High-precision topographic surveying and mapping and three-dimensional modeling system based on unmanned aerial vehicle

Abstract

The invention provides a high-precision topographic mapping and three-dimensional modeling system based on an unmanned aerial vehicle, and the system comprises an unmanned aerial vehicle cooperation module which is used for deploying a multi-source sensor and a sensor electronic interface on the unmanned aerial vehicle, and building a distributed network of the unmanned aerial vehicle; the real-time dynamic positioning module is used for acquiring unmanned aerial vehicle positioning data and an initial target terrain map; the autonomous control module is used for generating a flight path, performing dynamic adjustment, constructing an autonomous flight control model and outputting a flight control instruction; the data quality enhancement module is used for performing parameter adjustment on the real-time data and performing space and signal level combined processing on the real-time data so as to enhance the data quality; and the terrain model optimization module is used for processing the laser radar point cloud data so as to construct a three-dimensional terrain model post-processing module, and carrying out data visual display and flight suggestion generation. According to the method, the positioning and modeling precision is greatly improved, efficient surveying and mapping are achieved, the single-point fault risk is reduced, and it is guaranteed that tasks can be completed in a high-safety, high-precision and intelligent mode.

Description

Technical Field

[0001] The present invention relates to the technical field of terrain mapping and modeling, and in particular to a high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle (UAV). Background Art

[0002] Topographic surveying and 3D modeling are important techniques in surveying and mapping, with widespread application in multiple fields. Topographic surveying involves measuring and mapping the Earth's surface morphology and features. Using various surveying instruments and techniques, terrain coordinates, elevation, slope, aspect, and other data are acquired and then mapped into topographic maps to visually display the terrain's undulations and the distribution of features. 3D modeling utilizes collected data to construct three-dimensional spatial models using computer technology. This allows for a more intuitive and three-dimensional representation of terrain and features, providing richer information for subsequent analysis, design, and decision-making.

[0003] The high-precision terrain mapping and 3D modeling system based on drones is a technical system that uses drone-mounted sensors to collect data and processes and analyzes it through professional software to obtain high-precision terrain data and 3D models. However, due to the complexity of surveying and mapping, the current existing technology still has the following shortcomings: 1) Positioning accuracy depends on satellite signals. When drones are positioned in complex environments such as dense buildings, canyons, and forests, the signals are weak or blocked, resulting in large positioning errors, which affects the accuracy of 3D modeling. 2) Point cloud modeling often has high noise and low resolution, and the ability to complete and reconstruct is insufficient. 3) It is greatly affected by weather and environment: Weather conditions such as rain, fog, and strong winds affect flight and sensor collection effects, reducing data quality. 4) Insufficient degree of automation: Traditional processes require a lot of manual intervention, such as route planning, data collection, point cloud filtering, and model generation, which affects efficiency and consistency. Summary of the Invention

[0004] In order to overcome the shortcomings of the existing technology, the purpose of the present invention is to provide a high-precision terrain mapping and three-dimensional modeling system based on drones. Through multi-machine collaboration, the mapping efficiency is greatly improved and the risk of single-point failure is reduced. Through multiple positioning, flight control is strengthened and optimized to ensure that the mission can be completed with high safety, high precision and intelligence. It can also realize the construction of high-resolution three-dimensional models through data quality enhancement and point cloud optimization, thereby promoting the intelligent development of the spatial data industry.

[0005] To achieve the above objectives, the present invention provides the following solution: a high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle, comprising:

[0006] The UAV collaboration module is used to deploy multi-source sensors and sensor electronic interfaces on UAVs and build a distributed network of UAVs through wireless communication for collaborative control of UAVs;

[0007] A real-time dynamic positioning module is used to obtain UAV positioning data using an RTK-GNSS receiver, a ground base station, PPK technology, and an inertial navigation component, and then perform initial three-dimensional modeling of the surveying area based on the UAV positioning data to obtain an initial target terrain map;

[0008] an autonomous control module, configured to generate a flight path based on the initial target terrain map and target environment information, dynamically adjust the flight path to avoid obstacles, and construct an autonomous flight control model based on the flight path to output flight control instructions;

[0009] a data quality enhancement module, configured to adjust parameters of the real-time data acquired by the multi-source sensors according to a designed quality evaluation index system, and to perform joint processing of the real-time data at the spatial and signal levels to enhance data quality;

[0010] A terrain model optimization module is used to optimize, extract features, complete, and reconstruct the resolution of the LiDAR point cloud data, and then use the processed LiDAR point cloud data to construct a three-dimensional terrain model;

[0011] A post-processing module, which is used to visualize the 3D terrain model and flight path, and generate flight recommendations in conjunction with artificial intelligence;

[0012] Among them, the UAV collaboration module, the real-time dynamic positioning module, the autonomous control module, the data quality enhancement module, the terrain model optimization module and the post-processing module are interconnected.

