CN116605772A - A tower crane collision warning method based on multi-integrated system - Google Patents

Abstract

The invention provides a tower crane collision warning method based on a multi-integrated system, belonging to the technical field of building tower crane construction, including: adopting a dual-antenna combined positioning and attitude determination technology to obtain high-precision pose information; combining laser point clouds and pose information to optimize sensors Arrange parameters, generate the site cloud base map of the construction site, and extract the outer bounding box of the building; use the position information of the tower crane hook and the encoder data to segment the laser point cloud of the hoisting object, generate the optimal outer bounding box of the hoisting object, and combine the bounding box of the base map with the The bounding box of the hoisting object performs collision detection to obtain the collision warning information of the hoisting object; through the combination of the server and the client, a collision information visualization module is built to display the 3D environment and collision warning information of the tower crane operation in real time. The present invention effectively solves the collision warning problem in tower crane operation by constructing a tower crane collision warning system based on the integration of dual antennas, encoders and laser radars, improves tower crane operation efficiency, and provides assistance for tower crane remote operation.

Description

A tower crane collision warning method based on multi-integrated system

technical field

The invention relates to the technical field of building tower crane construction, in particular to a multi-integrated system-based tower crane collision warning method.

Background technique

Tower cranes are widely used lifting equipment in the field of engineering construction. The operation activities in the use stage are a dynamic process of complex interactions between machinery, personnel and the environment, involving tower crane drivers, signal workers and other types of special operators. The level of operation safety is affected by special The comprehensive influence of various factors such as the knowledge and skills of the operators, physical and mental conditions, communication and coordination, etc., has a high risk. The stage of tower crane use is a stage where tower crane production safety accidents frequently occur. Timely and accurate collision warning information during hoisting operations is of great significance for controlling the safety risk level of tower crane operations.

In the traditional tower crane operation safety management, the technical management personnel stand aside to supervise and patrol inspection as the main means to implement the safety technical requirements proposed by the standards and regulations. The management effect is often limited by the experience and subjective judgment of the technical management personnel, and it is difficult to achieve Comprehensive, timely and reliable. In order to reduce the occurrence of tower crane accidents and make the tower crane run safely, smoothly and effectively, encoders, ultrasonic sensors, laser range finders, cameras, radio frequency tags, Global Positioning System (Global Positioning System, GPS), Ultra Wide Band (Ultra Wide Band, UWB) and other sensors have been used for early warning of tower crane collision accidents, but these sensors also have some shortcomings in use. The encoder obtains the position information of the hoisting object by measuring the rotation angle of the tower crane and the moving distance of the wire rope, but the encoder cannot perceive the surrounding environment of the hoisting object, and cannot obtain the spatial relationship between the hoisting object and the building. It is usually used in tower crane groups Collision warning between objects; Ultrasonic sensors, laser rangefinders and other distance measuring sensors are usually installed on the hook, and the collision accidents are warned by directly measuring the distance between the hoisting object and the obstacle on the construction site. The measurement accuracy is high, but it cannot provide visualization. The hoisting environment information and directional auxiliary operation information; the camera is mostly installed on the boom and trolley of the tower crane, and the tower crane driver's perception of the hoisting environment is enhanced through real-time transmission of images, but it is still necessary to manually judge the hoisting objects and surrounding objects distance, and the shooting angle of the camera is fixed, the three-dimensional effect of the image is poor, which is not conducive to the judgment of the collision distance; position sensors such as radio frequency tags, GPS, and UWB are usually installed on the hook, which can accurately obtain the position information of the hoisting object, but It is impossible to obtain the surrounding environmental information, and it is usually necessary to combine technologies such as Building Information Modeling (BIM) to realize early warning of collision information.

Aiming at the deficiencies of the above-mentioned various sensor systems in tower crane collision warning, it is necessary to learn from each other and propose a new tower crane collision warning method.

Contents of the invention

The present invention provides a tower crane collision early warning method based on a multi-integrated system, which is used to solve the problem that various sensors used for tower crane collision early warning in the prior art generally have poor space environment perception, rely too much on cooperation with surrounding systems, and cannot Identify the defects in obtaining more accurate collision warning information.

In the first aspect, the present invention provides a tower crane collision warning method based on a multi-integrated system, including:

The industrial computer on the side of the tower crane collects the laser radar point cloud data and the observation data of the integrated navigation system, as well as the observation data of the receiving encoder and the positioning result of the carrier phase difference of the hook receiver;

Using the observation data of the integrated navigation system, performing dual-antenna integrated navigation, positioning, and attitude determination to obtain real-time attitude information;

Using the lidar point cloud data and the real-time pose information, through geographic orientation and point cloud registration, the point cloud base map of the construction site is obtained, and the upper and lower bottom surface contours combined with the bounding box extraction algorithm are used to extract the point cloud of the construction site The building outline bounding box corresponding to the base map;

Segmenting and extracting the lidar point cloud data or using the encoder observation data to calculate the coordinates of the hoisting object, and combining the carrier phase difference positioning results of the hook receiver to generate a hoisting object bounding box;

Collision detection is performed on the bounding box of the outline of the building and the bounding box of the hoisting object to obtain the spatial relationship between the hoisting object and the building and collision warning information.

In the second aspect, the present invention also provides a tower crane collision warning system based on a multi-integrated system, including:

The collection and receiving module is used to collect lidar point cloud data and integrated navigation system observation data by the industrial computer on the side of the tower crane, as well as receive encoder observation data and hook receiver carrier phase difference positioning results;

The positioning and calculating module is used to use the observation data of the integrated navigation system to perform dual-antenna integrated navigation positioning and attitude determination to obtain real-time attitude information;

The orientation registration module is used to obtain the point cloud base map of the construction site by using the lidar point cloud data and the real-time pose information through geographic orientation and point cloud registration, and adopts the upper and lower bottom surface contours combined with the bounding box extraction algorithm, Extracting the building outline bounding box corresponding to the point cloud base map of the construction site;

The segmentation calculation module is used to segment and extract the lidar point cloud data or use the encoder observation data to calculate the coordinates of the hoisting object, and combine the carrier phase difference positioning results of the hook receiver to generate a hoisting object bounding box;

The collision detection module is configured to perform collision detection on the bounding box of the outline of the building and the bounding box of the hoisting object, and obtain the spatial relationship between the hoisting object and the building and collision warning information.

