WO2023178727A1

WO2023178727A1 - Indoor three-dimensional barrier-free map generation method based on lidar point cloud and bim collision simulation

Info

Publication number: WO2023178727A1
Application number: PCT/CN2022/084953
Authority: WO
Inventors: 薛帆; 叶嘉安; 吴怡洁; 陈哲
Original assignee: 香港大学深圳研究院
Priority date: 2022-03-24
Filing date: 2022-04-02
Publication date: 2023-09-28
Also published as: CN114818051A

Abstract

An indoor three-dimensional barrier-free map generation method based on a LiDAR point cloud and BIM collision simulation, relating to the technical field of building construction, and for use in solving the problem of difficulty in planning indoor barrier-free access routes for special crowds, and constructing an indoor barrier-free access three-dimensional map capable of being matched with multiple types of barrier-free auxiliary passing equipment. The method comprises the following steps: S10, laser point cloud collection and preprocessing: performing laser point cloud collection on an indoor space, and preprocessing the collected multi-frame point cloud; S20, point cloud quantity counting and elevation fitting: performing point cloud quantity counting on the preprocessed point cloud in a vertical direction; S30, generating a three-dimensional voxel model corresponding to a ground point cloud; and S40, passage simulation and collision detection: performing passage simulation and collision detection to generate a plurality of passage domains adapted to different passage equipment.

Description

Indoor 3D barrier-free map generation method based on LiDAR point cloud and BIM collision simulation

Technical field

The invention relates to the technical field of building construction. More specifically, the invention relates to an indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation.

Background technique

In the process of constructing a barrier-free indoor map, if there is no original CAD drawing of the building for which the indoor map is to be constructed, staff need to measure the dimensions of the walls, doors and windows of each indoor room to obtain the indoor map of the building. At the same time, according to The measurement results further identify accessible indoor routes suitable for special groups such as wheelchair users.

The construction process of this kind of indoor map requires a large number of staff with professional mapping capabilities to personally survey and map the indoor environment, which requires high staff capabilities and a large amount of labor. It is also impossible to accurately plan a three-dimensional digital map of indoor barrier-free passages.

Contents of the invention

In order to overcome the shortcomings of the existing technology, the present invention provides an indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation.

The technical solution adopted by the present invention to solve its technical problems is: an indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation. The improvement is that the method includes the following steps:

S10, laser point cloud collection and preprocessing, collect laser point clouds in indoor spaces, and preprocess the collected multi-frame point clouds. The preprocessing process includes point cloud noise reduction processing and multi-frame point cloud registration;

S20. Point cloud quantity statistics and elevation fitting. Perform point cloud quantity statistics in the vertical direction on the preprocessed point cloud, perform elevation fitting on the floor and ceiling, and detect ground point clouds;

S30. Generate a three-dimensional voxel model corresponding to the ground point cloud, extract a universal barrier-free passable area on the three-dimensional voxel model, and segment the passable voxel area at the position where the height difference changes;

S40, traffic simulation and collision detection, conduct traffic simulation and collision detection, and generate multiple traffic domains suitable for different traffic equipment.

Further, in step S10, the python function for removing outlier points in the Open3D library is used to perform noise reduction processing on the point cloud.

Further, in step S10, a laser mobile scanning device is used to conduct laser point cloud scanning of the interior of the museum, using point or segment scanning mode.

Further, in step S10, the Python registration function provided by the Open3D library is used for point cloud registration, and the iterative nearest point algorithm is used to determine the relative pose between different point clouds with overlapping areas, including the rotation matrix R and translation. Vector t, and perform rigid body transformation RC+t on the point set C corresponding to the point cloud, and register multiple point clouds on the same floor and cross-floor point clouds to a unified spatial coordinate system.

Further, in step S20, the indoor space is a museum, and step S20 includes the following steps:

Step S20 includes the following steps:

Using the set value as a separation, count the frequency of points collected in each vertical separation of the museum point cloud;

According to the number of floors k in the museum, hybrid Gaussian fitting is performed. The number of Gaussian functions to be fitted is twice the number of floors. The derivative-free optimization method is used for optimization and solution to obtain the elevation of the ceiling and floor of each floor in the point cloud;

After estimating the floor elevation of each floor in the point cloud, the point cloud located in the corresponding elevation interval is segmented and used as the ground point cloud of each floor to provide reference data for ground height difference detection, barrier-free traffic simulation and collision detection.