[0013] Optionally, the UAV collaboration module includes:

[0014] A drone configuration unit is used to obtain target mapping requirements, select a drone based on the target mapping requirements, configure multi-source sensors and sensor electronic interfaces on the selected drone, and regularly perform sensor debugging and environmental adaptability testing in the laboratory to optimize sensor power; the multi-source sensors include a GNSS-INS integrated navigation system, a lidar, an RGB camera, a multispectral sensor, and an environmental sensor;

[0015] The control architecture building unit is used to divide the surveying area into multiple sub-areas according to the target surveying requirements, and assign the sub-areas to different drones. Through wireless communication, a distributed network of drones is built, and then the drone status information is obtained through the distributed network to perform drone task scheduling and path planning.

[0016] Optionally, the real-time dynamic positioning module includes:

[0017] A reference positioning unit, configured to be mounted on a UAV with an RTK-GNSS receiver to receive satellite signals and support real-time dynamic differential positioning, and to enable continuous communication between the UAV and the RTK-GNSS receiver and a ground base station;

[0018] Positioning enhancement unit, used to obtain UAV positioning data using RTK-GNSS receiver, ground base station, PPK technology and inertial navigation unit;

[0019] A continuous positioning unit, used to perform short-term autonomous navigation using the inertial navigation unit in areas where satellite signals are weak or blocked, so as to continuously position the UAV;

[0020] A terrain generation unit is used to perform three-dimensional modeling of the surveying area based on the UAV positioning data, obtain an initial terrain model and obstacle distribution data, and obtain an initial target terrain map by combining the initial terrain model and the obstacle distribution data.

[0021] Optionally, the positioning reinforcement unit includes:

[0022] The first positioning enhancement subunit is configured to calculate a satellite signal error correction value using an RTK system through a ground base station, and transmit the satellite signal error correction value to the UAV in real time to perform error correction processing and position information fusion on the UAV GNSS collected data to obtain first positioning data;

[0023] The second positioning enhancement subunit is used to synchronously import the UAV flight records and ground base station data into the post-processing platform, and use the PPK technology to perform post-differential positioning processing on the first positioning data to correct errors and occlusions in the acquisition process to obtain the second positioning data;

[0024] The third positioning reinforcement subunit is used to combine the IMU, accelerometer, gyroscope sensor and the RTK-GNSS receiver to obtain an inertial navigation component, and use the monitoring data of the inertial navigation component to correct the second positioning data to obtain UAV positioning data.

[0025] Optionally, the autonomous control module includes:

[0026] an intelligent route planning unit, configured to obtain weather station data and weather information collected by the UAV to obtain target environment information, generate an initial route based on the initial target terrain map and the target environment information, and then define an objective function based on the initial route to optimize the initial route and obtain a flight path;

[0027] a route self-adjustment unit, configured to utilize the multi-source sensor to identify sudden obstacles and local weather changes within the surveying area, perform dynamic environmental perception of the surveying area, and then dynamically adjust the flight path in real time based on the dynamic environmental perception to avoid obstacles;

[0028] An autonomous control strategy unit is used to train a deep reinforcement learning model using the initial target terrain map, the flight path, and the dynamic environment perception, and introduce a spatiotemporal coupling model during the training process to obtain an autonomous flight control model, and then use the flight control model to output flight control instructions.

[0029] Optionally, the intelligent route planning unit includes:

[0030] The target area subunit is used to obtain weather station data and weather information collected by the UAV to obtain target environment information, and to obtain the target area by combining the initial target terrain map and the target environment information;

[0031] An initial route subunit is configured to combine a digital elevation model and a digital surface model to obtain a spatial analysis model, perform spatial analysis on the target area using the spatial analysis model to obtain terrain relief and a safety height, generate a coverage path for the target area using an A* algorithm based on a grid map, automatically identify obstacles based on the LiDAR point cloud using an obstacle detection model, map the identified obstacles to the initial target terrain map, and then combine the terrain relief and safety height, the coverage path, and the initial target terrain map to obtain an initial route;

[0032] The route optimization subunit is used to define an objective function based on the initial route, and then use a genetic algorithm or a particle swarm optimization algorithm to solve the objective function to optimize the initial route and obtain a flight path; the calculation expression of the objective function is:

[0033] J=α·C dist +β·C energy +γ·C risk

[0034] Among them, J is the optimized flight path, C dist is the total distance of the flight path, C energy To estimate energy consumption, C risk is the risk function, and α, β, and γ are weight coefficients.