In a third aspect, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and operable on the processor. The tower crane collision warning method based on multi-integrated system is described.

In the fourth aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the multi-integrated system-based tower crane collision early warning as described in any one of the above is realized method.

The tower crane collision warning method based on the multi-integrated system provided by the present invention effectively solves the collision warning problem in the tower crane operation by constructing a tower crane collision warning system based on the integration of dual antennas, encoders and laser radars, and improves the tower crane operation efficiency. Remote operation provides assistance.

Description of drawings

In order to more clearly illustrate the present invention or the technical solutions in the prior art, the accompanying drawings that need to be used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings in the following description are the For some embodiments of the present invention, those of ordinary skill in the art can also obtain other drawings based on these drawings on the premise of not paying creative efforts.

Fig. 1 is the schematic flow chart of the tower crane collision early warning method based on multi-integrated system provided by the present invention;

Fig. 2 is the overall structural diagram of the multi-sensor integrated tower crane collision intelligent warning system provided by the present invention;

3 is a schematic diagram of the layout of the integrated navigation system and laser radar provided by the present invention;

Fig. 4 is the flow chart of the dual-antenna GNSS/SINS combined high-precision autonomous positioning and attitude determination algorithm provided by the present invention;

Fig. 5 is a flow chart of the laser point cloud base map and bounding box generation algorithm provided by the present invention;

Fig. 6 is a flow chart of the real-time monitoring and collision information measurement algorithm for hoisting objects provided by the present invention;

Fig. 7 is a flow chart of data interaction of the visualization system provided by the present invention;

Fig. 8 is a structural schematic diagram of a tower crane collision warning system based on a multi-integrated system provided by the present invention;

FIG. 9 is a schematic structural diagram of an electronic device provided by the present invention.

Reference signs:

1: Master GNSS antenna; 2: Slave GNSS antenna; 3: Integrated navigation system;

4: LiDAR.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention , but not all examples. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

Aiming at the deficiencies in the tower crane collision warning technology in the prior art, the present invention integrates a Global Navigation Satellite System (Global Navigation Satellite System, GNSS), an inertial strapdown navigation system (Strap-downInertial Navigation System, SINS), an encoder and a laser The tower crane collision intelligent early warning system composed of radar proposes a multi-integrated system tower crane collision early warning method, which can accurately obtain the spatial geometric information of the construction environment and hoisting objects, and has the characteristics of accurate early warning, high real-time performance and panoramic visualization of the hoisting environment.

Fig. 1 is a schematic flow chart of a tower crane collision warning method based on a multi-integrated system provided by an embodiment of the present invention, as shown in Fig. 1 , including:

Step 100: Collecting lidar point cloud data and integrated navigation system observation data by the industrial computer on the side of the tower crane, and receiving encoder observation data and hook receiver carrier phase differential positioning results;

Step 200: Use the observation data of the integrated navigation system to perform dual-antenna integrated navigation positioning and attitude determination to obtain real-time attitude information;

Step 300: Using the lidar point cloud data and the real-time pose information, through geographic orientation and point cloud registration, obtain the point cloud base map of the construction site, and use the upper and lower bottom surface contours combined with the bounding box extraction algorithm to extract the construction site The building outline bounding box corresponding to the site point cloud base map;

Step 400: Segment and extract the lidar point cloud data or use the encoder observation data to calculate the coordinates of the hoisting object, and combine the carrier phase difference positioning results of the hook receiver to generate a hoisting object bounding box;

Step 500: Perform collision detection on the bounding box of the outline of the building and the bounding box of the hanging object to obtain the spatial relationship between the hanging object and the building and collision warning information.

It should be noted that the embodiment of the present invention is based on the multi-sensor integrated intelligent early warning system for tower crane collision. The structure is shown in Figure 2. The laser radar and integrated navigation system are placed above the cockpit of the tower crane, and are connected to the industrial computer in the cockpit through cables; the GNSS receiver is installed above the hook of the tower crane to determine the position of the hook. The local area network transmits the position information to the industrial computer; the encoder is integrated in the hoist of the tower crane, which can measure the moving distance of the trolley on the boom and the distance of the hook drop, and transmit the measurement results to the industrial computer through the local area network. The integrated navigation system above the cockpit is connected to two GNSS antennas, which are fixedly connected to the laser radar. In the schematic diagram of the integrated navigation system and laser radar layout shown in Figure 3, the main GNSS antenna 1 and the slave GNSS antenna 2 are respectively arranged on the tower crane. On both sides of the boom, the integrated navigation system 3 and the lidar 4 are time-synchronized by sending Pulses Per Second (PPS) pulses.

Specifically, the industrial computer on the tower crane side receives the lidar point cloud data and the GNSS observation value and SINS observation value of the integrated navigation system in real time through the robot operating system (Robot Operating System, ROS), and at the same time receives the encoder data and the GNSS observation value at the hook through the local area network. Real-time dynamic carrier phase difference (Real-Time Kinematic, RTK) positioning results of the receiver.

Using the GNSS observations and SINS observations obtained in real time, the dual-antenna GNSS/SINS combined positioning and attitude determination is used to obtain real-time position and attitude information.

Rotate the tower crane 1.5 times clockwise and counterclockwise, save the received laser radar point cloud data and the calculated pose information, and optimize the relationship between the integrated navigation sensor and the laser radar through direct geo-orientation and point cloud registration. Spatial placement parameters, the point cloud base map of the construction site is obtained, and the upper and lower bottom contours combined with the bounding box extraction method is used to obtain a bounding box that is closer to the real outline of the building.

Carry out preliminary segmentation and extraction of the point cloud received in real time, use the observation data of the encoder to calculate the coordinates of the hoisting object, combine the position information of the RTK at the hook, filter out the laser point cloud of the hoisting object, and generate the optimal outer bounding box of the hoisting object, And carry out collision detection with the obtained base map bounding box, and obtain the spatial relationship between the hoisting object and the building and the collision warning information.

