloud and the annotated fault points are merged and rendered into a unified 3D scene to provide a panoramic and interactive display of the detection results.

[0022] This embodiment solves the problems of difficult underwater structure detection and positioning and the separation of land and water data in mining water conservancy projects, and significantly improves the accuracy, efficiency and safety of detection operations.

[0023] Furthermore, the movement and drive module includes two independent drive units arranged symmetrically on the left and right, each drive unit being a hollow, sealed coaxial double helix; The surface of the double helix is ​​equipped with continuous helical blades, and the interior integrates a waterproof motor and reducer; The double helix generates propulsion on land by meshing its blades with the ground, and in water by using its blades as a propeller to generate thrust, seamlessly switching between land and water propulsion modes.

[0024] Furthermore, the spiral amphibious robot platform described in this embodiment is the core of the system's mobile data acquisition, adopting an integrated spiral propulsion and modular structure, optimized for water-land transition environments.

[0025] The mobility and drive module, the core component of the helical amphibious robot platform, employs a symmetrical coaxial double helix as its sole propulsion mechanism. Each helix is ​​a hollow, sealed structure with continuous, high-lead helical blades on its surface, and integrates a waterproof motor and planetary reducer internally. This design allows the robot to move on land by engaging the helical blades with the ground to generate propulsion and obstacle-crossing capabilities; and to move in water by acting as propellers to generate thrust, achieving a seamless and unobtrusive transition between land and water modes.

[0026] Furthermore, the environmental perception module includes: The waterborne three-dimensional perception unit includes a three-dimensional lidar installed on the top of the robot cabin, used to acquire waterborne point cloud data of the waterborne structure; The underwater 3D sensing unit includes a multibeam imaging sonar installed at the bottom of the robot's cabin, used to acquire underwater point cloud data of underwater structures; The attitude sensing unit includes an inertial measurement unit rigidly mounted inside the robot cabin near the center of gravity, used to provide instantaneous attitude data of the carrier.

[0027] Furthermore, the environmental perception module involved in this embodiment is responsible for the synchronous acquisition of multi-source data, including: Waterborne 3D Sensing Unit: Employs a 3D lidar, installed on the top of the cabin, to acquire point cloud Q of the waterborne structure.

[0028] Underwater 3D sensing unit: Employs a multibeam imaging sonar, installed at the bottom of the hull, to acquire underwater structural point cloud P.

[0029] Attitude sensing unit: Employing a high-precision six-axis IMU, rigidly mounted inside the cabin near the center of gravity, it provides raw instantaneous attitude data of the carrier. Its data purpose has been redefined, primarily for generating physical constraints rather than motion integrals.

[0030] Furthermore, the control processing module executes the attitude-anchor point dual-constraint iterative nearest-point algorithm, including: The attitude constraint generation unit is used to perform spectral analysis on instantaneous attitude data, estimate the sway period, and perform integral averaging within the sway period to filter out sway noise, thereby solving for the instantaneous attitude rotation matrix used to construct physical constraints. The geometric anchor point extraction and matching unit is used to extract line segment features from surface point cloud data and underwater point cloud data respectively as geometric anchor points, and to establish corresponding sets of surface and underwater anchor point pairs through feature matching; The dual-constraint optimization solution unit is used to construct and solve the objective function of the fused standard point cloud registration term, the physical constraint term consisting of the instantaneous attitude rotation matrix, and the geometric constraint term consisting of the anchor point pair set, to obtain the rigid body transformation parameters.

[0031] Furthermore, the optimization process of constructing the objective function by the dual-constraint optimization solution unit simultaneously considers three constraints: aligning the surface point cloud and the underwater point cloud in space, approximating the instantaneous attitude rotation matrix with the solved rotation transformation, and ensuring that the surface anchor point and the corresponding underwater anchor point are precisely coincident after the transformation.

[0032] Furthermore, the geometric anchor point extraction and matching unit is also used to identify linear structures from the point cloud through local principal component analysis, extract the center point and direction vector of the linear structure as line segment features, and perform anchor point matching based on the positional similarity of the center point and the directional similarity of the direction vector.

[0033] Furthermore, after obtaining the rigid body transformation parameters, the control processing module is also used to map the location of the underwater detected fault point to a unified three-dimensional coordinate system based on the surface point cloud through coordinate mapping, thereby obtaining the surface coordinates of the fault point, and sending the fault point and its surface coordinates to the remote monitoring platform.