[0035] Optionally, the data quality enhancement module includes:

[0036] a quality monitoring unit, configured to obtain real-time data from the multi-source sensors, unify timestamps of the real-time data, and construct a quality evaluation index system; compare the real-time data with the quality evaluation index system to obtain data quality feedback; and adjust parameters of the lidar and the environmental sensor based on the data quality feedback;

[0037] The data fusion and enhancement unit is used to perform joint processing of the lidar point cloud data and optical imaging data in the real-time data at the spatial and signal levels to obtain fused data to enhance data quality. The calculation expression of the fused data is:

[0038] F f =w LiDAR ·F LiDAR +w Opt ·F Opt +ε

[0039] Among them, F f is the fused data, F LiDAR is the laser radar reflection intensity and point cloud spatial characteristics, F Opt is the optical image feature after compensation, w LiDAR 、w Opt are dynamic weights, and ε is the fusion error term.

[0040] Optionally, the terrain model optimization module includes:

[0041] a point cloud optimization unit, configured to identify and remove noise from the lidar point cloud data to obtain optimized point cloud data, and perform feature extraction, point cloud completion, and resolution reconstruction on the optimized point cloud data by combining a conditional generative adversarial network and a convolutional neural network;

[0042] The model optimization unit is used to adjust the initial terrain model using the optimized point cloud data after point cloud completion and resolution reconstruction, and then combine the adjusted initial terrain model with the spatial analysis model to obtain a three-dimensional terrain model.

[0043] Optionally, the point cloud optimization unit includes:

[0044] a point cloud processing subunit, configured to identify and remove noise from the laser radar point cloud data to obtain optimized point cloud data, and perform point cloud local feature extraction and point cloud global feature extraction based on the optimized point cloud data;

[0045] The neural network subunit is used to design a conditional generative adversarial network and a network objective function to complete the optimized point cloud data; the calculation expression of the network objective function is:

[0046]

[0047] in, For the completed and detailed enhanced point cloud data, G is the generator, D is the discriminator, and P real For real high-quality point cloud data, P in is a sparse or missing point cloud, and I is the auxiliary information of multi-source images;

[0048] The resolution reconstruction subunit is used to introduce the perceptual feature difference loss and geometric consistency loss based on the completed optimized point cloud data to obtain a total loss function, and perform super-resolution reconstruction of the optimized point cloud data in combination with the total loss function; the calculation expression of the total loss function is:

[0049] L t =L GAN +λ perc L perc +λ geo L geo

[0050] Among them, L t is the total loss function, L GAN To generate the adversarial loss, L perc To measure the difference loss of perceptual features between the generated point cloud and the real point cloud, L geo is the geometric consistency loss, λ perc ,λ geo are all weight coefficients.

[0051] Optionally, the post-processing module includes:

[0052] The visualization unit is used to develop a user interface based on Web technology to visualize the 3D terrain model and flight path, and provide data query and management functions, as well as data screening and downloading;

[0053] The decision-making suggestion unit is used to perform artificial intelligence-assisted analysis based on the flight path and the three-dimensional terrain model to obtain intelligent decision-making suggestions and early warning support.

[0054] The present invention provides a high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle, which discloses the following technical effects:

[0055] 1. High Positioning Accuracy: 1) Combining RTK-GNSS, ground base stations, PPK (post-processed differential positioning), and inertial navigation unit (IMU) multi-positioning technologies greatly improves positioning accuracy and stability, ensuring accurate spatial reference for terrain modeling and navigation, significantly superior to single positioning solutions. 2) In areas with weak satellite signals or obstructions, the IMU can maintain motion continuity for a short period of time, ensuring stable and continuous positioning of the drone, adapting to complex terrain and obstructed environments. This enables high-precision mapping and greater accuracy throughout the entire route process.

[0056] 2. Intelligent Planning and Efficient Autonomous Flight: 1) Based on initial terrain maps and environmental data, intelligent route optimization is performed to significantly improve flight safety and the ability to efficiently complete missions. 2) Real-time obstacle perception and avoidance are enabled, and flight control strategies are dynamically adjusted to achieve rapid obstacle avoidance. 3) Deep reinforcement learning models and spatiotemporal coupling optimization are introduced to ensure intelligent and safe path execution, achieve autonomous flight, adapt to sudden environmental changes and mission requirements, enhance mission flexibility, and ensure smooth mission execution.