The point cloud base map and base map bounding box, as well as the hoisting object bounding box and collision warning information obtained by the tower base side server are transmitted to the ground client through the LAN, and the ground terminal visualization software is based on Unreal Engine 4 (UnrealEngine 4 , UE4) display the 3D environment of hoisting operation, collision distance information and potential collision targets in real time under different viewing angles.

The present invention establishes an accurate three-dimensional point cloud base map of the construction site by integrating GNSS, inertial navigation, encoder and laser radar, and can display the three-dimensional environment and early warning information of the hoisting operation at multiple angles in real time, and the display effect is more intuitive, which is beneficial to reduce the The risk of tower crane collision accidents; using the dual-antenna GNSS/SINS combined positioning and attitude determination method, the position and attitude calculation results are more accurate, and the accuracy of the point cloud base map and collision warning information is improved; in addition, the use of building bounding boxes and hoisting objects The optimal outer bounding box for collision detection greatly improves the calculation efficiency while ensuring the redundancy of collision warning; it also builds a visualization module based on the separation of the client and the server, which reduces the amount of data transmission and avoids data blocking. , which improves the real-time performance of the visualization module.

Based on the above-mentioned embodiments, the embodiment of the present invention constructs a three-dimensional point cloud base map of the construction site and determines the position of the lifting object in the point cloud base map, so it is necessary to calculate high-precision lidar pose information.

As shown in Figure 3, the two GNSS antennas fixed on the acquisition device can form a short baseline to realize motion-to-motion RTK, determine the precise baseline vector, and obtain an accurate attitude. In the GNSS/SINS combination, the dual-antenna GNSS attitude measurement can be used for initial alignment, and can also provide external attitude information for the navigation phase, especially improving the observability of the heading angle. Dual-antenna GNSS attitude measurement is still affected by signal occlusion, and the result output is discontinuous and noisy. It is usually combined with SINS. On the one hand, the gyroscope in SINS can smooth the GNSS attitude measurement results and make up for the interruption when the signal is out of lock. On the one hand, the accelerometer can calculate the horizontal angle, so that the dual-antenna GNSS/SINS combination can provide continuous, smooth and reliable 3D attitude information. The present invention uses the misalignment angle equation in the GNSS/SINS combined basic model as the state equation, and uses the heading angle provided by the dual-antenna GNSS and the pitch angle and roll angle provided by the addition as observation values to jointly form the basic elements of Kalman filtering. The specific process of the high-precision autonomous positioning and attitude determination of the dual-antenna GNSS/SINS combination is shown in Figure 4, including:

When the dual-antenna GNSS fixes the ambiguity correctly, an accurate baseline vector can be obtained, and the heading angle and pitch angle can be calculated as follows:

In the formula, Indicates the projection of the baseline vector in the east direction, /> represents the projection of the baseline vector in the north direction, Indicates the projection of the baseline vector in the vertical direction;

According to the output of inertial navigation, the pitch angle and roll angle are derived, and the observation equation of attitude update is obtained:

The inertial accumulator output can be expressed by the specific force equation:

In the formula, is the accumulative output value, /> is the line-of-sight acceleration in l system (local horizontal coordinate system), /> is the Coriolis force, /> is the centripetal acceleration to the ground, /> For gravity, the obtained /> and /> After that, you can compose about /> The equation of , where the heading angle is given by /> Provided, therefore, the accumulatively derived pitch and roll angles can be computed while using the Butterworth filter for /> The original observations are smoothed and denoised. /> Indicates the projection of the angular velocity of the e system relative to the i system under the l system, /> Indicates the projection of the angular velocity of the l system relative to the i system under the l system, /> Indicates the accumulative output value under the l system, /> Indicates the coordinate transformation matrix from the b system to the l system.

After the attitude observation value is obtained, the relationship with the state to be estimated is established by the following formula, that is, the observation equation:

In the formula, Represents the coordinate transformation matrix from the e system to the l system, /> Represents the coordinate transformation matrix from the e system to the l system, /> represents the unit matrix, /> means /> antisymmetric matrix of angles, /> Indicates the coordinate conversion matrix from the b -system to the e -system obtained by mechanical arrangement at the current moment.

Since the attitude angle is expressed in the l system, the observed value implies that in, while pending evaluation/> Indicates that under the e system, using the above formula, use the attitude angle as /> After expressing, the two sides are differentiated, and the observation equation can be obtained:

In the formula, a vector of residuals representing observations, /> represents the design matrix, /> Indicates the parameter to be estimated, /> represents the observation noise, /> , /> and /> Respectively represent the residual vector of heading angle, pitch angle and roll angle, /> , /> and /> Respectively represent the observed values of the attitude angle, /> , /> and /> Indicates the Euler angle obtained by inertial navigation attitude update, /> Medium /> , , /> Indicates the observation coefficient of the heading angle, /> , /> , /> Indicates the observation coefficient of the pitch angle, /> , /> , /> Indicates the observed coefficient of the roll angle, /> Indicates the pending status.

The main antenna obtains the observation equation of position update through RTK:

In the formula, Represents the observation value residual vector of the position update under the e system, /> Indicates the position of the lower inertial navigation system of the e system obtained by mechanical arrangement, /> Indicates the projection of the lever arm under the e system, /> Indicates the position of the center of the GNSS antenna in the e system, /> Indicates the error status of the position, /> Indicates misalignment angle, /> represents the observation noise.

Then the integrated navigation calculation is performed to obtain high-precision pose information, and the acceleration and gyroscope's zero bias are fed back and corrected according to the calculation results.

The present invention obtains the heading angle of the acquisition device through dynamic-to-dynamic RTK, and constructs an observation equation for attitude update by combining the inertial navigation accumulative output, and the pose information obtained by the combined navigation solution provides a spatial reference for subsequent point cloud data processing

Based on the above embodiment, as shown in Figure 5, the specific algorithm flow of the laser point cloud base map and bounding box generation proposed by the embodiment of the present invention includes:

Perform point cloud geo-orientation on the original point cloud, and convert the point cloud from the lidar coordinate system to the local horizontal coordinate system;

Perform inter-frame matching on the converted point cloud to obtain a preliminary laser point cloud base map;

Optimize the spatial placement parameters of lidar and inertial navigation to improve the accuracy of point cloud basemap;

The cloth simulation filter algorithm is used to divide the point cloud into ground points and non-ground points, which is convenient for the extraction of the bounding box of the object within the monitoring range. The basic idea of the cloth simulation filtering algorithm is to invert the 3D point cloud data. Imagine a piece of cloth covering the inverted point cloud. The cloth is affected by gravity and will be close to the ground. The final position covered by the cloth is the position of the ground point;

Use the Euclidean clustering method to cluster the discrete point clouds belonging to the same non-ground object into a point cloud object, and use the octree structure or KD tree structure for point cloud neighborhood query to realize the Euclidean non-ground point cloud Clustering, which clusters points whose distance between point clouds is less than a threshold into the same object. for each object -The shape algorithm calculates the out-of-plane boundary, and the bounding box of the object is obtained after extracting the bottom and top points of the point cloud object.