[0034] Furthermore, the attitude-anchor point dual-constraint iterative nearest point (DC-ICP) algorithm described in this embodiment is key to achieving high-precision cross-medium registration. Its core lies in fusing physical attitude constraints with environmental geometric anchor point constraints. Specifically, it includes: (1) Improved attitude calculation for sway suppression: For water sway, the sway period is estimated by spectral analysis of IMU accelerometer data, and the gravity vector gest is robustly estimated by integral averaging over the complete period. Then, the roll angle ϕ and pitch angle θ are calculated, and the rotation matrix Rmeas is constructed as a constraint. This process avoids time integration of attitude data.

[0035] (2) Automatic detection and matching of geometric anchor points: Stable line segment features are extracted from the point cloud as anchor points. Linear structures are identified through local principal component analysis (PCA), and their center point ck and direction vector uk are extracted. This operation is performed on the surface point cloud Q and the underwater point cloud P respectively, and matching is performed by combining position and direction similarity to form a set of reliable anchor point pairs. .

[0036] (3) Double-constraint ICP optimization: Construct the following objective function and solve for the optimal rigid body transformation (R,t): The first term is the standard ICP data term, the second term is the attitude physical constraint term, and the third term is the geometric anchor point constraint term; R,t are the rotation matrix and translation vector to be solved. , These are the corresponding points in the underwater and surface point clouds, respectively. The rotation matrix is ​​calculated from the data of the six-axis accelerometer. , These are environmental anchor point pairs matched from surface and underwater point clouds. and For regularization weights.

[0037] The following steps are used to obtain the following: the measurements of the six-axis accelerometer are averaged over at least one water sloshing cycle to estimate the direction of the gravity vector, and then the roll and pitch angles are calculated; the yaw angle is calculated by combining the angular velocity data with adaptive weights. It is used only as a constraint in optimization problems and is not used for time integral prediction of robot motion.

[0038] Environmental anchor point pairs are obtained as follows: Objects with significant line segment geometric features are first detected from the surface point cloud, and their line segment center points c_k and direction vectors u_k are extracted as anchor points; similar feature detection and matching are performed in the underwater point cloud to form anchor point pairs. , During the optimization process, an optional direction alignment term can be added. To further constrain rotation.

[0039] (4) Cross-medium coordinate mapping: using the obtained high-precision transformation Establish mapping relationships. Underwater fault points. It is possible Transform to a unified coordinate system on water to achieve rapid positioning.

[0040] In this embodiment, after completing point cloud registration and obtaining the transformation relationship (R,t), the control processing module performs coordinate mapping. The fault points detected underwater The location is mapped to a unified 3D map coordinate system based on the above point cloud, realizing the location of underwater faults based on the above-water structure.

[0041] Furthermore, this embodiment also involves a control and support module, including a main controller (such as a high-performance embedded computing platform), a communication unit (wired / wireless and underwater acoustic communication), and a high-energy-density power supply system, all integrated within the waterproof main hull. The main platform is connected to the propeller via a support frame with shock absorbers to isolate it from land impacts.

[0042] Furthermore, the remote monitoring platform includes: The 3D map reconstruction module is used to generate an integrated 3D map of land and water in a unified coordinate system based on the received rigid body transformation parameters, surface point cloud data and underwater point cloud data. The path planning and navigation module is used to plan navigation paths and generate corresponding navigation instructions in a three-dimensional map integrating land and water, based on a set starting point and underwater target points located by the system.

[0043] Furthermore, the remote monitoring platform also includes: The visualization annotation module is used to fuse and render surface point clouds, underwater point clouds, and fault points identified or annotated by the system into a single interactive 3D scene for display.

[0044] Furthermore, the spiral amphibious robot platform also includes a waterproof main cabin, with the control and processing module and power system integrated within the waterproof main cabin; The movement and drive module is connected to the waterproof main body via a support frame with shock absorbers.

[0045] In this embodiment, the remote monitoring platform is connected to the robot to receive data and registration results collected by the robot, and to perform 3D map reconstruction, automatic analysis of structural defects, and generation of navigation guidance information.