[0057] 3. High data quality: 1) Time synchronization of multi-sensor data is achieved to ensure one-to-one correspondence between data; 2) A quality evaluation indicator system is designed to provide real-time feedback and dynamically adjust parameters to ensure that data collection tasks meet compliance standards; 3) Multi-source data, such as lidar and cameras, are jointly enhanced at the spatiotemporal and signal levels to improve overall data quality, especially in low visibility and rainy and snowy weather conditions, making up for the shortcomings of a single sensor.

[0058] 4. High 3D modeling accuracy: Through point cloud optimization, neural network-driven feature extraction, automatic completion, and resolution super-resolution processing, the quality of point cloud reconstruction can be improved, and high-resolution, complete, and truly reflective 3D models with geographic details can be achieved to meet the needs of sophisticated applications.

[0059] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 A schematic diagram of the system architecture provided by an embodiment of the present invention;

[0062] Figure 2 A schematic diagram of a dynamic positioning process according to an embodiment of the present invention;

[0063] Figure 3 A schematic diagram of the flow of autonomous flight path control provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0064] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0065] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0066] like Figure 1 As shown, the present invention provides a high-precision terrain mapping and three-dimensional modeling system based on drones, including an interconnected drone collaboration module, a real-time dynamic positioning module, an autonomous control module, a data quality enhancement module, a terrain model optimization module and a post-processing module.

[0067] 1. UAV collaboration module

[0068] It is used to deploy multi-source sensors and sensor electronic interfaces on drones and build a distributed network of drones through wireless communication to achieve coordinated control of drones. The drone coordination module includes:

[0069] 1.1 UAV configuration unit

[0070] Obtain the target mapping requirements, including the size of the survey area, terrain complexity, required accuracy, and mission duration. Select a drone based on these requirements. Consider a multi-rotor drone with a high payload capacity, at least enough to carry the combined weight of the LiDAR and multiple sensors, while ensuring a flight time sufficient for the mission.

[0071] Multi-source sensors and sensor electronic interfaces are configured on the selected UAV, and sensor debugging and environmental adaptability tests are regularly performed in the laboratory to optimize sensor power; the multi-source sensors include a GNSS-INS integrated navigation system, lidar, RGB camera, multispectral sensor, and environmental sensor.

[0072] 1.2 Control Architecture Building Unit

[0073] It is used to divide the surveying area into multiple sub-areas according to the target surveying requirements, and allocate the sub-areas to different drones. Through wireless communication, a distributed network of drones is built, and then the drone status information is obtained through the distributed network to perform drone task scheduling and path planning, ensure load balancing, and avoid task overlap or omission.

[0074] Among them, drone status information includes battery level, location, sensor operating status, data quality indicators, etc.

[0075] 2. Real-time dynamic positioning module

[0076] like Figure 2 As shown, it is used to obtain UAV positioning data using an RTK-GNSS receiver, a ground base station, PPK technology and an inertial navigation component, and then perform initial three-dimensional modeling of the surveying area based on the UAV positioning data to obtain an initial target terrain map; the real-time dynamic positioning module includes:

[0077] 2.1 Reference positioning unit

[0078] It is used to configure an RTK-GNSS receiver on the drone to receive satellite signals and support real-time dynamic differential positioning. The RTK-GNSS receiver is then combined with a ground base station to ensure continuous communication between the drone. The ground base station serves as the transmitter of RTK differential data and covers the surveying area, ensuring continuous and stable communication between the base station and the drone.

[0079] 2.2 Positioning reinforcement unit

[0080] Used to obtain UAV positioning data using RTK-GNSS receiver, ground base station, PPK technology and inertial navigation components; the positioning enhancement unit includes:

[0081] 2.2.1 The first positioning reinforcement subunit

[0082] It is used to calculate the satellite signal error correction value through the ground base station and the RTK system, and send the satellite signal error correction value to the drone in real time to perform error correction processing and position information fusion of the drone GNSS collected data to obtain the first positioning data, thereby improving centimeter-level positioning accuracy.

[0083] 2.2.2 Second Positioning Strengthening Subunit

[0084] It is used to synchronously import UAV flight records and ground base station data into the post-processing platform, and use PPK technology to perform post-differential positioning processing on the first positioning data, correct errors and occlusions in the acquisition process, obtain a more accurate and stable position trajectory, and obtain the second positioning data.

[0085] 2.2.3 The third positioning reinforcement subunit

[0086] It is used to combine the IMU, accelerometer, gyroscope sensor and the RTK-GNSS receiver to obtain an inertial navigation component, and use the monitoring data of the inertial navigation component to correct the second positioning data to obtain the UAV positioning data.

