Curve surveying and mapping device for engineering surveying

Abstract

The invention belongs to the technical field of surveying and mapping, and discloses a curve surveying and mapping device for engineering surveying. A laser radar, a visual camera and an inertial measurement unit are integrated, a multi-modal fusion measurement system is constructed, and the surveying and mapping precision is improved through a data fusion algorithm. The device is provided with a self-adaptive dynamic measurement range adjustment algorithm, and the measurement range can be intelligently adjusted according to the environment and curve complexity. Noise suppression and data restoration are realized by using deep learning, an intelligent path planning and automatic surveying and mapping algorithm is also provided, a path can be automatically generated, and parameters can be adjusted in real time. Meanwhile, a real-time online high-precision curve fitting algorithm completes curve fitting during data acquisition. The device is strong in environmental adaptability, high in precision, high in efficiency, capable of remarkably reducing labor intensity and suitable for various engineering curve surveying and mapping scenes.

Description

Technical Field

[0001] The present invention belongs to the field of surveying and mapping technology, and specifically relates to a curve surveying and mapping device for engineering surveying. Background Art

[0002] In modern engineering construction, curve surveying and mapping is an extremely important task. Whether it is the construction of roads, bridges, tunnels, or in the fields of water conservancy projects, urban planning, etc., accurately obtaining the shape and position information of curves is crucial for ensuring project quality and improving construction efficiency. Traditional curve surveying and mapping methods, such as total station surveying, GPS surveying, etc., although can meet engineering requirements to a certain extent, still expose many limitations in practical applications.

[0003] Total station surveying requires line of sight between the survey station and the observation point. For areas with complex terrain and poor line of sight conditions, the surveying work will be greatly hindered, and the surveying process is relatively cumbersome, requiring a large amount of manpower and time. Although GPS surveying has the advantages of fast positioning speed and high accuracy, in areas with severe signal occlusion, such as urban canyons, dense forests, etc., the signal is easily interfered, resulting in a decrease in surveying accuracy or even inability to work properly.

[0004] In addition, existing curve surveying and mapping devices also have deficiencies in data processing and algorithms. On the one hand, the fitting accuracy for complex curves is not high enough to accurately reflect the true shape of the curve; on the other hand, during the data acquisition process, due to the influence of measurement errors, environmental noise and other factors, the accuracy and reliability of the data need to be further improved. With the continuous expansion of the scale of engineering construction and the increasing requirements for surveying accuracy, there is an urgent practical need to develop a curve surveying and mapping device that can overcome the drawbacks of traditional surveying methods and has high accuracy and high reliability. Summary of the Invention

[0005] The present invention provides a curve surveying and mapping device for engineering surveying, including a multi-modal fusion measurement system. The multi-modal fusion measurement system integrates a lidar, a vision camera and an inertial measurement unit (IMU); the lidar is used to quickly obtain the three-dimensional point cloud data of the target object and measure the contour and distance information of the curve; the vision camera is used to capture details such as the texture and feature points of the curve by using image recognition technology; the inertial measurement unit is used to monitor the attitude and motion state of the device in real time and compensate for the measurement errors caused by the movement or vibration of the device; the multi-modal fusion measurement system also includes a data processing unit, and the data processing unit adopts a data fusion algorithm based on Kalman filtering to fuse the data collected by the lidar, the vision camera and the inertial measurement unit.

[0006] Furthermore, the device has an adaptive dynamic measurement range adjustment function, which is achieved in the following way: before measurement, the device evaluates the complexity of the measurement environment and the target curve through a preliminary scan of the laser radar and image analysis of the visual camera, and sets the initial measurement range according to the evaluation results; during the measurement process, the device continuously monitors the changes in the measurement data, and if the data within the current measurement range changes smoothly and has no obvious curve features, the measurement range is automatically narrowed; if the data changes dramatically and the curve features are rich, the measurement range is automatically expanded.