Among them, the point cloud data collected by LiDAR is based on the LiDAR coordinate system as a reference. During the data collection process, the LiDAR is constantly moving. Therefore, it is necessary to obtain the accurate space between the LiDAR and SINS when constructing the laser point cloud base map. relationship, and project the lidar coordinates to the local horizontal coordinate system.

Ground points measured by lidar The coordinates in the lidar coordinate system are /> , the relationship between the lidar and the inertial measurement unit is rigidly fixed. According to the coordinate transformation principle, the coordinates of the ground point in the inertial navigation coordinate system can be obtained as:

According to the position and attitude provided by the dual-antenna GNSS/SINS combination, the coordinates in the inertial navigation coordinate system can be transformed into the global coordinate system:

Further get:

Among the above formulas, Indicates any ground point /> Coordinates in the lidar coordinate system, /> Indicates the ground point Coordinates in the inertial navigation coordinate system, /> Indicates any ground point /> Coordinates in the WGS84 earth-centered earth-fixed coordinate system, /> Indicates the position of the origin of the lidar coordinate system in the inertial navigation coordinate system, /> Represents the rotation matrix for transforming the lidar coordinate system to the inertial navigation coordinate system, /> Indicates the position of the origin of the inertial navigation coordinate system in the WGS84 earth-centered ground-fixed space Cartesian coordinate system, Indicates the rotation matrix that transforms the inertial navigation coordinate system to the Cartesian coordinate system in WGS84 earth-centered ground-fixed space.

The attitude angle output by the inertial navigation system is the attitude of the inertial navigation coordinate system in the local horizontal coordinate system, and the local horizontal coordinate system can obtain the distance from the local horizontal coordinate system to the WGS84 earth-centered ground-fixed coordinate system through the geodetic coordinates of the origin of the local horizontal coordinate system. rotation matrix, so:

In the formula, is the rotation matrix from the inertial navigation coordinate system to the local horizontal coordinate system calculated according to the three attitude angles output by the inertial navigation system, /> It is the rotation matrix from the local horizontal coordinate system calculated according to the geodetic coordinates of the origin of the inertial navigation coordinate system to the WGS84 earth-centered earth-fixed space Cartesian coordinate system, and is substituted into:

The above formula is the positioning equation of tower crane laser scanning, which can directly calculate the coordinates of ground object points in the WGS84 space Cartesian coordinate system. Therefore, the coordinates of the feature points in the local horizontal coordinate system are:

in, is the coordinates of the origin of the local horizontal coordinate system in the WGS84 coordinate system, /> for

The rotation matrix of the WGS84 earth-centered earth-fixed space Cartesian coordinate system to the local horizontal coordinate system.

By projecting the point cloud from the lidar coordinates to the local horizontal coordinate system, the projection process involves the position of the origin of the lidar coordinate system in the inertial navigation coordinate system and the rotation matrix of the lidar coordinate system to the inertial navigation coordinate system , so need to optimize /> and /> , assuming that the laser scanning system scans the same feature point /> Two scans were performed to obtain:

In the formula, , /> , /> and /> , /> , /> is in two scans /> The measured value in the lidar coordinate system is a known value, /> , /> , /> and /> is the first scan /> Rotation matrix composed of position and attitude of inertial navigation at any time, /> , /> , /> and /> is the second scan /> The rotation matrix composed of the position and attitude of the inertial navigation at any time, they are all known values, only /> with /> is unknown, that is, only the lidar placement parameters that need to be solved are unknown. Subtracting the above two equations, we get:

In the calibration process, make the left side of the above equal sign the smallest, with /> unknown, /> Represents three translation parameters, /> It is a rotation matrix expressed by three angles around the X-axis, Y-axis and Z-axis respectively, so there are six independent unknowns:

Further, the nonlinear optimization algorithm (Levenberg-Marquardt, LM) is used to solve the six calibration parameters. The LM nonlinear algorithm obtains the least square sum of a set of nonlinear equations through iteration, and its mathematical model is as follows:

is a set of nonlinear equations, the LM algorithm finds a set of , such that /> Minimum, each point pair can form 3 equations, if there are n point pairs, 3 n equations can be formed, and the optimal solution can be obtained by LM algorithm , resulting in:

.

The invention uses direct geographical orientation to project the original point cloud to the local horizontal coordinate system, and uses a flexible and simple calibration method that does not require a three-dimensional control field, optimizes the placement parameters between the inertial navigation and the laser radar, and improves the accuracy of the point cloud base map.

Based on the above-mentioned embodiment, as shown in FIG. 6, the specific algorithm flow of the real-time monitoring of the hoisting object and the measurement of collision information proposed by the embodiment of the present invention includes:

Project the collected original laser point cloud to the local horizontal coordinate system, and convert the RTK positioning result of the hook from the WGS84 coordinate system to the local horizontal coordinate system as ;

Using point cloud filtering, the original point cloud is divided into ground points and non-ground points, and the surface point cloud is extracted from the non-ground point cloud as the target area for hoisting object extraction;

Carry out connectivity analysis on non-ground patches, merge adjacent non-ground patches, and use the merged result as the undetermined target area, and then there is a certain elevation difference between the building’s geometric size and the building’s boundary and the ground , remove the target area that is obviously not a building;

Using the position information of the hook The target closest to the hook is classified as the hoisting object target, and other targets are classified as large building targets. When the GNSS signal at the hook is blocked, the moving distance of the trolley and the descending distance of the hook measured by the encoder, and the dual-antenna GNSS /SINS combination to obtain the pose information, calculate the position of the hook /> ,Will Bringing in and enlarging the screening threshold can also get the target of the hoisting objects;

The three-dimensional outer bounding box of the hanging object is generated on the XY plane to generate a two-dimensional top-view projection. Since the hanging object will rotate around the z-axis during the hoisting process, the top-view projection is rotated around the center to simulate the possible rotation range of the hanging object and obtain The circumscribed rectangle of the rotation range is used as the optimal outer bounding box;

Use the AABB tree to index the base map primitives, then perform collision detection based on the AABB tree for the base map primitives and hanging objects, traverse the AABB trees of the two, and judge the nearest base from the outer bounding box of the hanging objects After drawing the primitives, the distance vector from the corner point of the bounding box of the lifting object to the nearest primitive is calculated in turn, so as to obtain the real minimum collision distance.