[0087] 2.3 Continuous Positioning Unit

[0088] It is used to perform short-term autonomous navigation using the inertial navigation component in areas where satellite signals are weak or blocked, so as to provide continuous positioning and anti-interference capabilities for the UAV, ensuring that the UAV positioning is stable and continuous.

[0089] 2.4 Terrain Generation Unit

[0090] It is used to perform three-dimensional modeling of the surveying area based on the UAV positioning data, obtain an initial terrain model and obstacle distribution data, obstacle distribution data such as buildings, trees, power lines, etc., and combine the initial terrain model and the obstacle distribution data to obtain an initial target terrain map.

[0091] 3. Autonomous control module

[0092] like Figure 3 As shown, it is used to generate a flight path based on the initial target terrain map and target environment information, dynamically adjust the flight path to avoid obstacles, and then build an autonomous flight control model based on the flight path to output flight control instructions; the autonomous control module includes:

[0093] 3.1 Intelligent route planning unit

[0094] The intelligent route planning unit is used to obtain weather information collected by meteorological stations and drones, such as wind speed, wind direction, and rainfall, to obtain target environment information, generate an initial route based on the initial target terrain map and the target environment information, and then define an objective function based on the initial route to optimize the initial route and obtain a flight path. The intelligent route planning unit includes:

[0095] 3.1.1 Target region subunit

[0096] It is used to obtain weather information collected by meteorological stations and drones, such as wind speed, wind direction, rainfall, etc., to obtain target environment information, and combine the initial target terrain map and the target environment information to obtain the target area.

[0097] 3.1.2 Initial route subunit

[0098] It is used to combine the digital elevation model and the digital surface model to obtain a spatial analysis model. Using this spatial analysis model, it performs spatial analysis on the target area to obtain terrain undulations and a safe height. Using the A* algorithm based on a grid map, it generates a coverage path for the target area that meets the mapping point cloud density requirements. Using the obstacle detection model, it automatically identifies obstacles based on the LiDAR point cloud and maps the identified obstacles to the initial target terrain map. The initial route is then derived by combining the terrain undulations and safe height, the coverage path, and the initial target terrain map.

[0099] 3.1.3 Route Optimization Subunit

[0100] It is used to define an objective function based on the initial route, and then use a genetic algorithm or a particle swarm optimization algorithm to solve the objective function to optimize the initial route and obtain a flight path; the calculation expression of the objective function is:

[0101] J=α·C dist +β·C energy +γ·C risk

[0102] Among them, J is the optimized flight path, C dist is the total distance of the flight path, C energy To estimate energy consumption, C risk is the risk function, and α, β, and γ are weight coefficients.

[0103] 3.2 Route self-adjustment unit

[0104] It is used to use the multi-source sensor to identify sudden obstacles and local weather changes in the surveying area, complete dynamic environmental perception of the surveying area, and then based on the dynamic environmental perception, make real-time dynamic adjustments to the flight path to avoid obstacles.

[0105] 3.3 Autonomous Control Strategy Unit

[0106] It is used to train a deep reinforcement learning model, such as one based on the DQN or PPO algorithm, using the initial target terrain map, the flight path, and the dynamic environment perception, and introduce a spatiotemporal coupling model during the training process to obtain an autonomous flight control model, which takes the flight state (such as speed, attitude, position information) and environmental observation as input to output flight control instructions.

[0107] 4. Data quality enhancement module

[0108] The module is used to adjust parameters of the real-time data acquired by the multi-source sensor according to the designed quality evaluation index system, and perform joint processing of the real-time data at the spatial and signal levels to enhance the data quality. The data quality enhancement module includes:

[0109] 4.1 Quality Monitoring Unit

[0110] It is used to obtain real-time data from the multi-source sensors, such as the lidar point cloud and RGB camera image data stream carried by the drone, and use multi-frequency GNSS time synchronization technology to give a unified timestamp to all sensor real-time data to ensure the consistency of the point cloud and image on the timeline.

[0111] A quality evaluation index system is constructed, including point cloud density, image clarity, signal integrity, etc. The real-time data is compared with the quality evaluation index system to obtain data quality feedback. Based on the data quality feedback, the parameters of the lidar and environmental sensors are adjusted to adjust the drone flight parameters, such as flight altitude, speed, sensor exposure, etc., to ensure that the collection mission complies with the standards.