[0046] The system in this embodiment is based on the aforementioned platform and algorithm, and the system implements three main functions: 1. Rapid positioning: After obtaining accurate cross-medium coordinate transformation through the DC-ICP algorithm, the system can map any underwater detected defect point to a unified three-dimensional map based on the surface laser point cloud in real time, and quickly describe the underwater fault location based on known surface structures (such as dam tops and gates).

[0047] 2. Precise Navigation: Using a unified 3D map, the optimal safe route is planned with the surface entry point and the located underwater fault point as the starting and ending points. This route is translated into intuitive navigation instructions (such as "forward X meters, descend Y meters") and sent to the diver's navigation terminal or ROV to guide them precisely to the target.

[0048] 3. Visual Annotation: The remote host computer system merges and renders the registered surface laser point cloud, underwater sonar point cloud, and annotated fault points into a single interactive 3D scene. Users can perform rotation, zoom, and sectioning operations to panoramically examine the overall appearance of the land and water structures and the details of potential hazards.

[0049] As a preferred implementation method, the spiral amphibious mapping system for mine hydraulic structure detection in this embodiment includes: Helical amphibious robot hardware platform; like Figure 1 As shown, the spiral amphibious robot serves as the system's data acquisition carrier, and its hardware design is closely aligned with the requirements for stable movement and high-quality perception in complex aquatic and terrestrial environments. This includes: Motion and Drive Module: The core of this module is a symmetrical double helix. The helices are made of PVC through a single injection molding process, with each helix completely sealed internally. They integrate a brushless DC motor and a planetary gear reducer, with the output shaft directly driving the helices' rotation. On land, the motor torque is converted into thrust against the ground through the blades; in water, the rotating blades propel the water, generating forward momentum. The left and right helices are driven independently, achieving steering through differential speed. The main cabin is mounted above the helices via an aluminum alloy frame with rubber damping shock absorbers, effectively filtering vibrations and shocks from land bumps onto the sensors inside the cabin.

[0050] Environmental perception module: The three-dimensional sensing system on the water uses a 16-line lidar, which is mounted on the center of the top cover of the cabin via a rigid bracket. It has a horizontal field of view of 360°, a vertical field of view of 30°, and a maximum range of 100 meters.

[0051] The underwater 3D sensing system uses a high-frequency multibeam imaging sonar, which is installed in a dedicated watertight tank at the bottom of the cabin. The transducer surface is flush with the bottom of the cabin to ensure unobstructed sound wave transmission. The maximum detection range is 50 meters and the beam opening angle is 130°×20°.

[0052] Attitude perception employs an industrial-grade six-axis IMU, containing a three-axis accelerometer and a three-axis gyroscope, with an accelerometer zero-bias stability of less than 0.01mg. This IMU is rigidly fixed to a metal base plate beneath the main circuit board inside the cabin by bolts; this location is calculated to be close to the robot's overall center of gravity.

[0053] A waterproof camera is also installed at the front of the cabin to assist in visual inspection.

[0054] Control and Communication Module: The main controller, an NVIDIA Jetson AGXXavier, is deployed within the waterproof main cabin and is responsible for sensor data synchronization, robot motion control, local algorithm computation, and communication scheduling. Above water, it communicates with the base station via a WiFi 6 module; underwater, it uses a low-power underwater acoustic communication module for emergency command transmission and reception. Power is supplied by a 24V / 20Ah lithium battery pack.

[0055] Furthermore, the data processing and function implementation process includes: The S1 robot performs inspections along a preset path, simultaneously collecting surface laser point cloud Q, underwater sonar point cloud P, six-axis accelerometer data, and hazard sensor data.

[0056] S2 Fast Positioning: Runs the DC-ICP algorithm, such as Figure 3 As shown. First, line segment features are extracted from Q and P and matched to obtain the anchor point pair set S. Simultaneously, the instantaneous attitude matrix is ​​calculated based on the improved formula. Subsequently, with Given S as constraints, optimize the transformation from Q to P. , When underwater sonar or cameras detect suspected crack points. At that time, utilize ( , The coordinates of the object on the water map are obtained through coordinate mapping. This enables rapid positioning.

[0057] S3 Precise Navigation Generation: Host Computer Receives The final coordinates. The path planner uses the water surface starting point and... Using the starting and ending points as references, and considering obstacles (identified from the point cloud), a safe and efficient 3D navigation path is generated. This path is then translated into a series of waypoint commands and sent to the diver's navigation terminal or ROV.