[0007] Furthermore, it also includes a noise suppression and data repair module based on deep learning. The implementation steps of this module are: collect a large amount of measurement data with noise and missing data and annotate them, and divide the data into training set, verification set and test set; select a suitable deep learning model architecture, such as convolutional neural network (CNN) or generative adversarial network (GAN), use the training set data to train the model, and use the verification set data to evaluate the model performance and adjust the training parameters; in actual measurement, input the collected measurement data into the trained model, and the model will identify and remove the noise and repair the missing data.

[0008] Furthermore, it has intelligent path planning and automatic mapping functional modules, including: using environmental data collected by lidar and visual cameras to build a three-dimensional map of the measurement area, marking obstacles, terrain undulations and other information in the map, and determining the starting point, end point and key areas of the measurement; using path search algorithms such as the A* algorithm or the Dijkstra algorithm to search for the optimal path from the starting point to the end point in the three-dimensional map, and optimizing the path by considering obstacles, terrain influence and measurement accuracy requirements; the device moves automatically according to the planned path, and during the movement, the automatic mapping algorithm adjusts the measurement parameters of the lidar, visual camera and inertial measurement unit in real time according to the current position, posture and environmental information.

[0009] Furthermore, it also includes a real-time online high-precision curve fitting algorithm module, the working process of which is as follows: receiving the measurement data collected by the measuring device in real time, including the point cloud data of the laser radar, the feature point data of the visual camera, etc.; preprocessing the received data, removing outliers and performing filtering operations; using a combination of the least squares method and spline interpolation to perform curve fitting, selecting a suitable curve model according to the data distribution characteristics, calculating the curve model parameters by the least squares method, combining spline interpolation to ensure the continuity and smoothness of the curve, and using a piecewise fitting method for complex curves; as new measurement data is input, the curve model is updated in real time, and the fitted curve is displayed on the device display in real time, and the curve data is stored in the data storage unit.

[0010] Furthermore, the lidar can set the scanning frequency and angular resolution according to measurement requirements. For large-area topographic mapping, the scanning frequency is appropriately reduced to improve the measurement speed; for fine curve mapping, the scanning frequency and angular resolution are increased to obtain more detailed data.

[0011] Furthermore, during the shooting process, the vision camera adjusts parameters such as the exposure time and aperture size according to the light conditions to ensure clear images, and uses image enhancement algorithms to improve the contrast and clarity of the images for subsequent feature extraction and matching; at the same time, the internal and external parameters of the camera are determined through camera calibration technology to convert the image pixel coordinates into world coordinates.

[0012] Furthermore, the inertial measurement unit is calibrated before use to eliminate zero drift and errors, and inertial data is recorded in real time during the measurement process and time-synchronized with the data of the lidar and vision camera.

[0013] Furthermore, the evaluation indicators for the measurement environment and the complexity of the target curve include the density distribution of lidar point cloud data, the curvature change of curves in visual images, etc., and potential interference factors such as the distribution of obstacles and terrain undulation in the surrounding environment are also considered.

[0014] Furthermore, the curve model includes polynomial curves, Bezier curves, etc. The sum of the squares of the errors between the curve and the measurement data is minimized by the least squares method to determine the parameters of the curve model.

[0016] Beneficial effects

[0017] Through the multi-modal fusion measurement system, the advantages of lidar, vision cameras, and inertial measurement units are fully utilized to complement each other's data, effectively improving the accuracy of curve mapping. The noise suppression and data repair algorithm based on deep learning can remove noise and repair missing data parts, further enhancing the accuracy of the data. The real-time online high-precision curve fitting algorithm fits the data in a timely manner during the measurement process, reducing error accumulation and ensuring high-precision curve fitting. The adaptive dynamic measurement range adjustment algorithm enables the device to automatically adjust the measurement range according to the measurement environment and the complexity of the target curve. Whether in an open plain or a complex mountainous area, it can efficiently and accurately complete the mapping task. The intelligent path planning and automatic mapping algorithm can plan the optimal mapping path based on the environmental modeling results, avoid obstacles, and adapt to various complex terrain and environmental conditions. The intelligent path planning and automatic mapping algorithm realizes the automation of the measurement process, reduces manual operation and intervention, and greatly improves the measurement efficiency. The adaptive dynamic measurement range adjustment algorithm reasonably adjusts the measurement range on the premise of ensuring measurement accuracy, avoiding unnecessary repeated measurements, and saving time and resources. The real-time online high-precision curve fitting algorithm performs curve fitting while collecting data, without waiting for all data to be collected before processing, further improving the work efficiency. The automated functions of the device of the present invention, such as automatic path planning and automatic measurement parameter adjustment, reduce the work burden of the operator and lower the labor intensity. The operator only needs to perform simple task settings and parameter adjustments before the measurement and necessary monitoring during the measurement, reducing the pressure of long-term manual operation in a complex environment. Description of the Drawings