[0112] 4.2 Data Fusion and Enhancement Unit

[0113] It is used to perform joint spatial and signal-level processing on the lidar point cloud data and optical imaging data in the real-time data to obtain fused data to enhance data quality, especially to supplement the shortcomings of a single sensor in low visibility and rainy and snowy weather conditions. The calculation expression of the fused data is:

[0114] F f =w LiDAR ·F LiDAR +w Opt ·F Opt +ε

[0115] Among them, F f is the fused data, F LiDAR is the laser radar reflection intensity and point cloud spatial characteristics, F Opt is the optical image feature after compensation, w LiDAR 、w Opt are dynamic weights, and ε is the fusion error term.

[0116] 5. Terrain model optimization module

[0117] It is used to optimize, extract features, complete and reconstruct the point cloud data of the laser radar, and then use the processed laser radar point cloud data to construct a three-dimensional terrain model; the terrain model optimization module includes:

[0118] 5.1 Point Cloud Optimization Unit

[0119] It is used to identify and remove noise from the laser radar point cloud data to obtain optimized point cloud data, and combine the conditional generative adversarial network and convolutional neural network to perform feature extraction, point cloud completion and resolution reconstruction on the optimized point cloud data; the point cloud optimization unit includes:

[0120] 5.1.1 Point Cloud Processing Subunit

[0121] It is used to identify and remove noise from the lidar point cloud data to obtain optimized point cloud data. Based on the optimized point cloud data, local feature extraction and global feature extraction of the point cloud are performed to generate a dense and high-quality point cloud, providing high-quality input for three-dimensional reconstruction.

[0122] 5.1.2 Neural Network Subunits

[0123] It is used to design a conditional generative adversarial network and a network objective function to complete the optimized point cloud data. The calculation expression of the network objective function is:

[0124]

[0125] in, For the completed and detailed enhanced point cloud data, G is the generator, D is the discriminator, and P real For real high-quality point cloud data, P in is a sparse or missing point cloud, and I is multi-source image auxiliary information.

[0126] Generator G: sparse / missing point cloud and multi-source image auxiliary information, outputting a completed and detail-enhanced point cloud.

[0127] Discriminator D: Determines whether the point cloud data is real and helps the generator improve the output quality.

[0128] 5.1.3 Resolution Reconstruction Subunit

[0129] Based on the completed optimized point cloud data, the perceptual feature difference loss and geometric consistency loss are introduced to obtain a total loss function. Combined with the total loss function, super-resolution reconstruction of the optimized point cloud data is performed to promote detail restoration and geometric rationality. The calculation expression of the total loss function is:

[0130] L t =L GAN +λ perc L perc +λ geo L geo

[0131] Among them, L t is the total loss function, L GAN To generate the adversarial loss, L perc To measure the difference loss of perceptual features between the generated point cloud and the real point cloud, L geo is the geometric consistency loss, λ perc ,λ geo are all weight coefficients.

[0132] 5.2 Model Optimization Unit

[0133] The method is used to adjust the initial terrain model by using the optimized point cloud data after point cloud completion and resolution reconstruction, and then combine the adjusted initial terrain model with the spatial analysis model to obtain a three-dimensional terrain model.

[0134] 6. Post-processing module

[0135] It is used to visualize the 3D terrain model and flight path, and generate flight suggestions in combination with artificial intelligence. The post-processing module includes:

[0136] 6.1 Visualization Unit

[0137] It is used to develop user interfaces based on web technologies for visualizing 3D terrain models and flight paths, and provides data query and management functions, as well as data filtering and downloading. For example, it supports 3D model rotation, scaling, layered display, and multi-view linkage display of point cloud and image data.

[0138] 6.2 Decision-making Recommendation Unit

[0139] It is used to perform artificial intelligence-assisted analysis based on the flight path and the three-dimensional terrain model to obtain intelligent decision-making suggestions and early warning support.

[0140] For example, integrated machine learning analysis can automatically identify trends and potential issues in surveying and mapping data, such as landform evolution and anomaly warnings. AI algorithms can be combined to generate personalized analysis recommendations to assist users in developing response strategies. Predictive models can be designed to simulate future short-term environmental and topographic changes, supporting disaster warnings and planning decisions.

[0141] Finally, the user interface and post-processing modules will be integrated and deployed on the cloud platform and local servers. A comprehensive user manual and online help will be designed, along with regular training and demonstrations to ensure users are proficient in system functions and enhance the user experience.

[0142] Therefore, the present invention provides a high-precision terrain mapping and three-dimensional modeling system based on drones, which greatly improves the mapping efficiency and reduces the risk of single-point failures through multi-machine collaboration. It strengthens and optimizes flight control through multiple positioning, ensuring that the mission can be completed with high safety, high precision and intelligence. It can also achieve the construction of high-resolution three-dimensional models through data quality enhancement and point cloud optimization, thereby promoting the intelligent development of the spatial data industry.