[0058] S4 Visual Annotation Presentation: The host computer displays the registered surface and underwater point clouds, as well as fault points labeled with their location and type. The data is then integrated and rendered into a unified 3D reality model. Users can view it interactively and obtain precise geographical information about the fault.

[0059] More specifically, such as Figure 3 As shown, the specific steps of the DC-ICP algorithm are as follows: Input: surface point cloud Q, underwater point cloud P, accelerometer data.

[0060] Preprocessing and feature extraction: Q and P are downsampled and denoised. In Q, features satisfying the following conditions are extracted through local PCA analysis. / Line segments with a length greater than 0.8m and a value less than 0.1 are considered as candidate anchor points. , A similar method is used to extract line segment features in P.

[0061] Anchor point matching: Calculate the positional and directional similarity between candidate anchor point pairs, and select the optimal matching pair to form a set S.

[0062] Attitude constraint generation: FFT analysis is performed on accelerometer data to estimate the water sloshing period. In the cycle The internal integral of the acceleration vector is calculated using the aforementioned improved formula. .

[0063] Dual-constraint optimization solution: Initialize transformation (R,t), iterative execution: a) Find the nearest neighbor in P for each point in Q; b) Construct an overall objective function that includes ICP error, attitude constraint error, and anchor point constraint error; c) Use the Gauss-Newton method to solve for the incremental update of (R,t) until convergence.

[0064] Output: Optimal Transformation ( , ).

[0065] Furthermore, regarding the application of cross-media coordinate mapping, suppose there is a known gate corner point in the 3D map of the water. .pass This allows for the prediction of its corresponding underwater location, which can be used to guide diving. Conversely, when the sonar is in its underwater position... A suspected cavity was detected. The projection point of the fault can be marked on the surface of the dam on the water map, realizing "water-based positioning underwater".

[0066] This embodiment achieves the following effects: High-precision cross-medium registration and positioning: By fusing instantaneous attitude physical constraints and environmental geometric anchor point constraints, the DC-ICP algorithm effectively resists registration errors caused by water flow sloshing, and achieves high-precision fusion and coordinate unification of water and land point clouds.

[0067] Robust anchor point strategy and sway suppression attitude calculation: Prioritize the use of rotation-sensitive line segment features as anchor points to improve matching reliability; Improved attitude calculation method filters out high-frequency sway noise through periodic integration, providing more reliable instantaneous attitude constraints.

[0068] An innovative amphibious mobile platform: The dual-helix propulsion design enables seamless switching between water and land travel modes. Its simple and reliable structure provides strong mobility in loose media, offering sensors a more stable mobile carrier.

[0069] Complete engineering functional closed loop: It integrates the entire process from data acquisition, fusion mapping, fault location, path planning, navigation guidance to visualization display, forming a complete detection solution that greatly improves the operational efficiency, accuracy and safety of underwater detection of mining and water conservancy structures.

[0070] This embodiment employs a helical propulsion system equipped with lidar, sonar, and a six-axis accelerometer. It combines instantaneous attitude constraints provided by the six-axis accelerometer with stable geometric anchor point constraints extracted from the environment using an attitude-anchor point dual-constraint iterative nearest-point algorithm. This jointly guides the registration process of surface and underwater point clouds, effectively resisting registration errors caused by water flow swaying. Simultaneously, it proposes prioritizing the use of line segment features in the scene as anchor points and improves the attitude calculation method under swaying environments. Finally, through high-precision cross-medium coordinate mapping, it achieves rapid and accurate localization and visualization of underwater fault points in the above-mentioned 3D map. This embodiment solves the problem of inaccurate 3D reconstruction and fault localization in complex water and land environments, significantly improving detection efficiency and accuracy.