[0018] Figure 1 Algorithm flowchart. Detailed Implementation Manner

[0019] Example 1

[0020] Implementation Steps of the Multi-modal Fusion Measurement System

[0021] Lidar data acquisition: The lidar emits laser beams, measures the time delay of the reflected light, and calculates the distance between the target object and the device to generate three-dimensional point cloud data. During the acquisition process, the scanning frequency and angular resolution of the lidar are set according to the measurement requirements. For example, for large-area terrain mapping, the scanning frequency can be appropriately reduced to improve the measurement speed; for fine curve mapping, the scanning frequency and angular resolution are increased to obtain more detailed data.

[0022] Visual camera data acquisition: The visual camera captures images of the target curve and uses an image sensor to record light information. Through camera calibration technology, the internal and external parameters of the camera are determined, and the image pixel coordinates are converted into world coordinates. During the shooting process, parameters such as the exposure time and aperture size of the camera are adjusted according to the light conditions to ensure clear images. At the same time, image enhancement algorithms are adopted to improve the contrast and clarity of the images, facilitating subsequent feature extraction and matching.

[0023] Inertial measurement unit data acquisition: The inertial measurement unit measures the acceleration and angular velocity of the device through an accelerometer and a gyroscope, and then calculates the attitude and motion state of the device. Before use, the inertial measurement unit is calibrated to eliminate zero drift and errors. During the measurement process, inertial data is recorded in real time and time-synchronized with the data of the lidar and visual camera.

[0024] Data fusion: The data collected by the lidar, visual camera, and inertial measurement unit are transmitted to the data processing unit. A data fusion algorithm based on Kalman filtering is adopted to fuse multi-source data. First, corresponding weights are assigned to each type of data according to the measurement accuracy and error characteristics of different sensors. Then, the data is predicted and updated through the Kalman filtering algorithm to obtain high-precision measurement data after fusion.

[0025] Implementation steps of the adaptive dynamic measurement range adjustment algorithm

[0026] Environment and target analysis: Before starting the measurement, the device evaluates the complexity of the measurement environment and the target curve through the preliminary scan of the lidar and the image analysis of the visual camera. For example, calculate indicators such as the density distribution of lidar point cloud data and the curvature change of the curve in the visual image to judge the complexity of the curve. At the same time, according to the distribution of obstacles and terrain undulations in the surrounding environment, the difficulty of the measurement and potential interference factors are evaluated.

[0027] Initial setting of the measurement range: According to the results of the environment and target analysis, the initial measurement range is set. For simple and regular curves, such as straight line segments or large-radius arc curves, a larger measurement range is set to improve the measurement efficiency. For complex and detailed curves, such as curves with multiple inflection points and small-radius arcs, a smaller measurement range is set to ensure that the detailed features of the curve can be accurately captured.

[0028] Real-time adjustment: During the measurement process, the device continuously monitors the changes in the measurement data. If it is found that the data changes gently within the current measurement range and there are no obvious curve features, it indicates that the current measurement range may be too large, and the measurement range is automatically reduced to improve the measurement resolution. On the contrary, if it is found that the data changes violently within the measurement range and the curve features are rich, it indicates that the current measurement range may be too small, and the measurement range is automatically expanded to obtain more comprehensive curve information.