It should be noted that due to the complex operating environment of the tower crane, the laser point cloud scanned in real time includes both hoisting objects and buildings and other construction equipment. Therefore, the embodiment of the present invention adopts a top-down strategy to identify the hoisting object area from the target area level, and then accurately extract the hoisting object point cloud and bounding box according to the difference in the target detail features. The data volume of the original point cloud is relatively large. In order to improve the speed of real-time monitoring of the hoisting object, the embodiment of the present invention uses point cloud filtering to extract the non-ground patch point cloud from the original point cloud. The process includes:

In point cloud filtering, the data quality of the point cloud has a great influence on its results and computational efficiency. For example, if the point cloud density is too high, sometimes reaching hundreds of points per square, a large amount of computation is required for point cloud local feature calculation and filtering processing; point cloud density is too low or the distribution is uneven, and the point cloud local feature calculation The accuracy has a serious impact, thus affecting the point cloud filtering results; data missing areas may cause discontinuity in the distribution of ground point clouds, resulting in weakening of the adjacent spatial relationship between point clouds, thereby increasing the probability of point cloud filtering errors. Therefore, the embodiment of the present invention adopts the virtual gridding technology to process the original point cloud, so as to make the distribution of the point cloud as uniform as possible and reduce data holes. First, the region is evenly divided by a grid of a certain size, where the grid size is set as . Then, traverse each grid, and take the lowest point in the grid as the grid point /> , while the remaining point clouds in the grid are labeled other point clouds /> . Finally, all grids without point clouds are detected and a virtual point is interpolated for each grid without point clouds. The point is at the center of the grid, and the elevation of the point is obtained by the nearest neighbor interpolation method.

Here, the point cloud filtering method used in the embodiment of the present invention mainly includes two steps: first, the grid point cloud is adaptively partitioned by using the point cloud segmentation method, and the grid point cloud is divided into a patch set and a discrete independent point cloud set Then, the two types of point clouds are processed by point cloud segmentation filtering and multi-scale morphological filtering methods, and the grid points are classified into ground grid points and non-ground grid points. In the laser point cloud data, the smoothness of different regions is different, and the geometric characteristics of the adjacent point clouds in the flat region are similar, and the probability of the point cloud having the same category is also very high; while the geometric characteristics of the adjacent point cloud in the violently changing region are greatly different, and the point and point cloud have similar geometric characteristics. Neighboring points may come from different target surfaces. Therefore, different types of regions should be expressed with different primitives. Gentle regions are suitable to be expressed as regional patch primitives, and point primitives should be used for sharply changing regions, so as to better express the characteristics of point clouds in different regions and their proximity relations, more robustly describe the difference between ground and non-ground point clouds. In order to use different primitives in different regions, the point cloud segmentation based on smooth constraints is used to divide the point cloud into smooth and gentle regions and regions with severe elevation changes. The point cloud segmentation based on smoothness constraints uses the region growing method to merge adjacent point clouds with similar normal vectors and plane fitting residuals into the same patch. The segmentation process is as follows:

Use principal component analysis to estimate the normal vector and Gaussian curvature of the point cloud, and then extract the point cloud corresponding to the minimum value of Gaussian curvature from the unsegmented point cloud as the seed point, which is recorded as , to retrieve the seed point /> The N neighbor points of , and the points that meet the growth conditions are used as new seed points, and this method is used until no point can meet the conditions. There are two growth conditions, one is that the angle between the normal vector of the neighborhood point and the normal vector of the seed point is less than a certain threshold, and the other is that the residual error of the local fitting plane from the neighborhood point to the seed point is less than a certain threshold. If all the points are assigned to the corresponding patches, the segmentation ends; otherwise, continue to extract the seed points from the unsegmented point cloud. The three threshold values in the segmentation process are very important and directly affect the point cloud segmentation results. If the parameters are improper, the result is prone to over-segmentation or sub-segmentation, which affects the point cloud filtering results, especially the sub-segmentation problem. Among them, the values of the three parameters are closely related to the point cloud density and point cloud coordinate error. If the point density is high and the point cloud coordinate error is small, N can be increased appropriately to improve the anti-noise ability of the normal vector calculation, and the included angle and residual threshold should be appropriately reduced; if the point density is low and the point cloud coordinate error is large, N can be appropriately reduced , the included angle and the residual threshold can be appropriately increased to ensure that the flat area can have a better segmentation effect. After the point cloud filtering process, the point cloud is divided into ground point cloud and non-ground point cloud, and the non-ground point cloud is represented as two types: non-ground patch and non-ground discrete independent point cloud.

The invention obtains the prior position of the hoisting object by using the RTK+ encoder, improves the extraction speed of the point cloud in the target area, reduces the interference of buildings and construction equipment on the rapid identification of the hoisting object, and is based on the AABB tree base map graphic element retrieval structure The collision detection method is suitable for the actual application scene of the tower crane, which speeds up the retrieval process of the bounding box of the base map.