Further, the set value is 0.3m.

Further, step S30 also includes screening accessible areas for special groups. Specific steps include:

S301. The side length of the three-dimensional voxel model is set according to the size requirements for special groups of people to pass indoors, and the planar connected areas in the three-dimensional voxel model are calculated;

S302. When the difference in elevation between adjacent ground points reaches the set value of the side length of the three-dimensional voxel model or more, the two adjacent points are regarded as non-horizontally connected;

S303. The construction of the horizontally connected region is completed through the region growing algorithm under horizontal eight-connectivity.

Further, the side length of the three-dimensional voxel model is set to 0.1m.

Further, in step S40, for wheelchairs or strollers used by special groups, traffic simulation and collision detection of wheelchairs or strollers are performed to ensure the accessibility of the access equipment in areas with a small range of activities.

Further, step S40 includes the following steps:

S401. Collect multiple representative wheelchair and stroller models, and generate three-dimensional digital maps of barrier-free passages corresponding to the multiple wheelchair and stroller models;

S402. Call the collision detection interface provided by Revit, input the wheelchair and stroller models one by one, and input the point cloud at the same time;

S403. Traverse all voxel positions of each accessible area, calculate when the horizontal center of the wheelchair or stroller model is placed at the voxel position, traverse the circumferential angle at set angle intervals, and check whether the wheelchair or stroller model is at this position. Whether each angle collides with the point cloud;

S404. If the result returned by the Revit interface indicates that there is a collision, remove the voxel position from the barrier-free voxels of the wheelchair or stroller model;

S405. After completing the traversal of all voxel areas and all angles, generate a barrier-free voxel model adapted to a single stroller or wheelchair model.

The beneficial effects of the present invention are: it aims to solve the problem of difficulty in planning indoor barrier-free passage routes for special groups; in the absence of building CAD drawings, it can utilize the currently rapidly developing LiDAR three-dimensional scanning technology and the relatively complete BIM software Collision detection function to build a three-dimensional map of indoor barrier-free passages that can adapt to multiple types of barrier-free auxiliary equipment.

Description of the drawings

Figure 1 is a schematic flow chart of the indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation according to the present invention.

Figures 2 and 3 are diagrams showing specific embodiments of the indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation of the present invention.

Figure 4 is a schematic diagram of eight horizontal connections of the three-dimensional voxel model in the present invention.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and examples.

The concept, specific structure and technical effects of the present invention will be clearly and completely described below in conjunction with the embodiments and drawings to fully understand the purpose, features and effects of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, other embodiments obtained by those skilled in the art without exerting creative efforts are all protection scope of the present invention. In addition, all the connections/connection relationships involved in the patent do not only refer to the direct connection of components, but refer to the fact that a better connection structure can be formed by adding or reducing connection auxiliary parts according to the specific implementation conditions. Various technical features in the invention can be combined interactively without conflicting with each other.

Referring to Figure 1, the present invention discloses an indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation. In this embodiment, the method includes steps S10-S40, the contents of which are as follows:

In this embodiment, the indoor space is a museum.

In step S10, a laser mobile scanning device is used to conduct laser point cloud scanning of the interior of the museum, using point or segment scanning mode. Mobile scanning equipment facilitates scanners to move flexibly in the museum space. They can more freely adjust the scanning angle of the lidar according to the surface of the scanned object, and can complete the scanning of the museum's multi-floor space in a short period of time.

After completing the scan, the point cloud needs to be quality checked, and there is generally a certain amount of noise. In order to reduce the impact of point cloud noise on subsequent processing, this embodiment uses the python function for removing outliers in the Open3D library to perform noise reduction processing on the point cloud.

In the scanning operation of large indoor spaces such as museums, in order to reduce the positioning drift error caused by scanning and limited by the endurance of the equipment, in this embodiment, a point or segment scanning mode is used, and then the acquisition is Multiple point cloud data are obtained for registration.