[0029] Implementation Steps of Noise Suppression and Data Repair Algorithm Based on Deep Learning

[0030] Data Preparation: Collect a large amount of measurement data with noise and data missing, including lidar point cloud data, visual image data, etc. Label these data to mark the noise part and the data missing part. Then, divide the data into training set, validation set and test set for training the deep learning model.

[0031] Model Training: Select a suitable deep learning model architecture, such as Convolutional Neural Network (CNN) or Generative Adversarial Network (GAN). Input the training set data into the model, and adjust the parameters of the model through the backpropagation algorithm, so that the model can accurately identify the noise and data missing parts, and learn how to remove noise and repair data. During the training process, use the validation set data to evaluate the performance of the model and adjust the training parameters to prevent the model from overfitting.

[0032] Noise Suppression and Data Repair: During the actual measurement process, input the collected measurement data into the trained deep learning model. The model first identifies the noise part in the data and removes the noise through a specific algorithm. Then, for the data missing part, the model generates reasonable data for repair according to the learned surrounding data features to obtain high-quality measurement data.

[0033] Implementation Steps of Intelligent Path Planning and Automatic Mapping Algorithm

[0034] Environmental Modeling: Use the environmental data collected by lidar and visual cameras to construct a three-dimensional map of the measurement area. Mark information such as obstacles and terrain undulations in the map. At the same time, according to the requirements of the measurement task, determine the starting point, ending point and key areas of the measurement.

[0035] Path Planning: Adopt path search algorithms such as A* algorithm or Dijkstra algorithm to search for the optimal path from the starting point to the ending point in the three-dimensional map. During the search process, consider the blocking of obstacles, the influence of terrain and the requirements of measurement accuracy, and optimize the path. For example, avoid the path passing through areas with dense obstacles and try to choose a route with flat terrain and good visibility conditions.

[0036] Automatic Mapping: The device moves automatically according to the planned path. During the movement, the automatic mapping algorithm adjusts the measurement parameters of the lidar, visual camera and inertial measurement unit in real time according to the current position, attitude and environmental information. For example, when the device approaches the key area of the curve, improve the scanning accuracy of the lidar and the resolution of the visual camera; when the device encounters bumps or vibrations, adjust the measurement parameters according to the data of the inertial measurement unit to ensure the stability and accuracy of the measurement data.

[0037] Implementation Steps of Real-time Online High-precision Curve Fitting Algorithm

[0038] Data Reception: The measurement data collected in real time by the measurement device, including the point cloud data of the lidar, the feature point data of the vision camera, etc., are quickly transmitted to the curve fitting algorithm module through the data transmission interface.

[0039] Data Preprocessing: The received data is preprocessed, including operations such as removing outliers and filtering. By setting reasonable thresholds, data points that deviate significantly from the normal range due to measurement errors or interference are removed. Then, a filtering algorithm, such as Gaussian filtering, is used to smooth the data and reduce the impact of noise on curve fitting.

[0040] Curve Fitting: A method combining the least squares method and spline interpolation is used for curve fitting. First, according to the distribution characteristics of the data, a suitable curve model is selected, such as a polynomial curve, a Bezier curve, etc. Then, the parameters of the curve model are calculated by the least squares method to minimize the sum of the squares of the errors between the curve and the measurement data. During the fitting process, the spline interpolation method is combined to ensure the continuity and smoothness of the curve. For complex curves, a segmented fitting method is used to improve the fitting accuracy.

[0041] Real-time Update and Display: As new measurement data is continuously input, the curve fitting algorithm updates the curve model in real time. The fitted curve is displayed on the display screen of the device in real time for the operator to view the measurement results in real time. At the same time, the curve data is stored in the data storage unit for subsequent analysis and processing.

Claims

1. A curve surveying and mapping device for engineering surveying, characterized in that: The invention comprises a multimodal fusion measurement system, which integrates a laser radar, a visual camera and an inertial measurement unit; the laser radar is used to quickly obtain three-dimensional point cloud data of a target object and measure the contour and distance information of a curve; the visual camera is used to capture the texture and feature point detail information of the curve by using image recognition technology; the inertial measurement unit is used to monitor the posture and motion state of the device in real time and compensate for the measurement error caused by the movement or vibration of the device; the multimodal fusion measurement system also comprises a data processing unit, which adopts a data fusion algorithm based on Kalman filtering to fuse the data collected by the laser radar, the visual camera and the inertial measurement unit.