[0071] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A spiral amphibious mapping system for detecting hydraulic structures in mines, characterized in that, include: Spiral amphibious robot platform and remote monitoring platform; The spiral amphibious robot platform includes a mobility and drive module, an environmental perception module, and a control and processing module. The mobility and drive module is used to carry the various modules of the system and drive the robot to move in an amphibious environment. The environmental perception module is used to simultaneously collect surface point cloud data of above-water structures, underwater point cloud data of underwater structures, and instantaneous attitude data of the robot. The control processing module is used to receive the surface point cloud data, underwater point cloud data and instantaneous attitude data, and execute the attitude-anchor point dual-constraint iterative nearest point algorithm. By fusing the physical constraints generated from the instantaneous attitude data and the environmental geometric anchor point constraints extracted from the point cloud data, the rigid body transformation parameters for unifying the surface and underwater coordinate systems are calculated. The remote monitoring platform is communicatively connected to the spiral amphibious robot platform and is used to receive the rigid body transformation parameters and point cloud data to construct and display an integrated three-dimensional map of land and water.

2. The system according to claim 1, characterized in that, The moving and driving module includes two independent driving units arranged symmetrically on the left and right, each driving unit being a hollow, sealed coaxial double helix. The surface of the double helix is ​​provided with continuous helical blades, and a waterproof motor and reducer are integrated inside; The double helix generates propulsion on land by meshing its blades with the ground, and in water by using its blades as a propeller to generate thrust, thus seamlessly switching between land and water propulsion modes.

3. The system according to claim 1, characterized in that, The environment sensing module includes: The waterborne three-dimensional perception unit includes a three-dimensional lidar installed on the top of the robot cabin, used to acquire waterborne point cloud data of the waterborne structure; The underwater 3D sensing unit includes a multibeam imaging sonar installed at the bottom of the robot's cabin, used to acquire underwater point cloud data of underwater structures; The attitude sensing unit includes an inertial measurement unit rigidly mounted inside the robot cabin near the center of gravity, used to provide instantaneous attitude data of the carrier.

4. The system according to claim 1, characterized in that, The control processing module executes the attitude-anchor point dual-constraint iterative nearest point algorithm, including: The attitude constraint generation unit is used to perform spectral analysis on the instantaneous attitude data, estimate the swaying period, and perform integral averaging within the swaying period to filter out swaying noise, thereby calculating the instantaneous attitude rotation matrix used to construct physical constraints. The geometric anchor point extraction and matching unit is used to extract line segment features from the surface point cloud data and the underwater point cloud data respectively as geometric anchor points, and to establish corresponding sets of surface and underwater anchor point pairs through feature matching. The dual-constraint optimization solution unit is used to construct and solve the objective function of the fused standard point cloud registration term, the physical constraint term consisting of the instantaneous attitude rotation matrix, and the geometric constraint term consisting of the anchor point pair set, to obtain the rigid body transformation parameters.

5. The system according to claim 4, characterized in that, The dual-constraint optimization solution unit considers three constraints in the optimization process of constructing the objective function: aligning the surface point cloud and the underwater point cloud in space, making the solved rotation transformation approximate the instantaneous attitude rotation matrix, and ensuring that the surface anchor point and the corresponding underwater anchor point are precisely coincident after transformation.

6. The system according to claim 4, characterized in that, The geometric anchor point extraction and matching unit is also used to identify linear structures from the point cloud through local principal component analysis, extract the center point and direction vector of the linear structure as line segment features, and perform anchor point matching based on the positional similarity of the center point and the directional similarity of the direction vector.

7. The system according to claim 1, characterized in that, The control processing module is also used to, after obtaining the rigid body transformation parameters, map the position of the underwater detected fault point to a unified three-dimensional coordinate system based on the surface point cloud through coordinate mapping, obtain the surface coordinates of the fault point, and send the fault point and its surface coordinates to the remote monitoring platform.

8. The system according to claim 1, characterized in that, The remote monitoring platform includes: The 3D map reconstruction module is used to generate an integrated 3D map of land and water in a unified coordinate system based on the received rigid body transformation parameters, surface point cloud data and underwater point cloud data. The path planning and navigation module is used to plan a navigation path and generate corresponding navigation instructions in the integrated land and water 3D map based on the set starting point and the underwater target point located by the system.

9. The system according to claim 8, characterized in that, The remote monitoring platform also includes: The visualization annotation module is used to fuse and render the above-water point cloud, underwater point cloud, and fault points identified or annotated by the system into the same interactive 3D scene for display.

10. The system according to claim 8, characterized in that, The spiral amphibious robot platform also includes a waterproof main cabin, and the control processing module and power system are integrated into the waterproof main cabin. The movement and drive module is connected to the waterproof main body via a support frame with shock absorbers.

Images