Based on the above embodiments, after obtaining the 3D point cloud base map of the construction site and the collision warning information of the hoisting object, the above information is visualized, which can assist the tower crane operator to perceive the 3D environment of the hoisting operation and improve the efficiency of the hoisting operation. The data interaction flow chart of the visualization system proposed by the embodiment of the present invention is shown in Figure 7. The tower crane collision warning visualization module is designed using the structure of "server + client", the tower crane side is used as the server, and the ground-side visualization software is used as the client. Realize functions such as viewing angle switching, safety threshold setting, and point cloud highlighting of potential collision areas. The overall implementation process includes:

The client and the server are connected through a LAN, configure parameters such as tower crane height and arm length, and load the tower crane model;

The client sends the map building command to the server, the tower crane rotates according to the map building requirements, the server collects the data required for map building, and calls the laser point cloud map and bounding box generation module to generate the point cloud base map and bounding box;

The client downloads the point cloud basemap and basemap bounding box from the server and loads them;

The client configures parameters such as the length of the lifting rope and sends them to the server. The client sends a collision detection command to the server. The server calls the real-time monitoring and collision information measurement module of the hoisting objects to obtain collision warning information: the coordinates of the starting point of the collision vector, and the end point of the collision vector Coordinates, the coordinates of the 8 corners of the bounding box;

After receiving the collision warning information, the client visualizes the bounding box and collision vector of the hoisting object, searches for potential collision areas in the base map point cloud according to the set warning range, and highlights them.

The invention improves the data transmission efficiency through the tower crane collision warning visualization module based on the "server + client" structure design, ensures the real-time performance of the collision warning information visualization, and can provide effective assistance for the remote operation of the tower crane.

The multi-integrated system-based tower crane collision warning system provided by the present invention is described below. The multi-integrated system-based tower crane collision warning system described below and the multi-integrated system-based tower crane collision warning method described above can be referred to in correspondence.

Fig. 8 is a schematic structural diagram of a tower crane collision warning system based on a multi-integrated system provided by an embodiment of the present invention. As shown in Fig. 8, it includes: an acquisition and reception module 81, a positioning and calculation module 82, an orientation registration module 83, and a segmentation calculation module 84 and collision detection module 85, wherein:

The collecting and receiving module 81 is used for collecting the laser radar point cloud data and the observation data of the integrated navigation system by the industrial computer on the side of the tower crane, and receiving the encoder observation data and the carrier phase difference positioning result of the hook receiver. The positioning solution module 82 is used for using the Combining the observation data of the navigation system, performing dual-antenna combined navigation, positioning and attitude determination, to obtain real-time position and attitude information; the directional registration module 83 is used to use the lidar point cloud data and the real-time position and attitude information, through geographical orientation and The point cloud registration obtains the point cloud base map of the construction site, and uses the upper and lower bottom surface contours combined with the bounding box extraction algorithm to extract the building outline bounding box corresponding to the point cloud base map of the construction site; the segmentation calculation module 84 is used for the laser The radar point cloud data is segmented and extracted or the coordinates of the hoisting object are calculated by using the encoder observation data, and the hoisting object bounding box is generated in combination with the carrier phase difference positioning result of the hook receiver; the collision detection module 85 is used to detect the building Collision detection is performed between the contour bounding box and the bounding box of the lifting object, and the spatial relationship between the lifting object and the building and collision warning information are obtained.

FIG. 9 illustrates a schematic diagram of the physical structure of an electronic device. As shown in FIG. 9, the electronic device may include: a processor (processor) 910, a communication interface (Communications Interface) 920, a memory (memory) 930, and a communication bus 940, Wherein, the processor 910 , the communication interface 920 , and the memory 930 communicate with each other through the communication bus 940 . The processor 910 can call the logic instructions in the memory 930 to execute the tower crane collision warning method based on the multi-integrated system, the method includes: collecting the laser radar point cloud data and the observation data of the integrated navigation system by the tower crane side industrial computer, and receiving the code The observation data of the receiver and the carrier phase difference positioning result of the hook receiver; the observation data of the integrated navigation system is used to perform the dual-antenna integrated navigation positioning and attitude determination to obtain real-time position and attitude information; the laser radar point cloud data and the described Real-time pose information, through geographic orientation and point cloud registration, obtain the point cloud base map of the construction site, and use the upper and lower bottom surface contours combined with the bounding box extraction algorithm to extract the building outline bounding box corresponding to the point cloud base map of the construction site; The lidar point cloud data is segmented and extracted or the coordinates of the hoisting objects are calculated by using the encoder observation data, and the hoisting object bounding box is generated in combination with the carrier phase difference positioning result of the hook receiver; the bounding box of the building outline Collision detection is performed with the bounding box of the hoisting object to obtain the spatial relationship between the hoisting object and the building and collision warning information.

In addition, the above-mentioned logic instructions in the memory 930 may be implemented in the form of software function units and be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the essence of the technical solution of the present invention or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk, and other media that can store program codes. .

On the other hand, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it is implemented to perform the multi-integrated system-based tower crane collision warning provided by the above methods The method includes: collecting lidar point cloud data and integrated navigation system observation data by the industrial computer on the side of the tower crane, and receiving the encoder observation data and the carrier phase difference positioning result of the hook receiver; using the integrated navigation system observation data, Perform dual-antenna combined navigation, positioning, and posture determination to obtain real-time pose information; use the lidar point cloud data and the real-time pose information to obtain the construction site point cloud base map through geographic orientation and point cloud registration, Using the upper and lower bottom contour combined with the bounding box extraction algorithm to extract the building outline bounding box corresponding to the point cloud base map of the construction site; segment and extract the lidar point cloud data or use the encoder observation data to calculate the hoisting object Coordinates, combined with the carrier phase difference positioning results of the hook receiver, generate the bounding box of the hoisting object; perform collision detection on the bounding box of the outline of the building and the bounding box of the hoisting object, and obtain the spatial relationship and collision between the hoisting object and the building Early warning information.

The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network elements. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. It can be understood and implemented by those skilled in the art without any creative efforts.