Referring to Figures 2 and 3, for laser point cloud scanning, the present invention provides a specific embodiment. The scanning route is shown with reference to Figures 2 and 3. Taking the G and 1 floors of the Proving Ground Museum as an example, it is To reduce the positioning drift problem caused by the scanning process of mobile scanning equipment, the scanning route is mostly planned in a closed loop to cover the halls and theme exhibition halls in the museum, as shown in Figure 3, other scanning segments except segment 4. Secondly, in order to prepare for subsequent registration, it is necessary to ensure that there is sufficient overlapping area between adjacent scanning segments, that is, it is necessary to adjust the scanning angle at the connection points such as the entrance and exit between two halls, stairs and corridors, and collect adjacent areas within the visual range. Scan the regional point cloud corresponding to the segment. In addition, during the scanning process, special attention should be paid to stairs, etc. that appear on the same floor.

After completing the laser point cloud scanning, in this embodiment, the Python registration function provided by the Open3D library is used to perform point cloud registration, and the iterative closest point algorithm is used to determine the relative pose between different point clouds in overlapping areas, including The rotation matrix R and the translation vector t are used, and the point set C corresponding to the point cloud is subjected to rigid body transformation RC+t, and multiple point clouds on the same floor and cross-floor point clouds are registered to a unified spatial coordinate system.

Step S20 includes the following steps:

Using the set value as a separation, count the frequency of points collected in each vertical separation of the museum point cloud; in this embodiment, the setting value is 0.3m;

After the floor elevation of each floor in the point cloud is estimated, the point cloud located in the corresponding elevation interval is segmented and used as the ground point cloud of each floor, providing reference data for ground height difference detection, barrier-free traffic simulation and collision detection.

Access facilities such as wheelchairs and strollers generally cannot easily pass on floors with height differences such as steps. Therefore, step three also requires screening of accessible areas for special groups based on height differences. Specific steps include:

S301. The side length of the three-dimensional voxel model is set according to the size requirements for special groups of people to pass indoors, and the plane connected areas in the three-dimensional voxel model are calculated; in this embodiment, the setting value of the side length of the three-dimensional voxel model is 0.1m;

S303. The construction of the horizontally connected region is completed through the region growing algorithm under horizontal eight-connectivity. In this embodiment, the implementation of this algorithm can refer to the following function 1:

Combined with what is shown in Figure 4, it is a schematic diagram of the horizontal eight-connectivity of the three-dimensional voxel model.

The result of this step is a barrier-free ground voxel area filtered by the height difference. Barrier-free access is ensured between adjacent voxels, but barrier-free access is not supported outside the edge of the voxel area.

S40, traffic simulation and collision detection, conduct traffic simulation and collision detection, and generate multiple traffic domains suitable for different traffic equipment; for wheelchairs or strollers used by special groups, conduct traffic simulation and collision detection of wheelchairs or strollers, To ensure the accessibility of access equipment in areas with a small range of activities.

In step S40, BIM software needs to be used to conduct collision detection on a variety of representative tools for special people, including wheelchairs and strollers, to build a barrier-free three-dimensional digital map that adapts to different tools.

Step S40 includes the following steps:

S401. Collect multiple representative wheelchair and stroller models, and generate a three-dimensional digital map of barrier-free passages corresponding to the multiple wheelchair and stroller models; in this embodiment, collect 10 representative wheelchair and stroller models;

S403. Traverse all voxel positions of each accessible area, calculate when the horizontal center of the wheelchair or stroller model is placed at the voxel position, traverse the circumferential angle at set angle intervals, and check whether the wheelchair or stroller model is at this position. Whether each angle collides with the point cloud; in this embodiment, the angle is set to 15°;

Subsequent path planning and navigation functions can use the A* algorithm to calculate the shortest path on the barrier-free voxel map.

The invention provides a method for generating a three-dimensional digital map of indoor barrier-free passages based on LiDAR point cloud and BIM software collision detection, aiming to solve the problem of difficult indoor barrier-free passage route planning for special groups of people. In the absence of building CAD drawings, the currently rapidly developing LiDAR 3D scanning technology and the relatively complete collision detection function in BIM software can be used to build a high-precision, fine-grained indoor environment that can adapt to multiple types of barrier-free auxiliary traffic equipment. The barrier-free three-dimensional map can meet the special indoor traffic needs of wheelchairs, cars, etc. At the same time, the semi-automatic solution proposed by the present invention also reduces the manpower requirements for mapping work and reduces the cost of mapping.

The above is a detailed description of the preferred implementation of the present invention, but the present invention is not limited to the embodiments. Those skilled in the art can also make various equivalent modifications or substitutions without violating the spirit of the present invention. , these equivalent modifications or substitutions are included in the scope defined by the claims of this application.

