Search

CN-121982242-A - Underground mine live-action three-dimensional modeling system and method based on unmanned plane SLAM and multi-sensor fusion

CN121982242ACN 121982242 ACN121982242 ACN 121982242ACN-121982242-A

Abstract

The invention provides an underground mine live-action three-dimensional modeling system and method based on unmanned aerial vehicle SLAM and multi-sensor fusion, wherein the system comprises a hardware platform and a software platform, the hardware platform takes an unmanned aerial vehicle as a main body, integrates a visual sensor, a laser radar, an IMU (inertial measurement unit), a self-adaptive lighting system, a visual laser acquisition control point device, a top plate control point identification camera and a wireless communication module, the wireless communication module is used for realizing real-time transmission of acquisition data to the software system and remote control instruction transmission among all components, the software platform comprises a front end SLAM module and a modeling and rendering module, the front end SLAM module adopts a visual-laser radar-IMU tight coupling SLAM algorithm, fuses control point data acquired by the top plate control point identification camera, realizes real-time point cloud registration and map construction without absolute coordinate constraint under a GPS environment, and the modeling and rendering module is characterized in that a Mesh and 3DGS parallel generation technology is adopted as a core, and the parallel computing capability of synchronously completing Mesh model generation and 3DGS model training is provided.

Inventors

  • HOU WENDA
  • ZHAO XUAN
  • LUO YE
  • XIONG GUOTAO
  • LI XIAOYU

Assignees

  • 中国十五冶金建设集团有限公司

Dates

Publication Date
20260505
Application Date
20260204

Claims (9)

  1. 1. The underground mine live-action three-dimensional modeling system based on unmanned plane SLAM and multi-sensor fusion is characterized by comprising a hardware platform and a software platform; the hardware platform comprises an unmanned aerial vehicle main body, wherein the unmanned aerial vehicle main body is integrated with a visual sensor, a laser radar, an IMU, a self-adaptive lighting system, a visual laser acquisition control point device, a top plate control point identification camera and a wireless communication module; The system comprises a top plate control point identification camera, a self-adaptive lighting system, a visual laser acquisition control point device and a wireless communication module, wherein the top plate control point identification camera is used for detecting and identifying an identified control point of a roadway top plate, the self-adaptive lighting system is used for supplementing light to a complex illumination environment of the roadway, the visual laser acquisition control point device is used for accurately acquiring top plate control point data, and the wireless communication module is used for realizing remote control instruction transmission of acquisition data which is transmitted back to a software system and between all components in real time; the software platform comprises a front-end SLAM module and a modeling and rendering module; the front-end SLAM module adopts a vision-laser radar-IMU tight coupling SLAM algorithm, fuses control point data acquired by a top plate control point identification camera, and realizes real-time point cloud registration and map construction without absolute coordinate constraint under a GPS environment; the modeling and rendering module is used for generating the Mesh model and the 3DGS model in parallel, and achieving parallel operation of Mesh model generation and 3DGS model training synchronously.
  2. 2. The underground mine live-action three-dimensional modeling system based on unmanned plane SLAM and multi-sensor fusion, which is disclosed in claim 1, is characterized in that the unmanned plane main body supports remote control flight, an obstacle avoidance sensor and an emergency processing module are configured, and the emergency processing module is used for realizing automatic hovering or obstacle avoidance around when collision early warning occurs in the unmanned plane flight process, and autonomous returning to a flying spot when signal interruption occurs.
  3. 3. The underground mine live-action three-dimensional modeling system based on unmanned aerial vehicle SLAM and multi-sensor fusion of claim 1 is characterized in that the unmanned aerial vehicle main body and each integrated sensor component meet IP 65-level dustproof and waterproof protection level, and the working temperature range is-10 ℃ to 60 ℃.
  4. 4. The underground mine live-action three-dimensional modeling system based on unmanned plane SLAM and multi-sensor fusion according to claim 1, wherein the wireless communication module adopts a Mesh ad hoc network and microwave communication hybrid architecture, the data transmission distance is more than or equal to 500m, the transmission rate is more than or equal to 10Mbps in a deep roadway without base station coverage, and real-time return of acquired data and no delay interaction of control instructions are supported.
  5. 5. The underground mine live-action three-dimensional modeling method based on unmanned plane SLAM and multi-sensor fusion is characterized by adopting the system of any one of claims 1-4, and specifically comprising the following steps: S1, establishing an absolute positioning reference, namely accurately measuring absolute coordinates of at least 3 marking control points uniformly arranged on a tunnel roof by a total station, pasting labels, planning an optimal flight path containing obstacle avoidance nodes according to the section size and length of the tunnel, and completing sensor calibration, communication links and unmanned aerial vehicle gesture self-checking; S2, integrated data acquisition, namely integrating light self-adaptive processing and dust interference suppression in the acquisition process according to the operations of taking off, IMU initial attitude calibration, preset control point matching initialization positioning, flying along a route, real-time SLAM mapping, merging roof control point data, roof control point identification, automatic control point acquisition, real-time data return, exception handling and safe landing flow operation of the unmanned aerial vehicle; S3, double model generation and control point constraint splicing, namely, mesh model generation and 3DGS model training are completed in parallel, and multi-section model fusion is realized through a control point constraint splicing method of multi-section acquisition data; And S4, incremental updating, namely based on a local incremental updating mechanism of the underground mine live-action model, comparing the new model with the old model to identify a change area, and only acquiring and updating the model for the increment of the change area.
  6. 6. The underground mine live-action three-dimensional modeling method based on unmanned plane SLAM and multi-sensor fusion according to claim 5 is characterized in that a roof control point in the step S1 is used for establishing an underground mine absolute coordinate system, known absolute coordinates of the control point are added into an SLAM optimization chart as strong constraint, and according to the recognition quality of the control point, the recognition quality takes the recognition success rate of more than or equal to 95% and the single-point positioning error of less than or equal to 5cm as an evaluation standard, constraint weight is dynamically adjusted, so that positioning drift suppression is realized.
  7. 7. The underground mine live-action three-dimensional modeling method based on unmanned plane SLAM and multi-sensor fusion of claim 5 is characterized in that the illumination self-adaptive processing in the step S2 automatically adjusts the parameters of an illumination system and the exposure parameters of a camera according to the intensity of ambient light; The dust interference suppression processing adopts an anti-interference point cloud filtering algorithm and an image defogging algorithm, so that the data quality of laser scanning and image acquisition is ensured.
  8. 8. The underground mine live-action three-dimensional modeling method based on unmanned plane SLAM and multi-sensor fusion of claim 5 is characterized in that the Mesh model generation flow in the step S3 comprises the steps of point cloud denoising, point cloud registration under control point constraint, triangular Mesh generation, texture mapping and model optimization; the absolute coordinate constraint of the top plate control point is fused in the training process of the 3DGS model, so that the model positioning accuracy is ensured to be consistent with the Mesh model; and the multi-section model fusion takes a top plate control point as a reference, and the accumulated error of the multi-section acquisition data is eliminated by a control point constraint splicing method, and the splicing error is less than or equal to 8cm.
  9. 9. The underground mine live-action three-dimensional modeling method based on unmanned plane SLAM and multi-sensor fusion according to claim 5 is characterized in that the core of the local increment updating mechanism in the step S4 is that accurate identification of a mine environment change area is completed through a detection result that the geometric deviation of a new model is more than or equal to 10cm or the volume change of the old model is more than or equal to 1m < 3 >, and the re-flying acquisition and the local updating of the model are only carried out on the change area and the expansion 5m range.