[0143] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0144] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A high-precision terrain mapping and 3D modeling system based on drones, characterized by: include: The UAV collaboration module is used to deploy multi-source sensors and sensor electronic interfaces on UAVs and build a distributed network of UAVs through wireless communication for collaborative control of UAVs; A real-time dynamic positioning module is used to obtain UAV positioning data using an RTK-GNSS receiver, a ground base station, PPK technology, and an inertial navigation component, and then perform initial three-dimensional modeling of the surveying area based on the UAV positioning data to obtain an initial target terrain map; an autonomous control module, configured to generate a flight path based on the initial target terrain map and target environment information, dynamically adjust the flight path to avoid obstacles, and construct an autonomous flight control model based on the flight path to output flight control instructions; a data quality enhancement module, configured to adjust parameters of the real-time data acquired by the multi-source sensors according to a designed quality evaluation index system, and to perform joint processing of the real-time data at the spatial and signal levels to enhance data quality; A terrain model optimization module is used to optimize, extract features, complete, and reconstruct the resolution of the LiDAR point cloud data, and then use the processed LiDAR point cloud data to construct a three-dimensional terrain model; A post-processing module, which is used to visualize the 3D terrain model and flight path, and generate flight recommendations in conjunction with artificial intelligence; Among them, the UAV collaboration module, the real-time dynamic positioning module, the autonomous control module, the data quality enhancement module, the terrain model optimization module and the post-processing module are interconnected.

2. The high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle according to claim 1 is characterized in that: The UAV collaboration module includes: A drone configuration unit is used to obtain target mapping requirements, select a drone based on the target mapping requirements, configure multi-source sensors and sensor electronic interfaces on the selected drone, and regularly perform sensor debugging and environmental adaptability testing in the laboratory to optimize sensor power; the multi-source sensors include a GNSS-INS integrated navigation system, a lidar, an RGB camera, a multispectral sensor, and an environmental sensor; The control architecture building unit is used to divide the surveying area into multiple sub-areas according to the target surveying requirements, and assign the sub-areas to different drones. Through wireless communication, a distributed network of drones is built, and the drone status information is obtained through the distributed network to perform drone task scheduling and path planning.

3. The high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle according to claim 2, characterized in that: The real-time dynamic positioning module includes: A reference positioning unit, configured to be mounted on a UAV with an RTK-GNSS receiver to receive satellite signals and support real-time dynamic differential positioning, and to enable continuous communication between the UAV and the RTK-GNSS receiver and a ground base station; Positioning enhancement unit, used to obtain UAV positioning data using RTK-GNSS receiver, ground base station, PPK technology and inertial navigation unit; A continuous positioning unit, used to perform short-term autonomous navigation using the inertial navigation unit in areas where satellite signals are weak or blocked, so as to continuously position the UAV; A terrain generation unit is used to perform three-dimensional modeling of the surveying area based on the UAV positioning data, obtain an initial terrain model and obstacle distribution data, and obtain an initial target terrain map by combining the initial terrain model and the obstacle distribution data.

4. The high-precision terrain mapping and 3D modeling system based on an unmanned aerial vehicle according to claim 3, characterized in that: The positioning reinforcement unit comprises: The first positioning enhancement subunit is configured to calculate a satellite signal error correction value using an RTK system through a ground base station, and transmit the satellite signal error correction value to the UAV in real time to perform error correction processing and position information fusion on the UAV GNSS collected data to obtain first positioning data; The second positioning enhancement subunit is used to synchronously import the UAV flight records and ground base station data into the post-processing platform, and use the PPK technology to perform post-differential positioning processing on the first positioning data to correct errors and occlusions in the acquisition process to obtain the second positioning data; The third positioning reinforcement subunit is used to combine the IMU, accelerometer, gyroscope sensor and the RTK-GNSS receiver to obtain an inertial navigation component, and use the monitoring data of the inertial navigation component to correct the second positioning data to obtain UAV positioning data.

5. The high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle according to claim 4 is characterized in that: The autonomous control module includes: an intelligent route planning unit, configured to obtain weather station data and weather information collected by the UAV to obtain target environment information, generate an initial route based on the initial target terrain map and the target environment information, and then define an objective function based on the initial route to optimize the initial route and obtain a flight path; a route self-adjustment unit, configured to utilize the multi-source sensor to identify sudden obstacles and local weather changes within the surveying area, perform dynamic environmental perception of the surveying area, and then dynamically adjust the flight path in real time based on the dynamic environmental perception to avoid obstacles; An autonomous control strategy unit is used to train a deep reinforcement learning model using the initial target terrain map, the flight path, and the dynamic environment perception, and introduce a spatiotemporal coupling model during the training process to obtain an autonomous flight control model, and then use the flight control model to output flight control instructions.