Claims

An indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation, which is characterized in that the method includes the following steps:

S10, laser point cloud collection and preprocessing, collect laser point clouds in indoor spaces, and preprocess the collected multi-frame point clouds. The preprocessing process includes point cloud noise reduction processing and multi-frame point cloud registration;

S20. Point cloud quantity statistics and elevation fitting. Perform point cloud quantity statistics in the vertical direction on the preprocessed point cloud, perform elevation fitting on the floor and ceiling, and detect ground point clouds;

S30. Generate a three-dimensional voxel model corresponding to the ground point cloud, extract a universal barrier-free passable area on the three-dimensional voxel model, and segment the passable voxel area at the position where the height difference changes;

S40, traffic simulation and collision detection, conduct traffic simulation and collision detection, and generate multiple traffic domains suitable for different traffic equipment.
The indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation according to claim 1, characterized in that, in step S10, the python function for removing outliers in the Open3D library is used to denoise the point cloud. deal with.
The indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation according to claim 2, characterized in that, in step S10, a laser mobile scanning device is used to scan the interior of the museum with a laser point cloud, using point points or segmented scan mode.
The indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation according to claim 2, characterized in that, in step S10, the Python registration function provided by the Open3D library is used to perform point cloud registration, using iteration The closest point algorithm determines the relative pose between different point clouds with overlapping areas, including the rotation matrix R and the translation vector t, and performs rigid body transformation RC+t on the point set C corresponding to the point cloud, and transforms multiple points on the same floor into The individual point clouds and cross-floor point clouds are registered to a unified spatial coordinate system.
The indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation according to claim 1, characterized in that in step S20, the indoor space is a museum, and step S20 includes the following steps:

Using the set value as a separation, count the frequency of points collected in each vertical separation of the museum point cloud;

According to the number of floors k in the museum, hybrid Gaussian fitting is performed. The number of Gaussian functions to be fitted is twice the number of floors. The derivative-free optimization method is used for optimization and solution to obtain the elevation of the ceiling and floor of each floor in the point cloud;

After estimating the floor elevation of each floor in the point cloud, the point cloud located in the corresponding elevation interval is segmented and used as the ground point cloud of each floor to provide reference data for ground height difference detection, barrier-free traffic simulation and collision detection.
The indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation according to claim 5, characterized in that the setting value is 0.3m.
The indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation according to claim 1, characterized in that step S30 also includes screening of accessible areas for special groups, and the specific steps include:

S301. The side length of the three-dimensional voxel model is set according to the size requirements for special groups of people to pass indoors, and the planar connected areas in the three-dimensional voxel model are calculated;

S302. When the difference in elevation between adjacent ground points reaches the set value of the side length of the three-dimensional voxel model or more, the two adjacent points are regarded as non-horizontally connected;

S303. The construction of the horizontally connected region is completed through the region growing algorithm under horizontal eight-connectivity.
The indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation according to claim 7, characterized in that the set value of the side length of the three-dimensional voxel model is 0.1m.
The indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation according to claim 7, characterized in that, in step S40, the wheelchair or stroller used by special groups is provided with a wheelchair or stroller to pass. Simulation and collision detection to ensure the passability of access equipment in areas with small activity ranges.
The indoor three-dimensional barrier-free map generation method based on LiDAR point cloud and BIM collision simulation according to claim 9, characterized in that step S40 includes the following steps:

S401. Collect multiple representative wheelchair and stroller models, and generate three-dimensional digital maps of barrier-free passages corresponding to the multiple wheelchair and stroller models;

S402. Call the collision detection interface provided by Revit, input the wheelchair and stroller models one by one, and input the point cloud at the same time;

S403. Traverse all voxel positions of each accessible area, calculate when the horizontal center of the wheelchair or stroller model is placed at the voxel position, traverse the circumferential angle at set angle intervals, and check whether the wheelchair or stroller model is at this position. Whether each angle collides with the point cloud;

S404. If the result returned by the Revit interface indicates that there is a collision, remove the voxel position from the barrier-free voxels of the wheelchair or stroller model;

S405. After completing the traversal of all voxel areas and all angles, generate a barrier-free voxel model adapted to a single stroller or wheelchair model.