Description

Underground mine live-action three-dimensional modeling system and method based on unmanned plane SLAM and multi-sensor fusion Technical Field The invention relates to the technical field of mine mapping and three-dimensional live-action modeling, in particular to an underground mine live-action three-dimensional modeling system and method based on unmanned plane SLAM and multi-sensor fusion. Background The current underground mine building live-action model mainly depends on handheld SLAM equipment, and when the control point data is collected, the control point of the tunnel top plate is required to be introduced into the bottom plate through a total station, or visible laser is utilized for alignment collection. The method has the defects of low operation efficiency, limited movement range of the handheld device, long data acquisition period, difficulty in ensuring positioning accuracy, easiness in being influenced by manual operation errors, high personnel safety risk, complex underground mine environment, hidden dangers such as collapse, toxic gas and the like, great danger for operators, incapability of adapting to the dynamic production requirements of mines, high cost for model updating, and no GPS signal coverage of underground mines, and further limitation of the application of the traditional positioning modeling technology. Therefore, an underground mine live-action three-dimensional modeling system and method based on unmanned plane SLAM and multi-sensor fusion are provided at present for solving the problems. Disclosure of Invention The invention mainly aims to solve the problems in the prior art and provides an underground mine live-action three-dimensional modeling system and method based on unmanned plane SLAM and multi-sensor fusion. The underground mine live-action three-dimensional modeling system based on unmanned plane SLAM and multi-sensor fusion comprises a hardware platform and a software platform; the hardware platform comprises an unmanned aerial vehicle main body, wherein the unmanned aerial vehicle main body is integrated with a visual sensor, a laser radar, an IMU, a self-adaptive lighting system, a visual laser acquisition control point device, a top plate control point identification camera and a wireless communication module; The system comprises a top plate control point identification camera, a self-adaptive lighting system, a visual laser acquisition control point device and a wireless communication module, wherein the top plate control point identification camera is used for detecting and identifying an identified control point of a roadway top plate, the self-adaptive lighting system is used for supplementing light to a complex illumination environment of the roadway, the visual laser acquisition control point device is used for accurately acquiring top plate control point data, and the wireless communication module is used for realizing remote control instruction transmission of acquisition data which is transmitted back to a software system and between all components in real time; the software platform comprises a front-end SLAM module and a modeling and rendering module; the front-end SLAM module adopts a vision-laser radar-IMU tight coupling SLAM algorithm, fuses control point data acquired by a top plate control point identification camera, and realizes real-time point cloud registration and map construction without absolute coordinate constraint under a GPS environment; The modeling and rendering module is characterized in that the modeling and rendering module is a Mesh and 3DGS dual-model parallel generation technology and has parallel operation capability of synchronously completing Mesh model generation and 3DGS model training. Furthermore, the unmanned aerial vehicle main body supports remote control flight, and is provided with an obstacle avoidance sensor and an emergency processing module, so that real-time return of acquired data can be realized; When collision early warning occurs in the flight process, the vehicle automatically hovers or bypasses the flight obstacle, and when signal interruption occurs, the vehicle automatically returns to the departure point without operators entering a dangerous roadway area. Furthermore, the unmanned plane main body and each integrated sensor component meet IP 65-level dustproof and waterproof protection level, adapt to the high-dust and humid environment characteristics of underground mines, have the working temperature range of-10 ℃ to 60 ℃, and can resist extreme temperature fluctuation in mine roadways. Furthermore, the wireless communication module adopts a Mesh ad hoc network and microwave communication mixed architecture, the data transmission distance is more than or equal to 500m, the transmission rate is more than or equal to 10Mbps in a deep roadway without base station coverage, and real-time data acquisition and control instruction feedback are supported without delay interaction. S1, establishing absolu