6. The high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle according to claim 5, characterized in that: The intelligent route planning unit includes: The target area subunit is used to obtain weather station data and weather information collected by the UAV to obtain target environment information, and to obtain the target area by combining the initial target terrain map and the target environment information; An initial route subunit is configured to combine a digital elevation model and a digital surface model to obtain a spatial analysis model, perform spatial analysis on the target area using the spatial analysis model to obtain terrain relief and a safety height, generate a coverage path for the target area using an A* algorithm based on a grid map, automatically identify obstacles based on the LiDAR point cloud using an obstacle detection model, map the identified obstacles to the initial target terrain map, and then combine the terrain relief and safety height, the coverage path, and the initial target terrain map to obtain an initial route; The route optimization subunit is used to define an objective function based on the initial route, and then use a genetic algorithm or a particle swarm optimization algorithm to solve the objective function to optimize the initial route and obtain a flight path; the calculation expression of the objective function is: J=α·C dist +β·C energy +γ·C risk Among them, J is the optimized flight path, C dist is the total distance of the flight path, C energy To estimate energy consumption, C risk is the risk function, and α, β, and γ are weight coefficients.

7. The high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle according to claim 6, characterized in that: The data quality enhancement module includes: a quality monitoring unit, configured to obtain real-time data from the multi-source sensors, unify timestamps of the real-time data, and construct a quality evaluation index system; compare the real-time data with the quality evaluation index system to obtain data quality feedback; and adjust parameters of the lidar and the environmental sensor based on the data quality feedback; The data fusion and enhancement unit is used to perform joint processing of the lidar point cloud data and optical imaging data in the real-time data at the spatial and signal levels to obtain fused data to enhance data quality. The calculation expression of the fused data is: F f =w LiDAR ·F LiDAR +w Opt ·F Opt +ε Among them, F f is the fused data, F LiDAR is the laser radar reflection intensity and point cloud spatial characteristics, F Opt is the optical image feature after compensation, w LiDAR 、w Opt are dynamic weights, and ε is the fusion error term.

8. The high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle according to claim 7, characterized in that: The terrain model optimization module includes: a point cloud optimization unit, configured to identify and remove noise from the lidar point cloud data to obtain optimized point cloud data, and perform feature extraction, point cloud completion, and resolution reconstruction on the optimized point cloud data by combining a conditional generative adversarial network and a convolutional neural network; The model optimization unit is used to adjust the initial terrain model using the optimized point cloud data after point cloud completion and resolution reconstruction, and then combine the adjusted initial terrain model with the spatial analysis model to obtain a three-dimensional terrain model.

9. The high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle according to claim 8, characterized in that: The point cloud optimization unit includes: a point cloud processing subunit, configured to identify and remove noise from the laser radar point cloud data to obtain optimized point cloud data, and perform point cloud local feature extraction and point cloud global feature extraction based on the optimized point cloud data; The neural network subunit is used to design a conditional generative adversarial network and a network objective function to complete the optimized point cloud data; the calculation expression of the network objective function is: in, For the completed and detailed enhanced point cloud data, G is the generator, D is the discriminator, and P real For real high-quality point cloud data, P in is a sparse or missing point cloud, and I is multi-source image auxiliary information; The resolution reconstruction subunit is used to introduce the perceptual feature difference loss and geometric consistency loss based on the completed optimized point cloud data to obtain a total loss function, and perform super-resolution reconstruction of the optimized point cloud data in combination with the total loss function; the calculation expression of the total loss function is: L t =L GAN +λ perc L perc +λ geo L geo Among them, L t is the total loss function, L GAN To generate the adversarial loss, L perc To measure the difference loss of perceptual features between the generated point cloud and the real point cloud, L geo is the geometric consistency loss, λ perc ,λ geo are all weight coefficients.

10. The high-precision terrain mapping and three-dimensional modeling system based on an unmanned aerial vehicle according to claim 9, characterized in that: The post-processing module includes: The visualization unit is used to develop a user interface based on Web technology to visualize the 3D terrain model and flight path, and provide data query and management functions, as well as data screening and downloading; The decision-making suggestion unit is used to perform artificial intelligence-assisted analysis based on the flight path and the three-dimensional terrain model to obtain intelligent decision-making suggestions and early warning support.

Images