Through the above description of the implementations, those skilled in the art can clearly understand that each implementation can be implemented by means of software plus a necessary general-purpose hardware platform, and of course also by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of software products, and the computer software products can be stored in computer-readable storage media, such as ROM/RAM, magnetic Disks, CDs, etc., including several instructions to make a computer device (which may be a personal computer, server, or network device, etc.) execute the methods described in various embodiments or some parts of the embodiments.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still be Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent replacements are made to some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims (10)

1. A tower crane collision early warning method based on a multi-integrated system, characterized in that it comprises: The industrial computer on the side of the tower crane collects the laser radar point cloud data and the observation data of the integrated navigation system, as well as the observation data of the receiving encoder and the positioning result of the carrier phase difference of the hook receiver; Using the observation data of the integrated navigation system, performing dual-antenna integrated navigation, positioning, and attitude determination to obtain real-time attitude information; Using the lidar point cloud data and the real-time pose information, through geographical orientation and point cloud registration, the point cloud base map of the construction site is obtained, and the upper and lower bottom surface contours combined with the bounding box extraction algorithm are used to extract the point cloud of the construction site The building outline bounding box corresponding to the base map; Segmenting and extracting the lidar point cloud data or using the encoder observation data to calculate the coordinates of the hoisting object, and combining the carrier phase difference positioning results of the hook receiver to generate a hoisting object bounding box; Collision detection is performed on the bounding box of the outline of the building and the bounding box of the hoisting object to obtain the spatial relationship between the hoisting object and the building and collision warning information. 2. The tower crane collision warning method based on multi-integrated systems according to claim 1, characterized in that, using the observation data of the integrated navigation system, the dual-antenna integrated navigation, positioning, and posture determination are used to obtain real-time pose information, including : Carry out dynamic-to-dynamic carrier phase differential RTK correction to the main antenna and the slave antenna in the GNSS dual antenna to obtain a baseline vector, from which the heading angle and the pitch angle are obtained: in, Indicates heading angle, /> Indicates pitch angle, /> Indicates the projection of the baseline vector in the east direction, /> Indicates the projection of the baseline vector in the north direction, /> Indicates the projection of the baseline vector in the vertical direction; According to the output of inertial strapdown navigation SINS, the pitch angle and roll angle are derived, and the attitude update observation equation is obtained from the pitch angle and roll angle; The position update observation equation is obtained by the main antenna through RTK: in, Represents the observation value residual vector of the position update under the e system, /> Indicates the position of the lower inertial navigation system of the e system obtained by mechanical arrangement, /> Indicates the projection of the lever arm under the e system, /> Indicates the position of the center of the GNSS antenna in the e system, /> Indicates the error status of the position, /> Indicates misalignment angle, /> Indicates the observation noise; Based on the heading angle, the attitude update observation equation and the position update observation equation, perform integrated navigation settlement to obtain the real-time pose information; Using the real-time pose information, the acceleration and the zero offset of the gyro are fed back and corrected. 3. the tower crane collision early warning method based on multi-integrated system according to claim 2, is characterized in that, obtains pitch angle and roll angle according to the output derivation of inertial strapdown navigation SINS, obtains attitude update observation equation by pitch angle and roll angle ,include: The attitude update observation equation is obtained according to SINS, and the SINS accelerometer output is expressed by the specific force equation: in, is the accumulative output value, /> is the line-of-sight acceleration in the local horizontal coordinate system l, /> is the Coriolis force, is the centripetal acceleration to the ground, /> for gravity, /> Indicates the projection of the angular velocity of the e system relative to the i system under the l system, /> Indicates the projection of the angular velocity of the l system relative to the i system under the l system, /> Indicates the accumulative output value under the l system, /> Indicates the coordinate transformation matrix from the b system to the l system; in, Represents the coordinate transformation matrix from the e system to the l system, /> Represents the coordinate transformation matrix from the e system to the l system, /> represents the unit matrix, /> means /> antisymmetric matrix of angles, /> Indicates the coordinate transformation matrix from the b -system to the e -system obtained by mechanical arrangement at the current moment; attitude angle Differentiate to obtain the attitude update observation equation: in, a vector of residuals representing observations, /> represents the design matrix, /> Indicates the parameter to be estimated, /> represents the observation noise, /> , /> and /> Respectively represent the residual vector of heading angle, pitch angle and roll angle, /> , /> and /> Respectively represent the observed values of the attitude angle, /> , /> and /> Indicates the Euler angle obtained by inertial navigation attitude update, /> Medium /> , , /> Indicates the observation coefficient of the heading angle, /> , /> , /> Indicates the observation coefficient of the pitch angle, /> , /> , /> Indicates the observed coefficient of the roll angle, /> Indicates the pending status. 4. The tower crane collision warning method based on multiple integrated systems according to claim 1, characterized in that, using the laser radar point cloud data and the real-time pose information, through geographic orientation and point cloud registration, construction The site point cloud base map uses the upper and lower bottom surface contours combined with the bounding box extraction algorithm to extract the building outline bounding box corresponding to the construction site point cloud base map, including: Carrying out point cloud geographic orientation to the lidar point cloud data, converting the lidar point cloud data from a lidar coordinate system to a local horizontal coordinate system, and obtaining converted lidar point cloud data; Perform frame-to-frame matching on the converted lidar point cloud data to obtain an initial laser point cloud base map; Using the nonlinear least squares estimation LM algorithm to optimize the spatial placement parameters of the laser radar to obtain the point cloud base map of the construction site; Using a cloth simulation filter algorithm to divide the initial laser point cloud base map into ground points and non-ground points; The non-ground points belonging to the same non-ground object are clustered into a point cloud object through European clustering, using -The shape algorithm calculates the out-of-plane boundary of each point cloud object, and extracts the lower bottom surface point and top surface point of each point cloud object to obtain the building outline bounding box. 5. the tower crane collision early warning method based on multi-integrated system according to claim 4, is characterized in that, carry out point cloud geographic orientation to described laser radar point cloud data, described laser radar point cloud data is from laser radar coordinate system Convert to the local horizontal coordinate system to obtain the converted lidar point cloud data, including: Obtain any ground point in the lidar point cloud data Coordinates in the lidar coordinate system /> , the coordinates /> Convert to coordinates in the inertial navigation coordinate system /> : According to the position and attitude of the combination of dual-antenna GNSS and SINS, the coordinates Convert to global coordinate system coordinates: in, Indicates any ground point /> Coordinates in