2. The curve surveying and mapping device for engineering surveying according to claim 1, characterized in that: The device is equipped with an adaptive dynamic measurement range adjustment function module, which is achieved in the following way: before measurement, the device evaluates the complexity of the measurement environment and the target curve through preliminary scanning of the laser radar and image analysis of the visual camera, and sets the initial measurement range according to the evaluation results.

3. The curve surveying and mapping device for engineering surveying according to claim 1, characterized in that: It also includes a noise suppression and data repair module based on deep learning. The implementation steps of this module are: collect a large amount of measurement data with noise and missing data and annotate them, divide the data into training set, validation set and test set; select a suitable deep learning model architecture.

4. The curve surveying and mapping device for engineering surveying according to claim 1, characterized in that: It has intelligent path planning and automatic mapping functional modules, including: using the environmental data collected by the lidar and visual camera to build a three-dimensional map of the measurement area, marking obstacles, terrain undulations and other information in the map, and determining the starting point, end point and key areas of the measurement; using the A* algorithm or Dijkstra algorithm path search algorithm to search for the optimal path from the starting point to the end point in the three-dimensional map, while optimizing the path by considering obstacles, terrain influence and measurement accuracy requirements; the device moves automatically along the planned path, and during the movement, the automatic mapping algorithm adjusts the measurement parameters of the lidar, visual camera and inertial measurement unit in real time according to the current position, posture and environmental information.

5. The curve surveying and mapping device for engineering surveying according to claim 1, characterized in that: It also includes a real-time online high-precision curve fitting algorithm module, the workflow of which is as follows: receiving measurement data collected by the measuring device in real time, including point cloud data of the laser radar and feature point data of the visual camera; preprocessing the received data, removing outliers and performing filtering operations; using a combination of the least squares method and spline interpolation to perform curve fitting, selecting a suitable curve model based on the data distribution characteristics, calculating the curve model parameters through the least squares method, combining spline interpolation to ensure the continuity and smoothness of the curve, and using a segmented fitting method for complex curves; as new measurement data is input, the curve model is updated in real time, and the fitted curve is displayed on the device display in real time, and the curve data is stored in the data storage unit.

6. The curve surveying and mapping device for engineering surveying according to claim 1, characterized in that: The laser radar can set the scanning frequency and angular resolution according to the measurement requirements. For large-area terrain mapping, the scanning frequency can be appropriately reduced to increase the measurement speed; for fine curve mapping, the scanning frequency and angular resolution can be increased to obtain more detailed data.

7. The curve surveying and mapping device for engineering surveying according to claim 1, characterized in that: During the shooting process, the visual camera adjusts parameters such as exposure time and aperture size according to light conditions to ensure image clarity, and uses an image enhancement algorithm to improve image contrast and clarity to facilitate subsequent feature extraction and matching; at the same time, the camera's intrinsic and extrinsic parameters are determined through camera calibration technology to convert image pixel coordinates into world coordinates.

8. The curve surveying and mapping device for engineering surveying according to claim 1, characterized in that: The inertial measurement unit is calibrated before use to eliminate zero drift and error, and inertial data is recorded in real time during the measurement process and synchronized with the data of the lidar and visual camera.

9. The curve surveying and mapping device for engineering surveying according to claim 2, characterized in that: The evaluation indicators for the complexity of the measurement environment and target curve include the density distribution of lidar point cloud data, the curvature change of the curve in the visual image, etc., while taking into account the distribution of obstacles in the surrounding environment and potential interference factors such as terrain undulations.

10. The curve surveying and mapping device for engineering surveying according to claim 5, characterized in that: The curve model includes a polynomial curve and a Bezier curve. The sum of square errors between the curve and the measured data is minimized by the least square method to determine the parameters of the curve model.

Images