the lidar coordinate system, /> Indicates ground point /> Coordinates in the inertial navigation coordinate system, /> Indicates any ground point /> Coordinates in the WGS84 earth-centered earth-fixed coordinate system, /> Indicates the position of the origin of the lidar coordinate system in the inertial navigation coordinate system, /> Represents the rotation matrix for transforming the lidar coordinate system to the inertial navigation coordinate system, /> Indicates the position of the origin of the inertial navigation coordinate system in the WGS84 earth-centered ground-fixed space Cartesian coordinate system, /> Indicates the transformation matrix of the inertial navigation coordinate system to the Cartesian coordinate system in the WGS84 earth-centered and ground-fixed space; in, is the rotation matrix from the inertial navigation coordinate system to the local horizontal coordinate system calculated according to the three attitude angles output by the inertial navigation system, /> is the rotation matrix from the local horizontal coordinate system to the WGS84 earth-centered ground-fixed space Cartesian coordinate system calculated according to the geodetic coordinates of the origin of the inertial navigation coordinate system; Will substitute /> Obtain the tower crane laser scanning positioning equation: Calculate the ground point according to the tower crane laser scanning positioning equation Coordinates in the local horizontal coordinate system: in, is the coordinates of the origin of the local horizontal coordinate system in the WGS84 coordinate system, /> for The rotation matrix of the WGS84 earth-centered earth-fixed space Cartesian coordinate system to the local horizontal coordinate system. 6. The tower crane collision early warning method based on multi-integrated systems according to claim 5, wherein the LM algorithm is used to optimize the laser radar space placement parameters to obtain the point cloud base map of the construction site, including: Determine for any ground point Perform two laser scans to obtain two laser point cloud positioning equations: in, , /> , /> and /> , /> , /> is in two scans /> Measured values in the lidar coordinate system, , /> , /> and /> is the first scan /> The rotation matrix composed of the position and attitude of the inertial navigation at all times, , /> , /> and /> is the second scan /> The rotation matrix composed of the position and attitude of the inertial navigation at all times; the first scan The measured value in the lidar coordinate system minus the second scan /> The measured value in the lidar coordinate system is: Will The three translation parameters represented by , and /> The rotation matrix represented by three angles that rotate around the X-axis, Y-axis, and Z-axis respectively, converted to /> , where /> respectively with /> The three translation parameters represented correspond to, /> respectively with /> The rotation matrices represented by the three angles that rotate around the X-axis, Y-axis, and Z-axis respectively; Through the LM algorithm for Carry out optimization and solve, iteratively obtain the mathematical model , where /> is a set of nonlinear equations, /> Indicates any set of numbers, /> Indicates the total number of groups, based on the mathematical model /> Solve for the minimum value to get the optimal solution /> get: . 7. The multi-integrated system-based tower crane collision warning method according to claim 1, characterized in that, the lidar point cloud data is segmented and extracted or the encoder observation data is used to calculate the hoisting object coordinates, combined with the Based on the carrier phase difference positioning results of the above-mentioned hook receiver, the bounding box of the hoisting object is generated, including: Project the lidar point cloud data to the local horizontal coordinate system, convert the carrier phase difference positioning result of the hook receiver from the WGS84 coordinate system to the local horizontal coordinate system to obtain the hook position information ; Using point cloud filtering to divide the lidar point cloud data into ground point clouds and non-ground point clouds, extracting non-ground patch point clouds from the non-ground point clouds, and extracting target areas as hoisting objects; Carrying out connectivity analysis to the point cloud of the non-ground patch, merging adjacent non-ground patches, as an undetermined target area, and eliminating the non-building target area according to the geometric size of the building and the elevation difference between the building boundary and the ground; Using the hook location information Determine the distance as described in /> The area within the preset distance range is the target of the hoisting object, and the rest of the area is the target of the building; or, if it is confirmed that the GNSS signal at the hook is blocked, the moving distance of the trolley and the descending distance of the hook obtained from the encoder measurement, and the dual The pose information obtained by the combination of antenna GNSS and SINS is used to calculate the position of the hook /> , according to /> Screening is carried out to obtain the target of the hoisting object; Projecting the three-dimensional outer bounding box of the lifting object target on a two-dimensional plane to generate a two-dimensional top view projection, rotating the two-dimensional top view projection around the projection center, and using the circumscribed rectangle of the rotation range as the lifting object bounding box. 8. the tower crane collision early warning method based on multi-integrated system according to claim 7, is characterized in that, adopts point cloud filtering to divide described lidar point cloud data into ground point cloud and non-ground point cloud, from described non-ground point cloud Extract the non-ground patch point cloud from the ground point cloud, and extract the target area as a hoisting object, including: Use the virtual grid to process the lidar point cloud data, determine the preset grid size, traverse each grid, determine the lowest point in each grid as each grid point, and the remaining points in each grid The point cloud is other point clouds, and the grid without point cloud is supplemented by virtual point interpolation to obtain the grid point cloud; performing adaptive partitioning on the grid point cloud by point cloud segmentation to obtain a facet set and a discrete independent point cloud set; Process the patch set and the discrete independent point cloud set respectively by point cloud segmentation filtering and multi-scale morphological filtering to obtain a ground grid point cloud and a non-ground grid point cloud; Segmenting the ground grid point cloud and the non-ground grid point cloud into the ground point cloud and the non-ground point cloud, respectively, using smooth constraint point cloud segmentation, the non-ground point cloud including non-ground patches and non-ground discrete independent point clouds. 9. The multi-integrated system-based tower crane collision warning method according to claim 1, wherein collision detection is performed on the building outline bounding box and the hoisting object bounding box to obtain the space between the hoisting object and the building Relationship and collision warning information, including: An axis-parallel bounding box AABB tree is used to index the building outline bounding box; Perform collision detection on the building outline bounding box and the hoisting object bounding box based on the AABB tree, traverse the AABB tree of the building outline bounding box and the hoisting object bounding box, and obtain the distance from the hoisting object bounding box The nearest building outline bounding box basemap primitive; The distance vectors from each corner point in the bounding box of the lifting object to the base map primitive of the bounding box of the building outline are calculated in turn, and the minimum collision distance is determined by the distance vector. 10. The multi-integrated system-based tower crane collision early warning method according to claim 1, characterized in that, the collision detection is carried out to the outline bounding box of the building and the optimal outer bounding box of the hoisting object, and the hoisting object and the building are obtained. After the spatial relationship of objects and collision warning information, it also includes: Send the cloud base map of the construction site, the building outline bounding box, the hoisting object bounding box and the collision warning information to the ground client by the tower base side server; The ground client uses visualization software to display the three-dimensional environment of the hoisting operation, collision distance information and potential collision targets.