A crane trolley active anti-collision method and system

Abstract

The application provides a crane trolley active anti-collision method and system, and relates to the technical field of crane anti-collision. In view of various collision risks that may occur in the operation process of the crane, the active anti-collision method and system capture the surrounding environment information in real time through high-precision and non-contact laser scanning technology; and through coordinate conversion, downsampling, clustering analysis and other preprocessing operations on the obtained point cloud data, the accurate identification of the obstacle contour size and coordinates is realized. On this basis, the active anti-collision method and system can intelligently set and dynamically adjust the sizes of protection areas of different levels in combination with the actual running state of the walking mechanism of the crane, so that the safety level and overall work efficiency in the operation process of the crane trolley can be ensured to be rapidly reacted under any working condition, and the mutual collision between machines, between machines and people or between machines and obstacles can be effectively prevented.

Description

Technical Field

[0001] This invention relates to the field of crane collision avoidance technology, and in particular to an active collision avoidance method and system for crane trolleys. Background Technology

[0002] Early collision avoidance systems mostly used mechanical contact or electromagnetic induction limit switches, triggering a braking mechanism to prevent collisions when the vehicle reached a predetermined position. With technological advancements, lidar and ultrasonic rangefinders began to be used in such applications. These systems determine the presence of obstacles and their distance from the vehicle by emitting signals and receiving reflected echoes. However, these technologies still fall short in complex and changing operating conditions, especially when processing dynamically changing environmental information in three-dimensional space. Due to hardware limitations and external interference, false alarms or missed alarms are common, indicating insufficient intelligence and an inability to flexibly adjust the protection zone to adapt to different operating conditions.

[0003] The radar-based anti-collision monitoring device for ship unloaders is one of the most similar technical solutions currently available. This device includes a millimeter-wave radar mounted on the ship unloader's gantry legs. It enables long-range, high-precision non-contact ranging and possesses strong anti-interference capabilities, allowing it to operate normally in adverse weather conditions. The radar transmits the acquired data to the main control PLC module and sub-control PLC modules, achieving high-speed signal transmission and processing via fiber optic connections. The main control PLC module controls the operation of the traveling motors via a frequency converter. Based on obstacle information identified by the radar, it sets warning zones, deceleration zones, and stop zones, providing intuitive prompts to operators using warning indicator lights, deceleration indicator lights, and stop indicator lights. It also automatically adjusts the speed of the ship unloader trolley or stops it according to the actual situation, thus achieving a certain degree of active collision avoidance.

[0004] However, while the aforementioned radar-based solutions improve detection range and accuracy, there is still room for improvement in areas such as precise contour recognition and intelligent dynamic adjustment of the protected area size in complex three-dimensional environments. To address these issues, this invention proposes a more advanced active collision avoidance method and system for crane trolleys. It utilizes two-dimensional laser scanning technology to capture detailed information about the surrounding environment and optimizes the data processing flow through algorithms such as coordinate transformation, downsampling, and cluster analysis to achieve accurate identification of obstacle contour dimensions and coordinates. Simultaneously, by combining the actual operating parameters of the crane's traveling mechanism, such as speed and direction, the system can intelligently set and adjust different warning levels and protected area sizes in a timely manner, ensuring that the crane can react quickly and accurately under various operating conditions, effectively avoiding various collision accidents, thereby significantly improving the safety performance and overall work efficiency of the crane trolley. Summary of the Invention

[0005] Purpose of the invention: This invention proposes an active collision avoidance method and system for crane trolleys. Addressing various collision risks that may occur during crane operation, it effectively prevents collisions between machines, between machines and personnel, or between machines and obstacles through high-precision point cloud data processing technology and dynamic safety zone setting, ensuring safe and efficient crane operation and thus effectively solving the aforementioned problems existing in the prior art.

[0006] Firstly, a method for active collision avoidance of crane trolleys is proposed, with the following steps:

[0007] S1. Install one single-line lidar sensor on each of the crane trolleys. The detection range of the multiple single-line lidar sensors can cover the three-dimensional spatial environment on both sides of the crane trolley's movement trajectory.

[0008] S2. The original point cloud data set of obstacles in the current three-dimensional spatial environment is scanned by multiple single-line lidar sensors;

[0009] S3. Perform a downsampling operation on the original point cloud data set to obtain a downsampled point cloud data set;

[0010] S4. Perform a clustering operation on the downsampled point cloud data set to cluster it into different point cloud clusters;

[0011] S5. Based on the point cloud cluster, extract the outline size of the current obstacle;

[0012] S6. Based on the current obstacle's outline dimensions, establish an active collision avoidance model;

[0013] S7. During the operation of the crane trolley, the spatial relationship between the current working space of the crane trolley and the active anti-collision model is determined in real time, and corresponding anti-collision measures are taken.

[0014] In a further embodiment of the first aspect, a set of original point cloud data of obstacles in the current three-dimensional spatial environment is scanned by a plurality of said single-line lidar sensors. The set of original point cloud data is in a polar coordinate system, and the original point cloud data at time t is represented as:

[0015]

[0016] In the formula, point i Represents the i-th sampling line; ρ i θ represents the distance value returned by the sampling point i. i This indicates the sampling angle corresponding to the current distance value; This indicates the maximum scanning angle.

[0017] In a further embodiment of the first aspect, step S2 further includes converting the original point cloud data set in the polar coordinate system into the original point cloud data set in the rectangular coordinate system, wherein the coordinate expression in the rectangular coordinate system at time t is:

[0018] x point =ρ i cosθ i

[0019] y point =ρ i sinθ i

[0020] In the formula, x point The x-coordinate in a rectangular coordinate system; y point This represents the ordinate in a rectangular coordinate system.

[0021] In a further embodiment of the first aspect, a downsampling operation is performed on the original point cloud data set to obtain a downsampled point cloud data set, including:

[0022] A 3D voxel grid is formed based on the given original point cloud data set. The center point of each grid is used to approximate all points within the grid. The formula for calculating the grid center point is:

[0023]

[0024] In the formula, x, y, and z are the coordinate values ​​of the center point in each dimension; N is the number of points contained in the voxel cell; x i y i z i Let be the coordinates of the i-th point in each dimension;

[0025] By retaining the center point of each grid, a set of downsampled point cloud data is obtained.

[0026] In a further embodiment of the first aspect, a clustering operation is performed on the downsampled point cloud data set to cluster it into different point cloud clusters, including:

[0027] Randomly select an object point from the downsampled point cloud data set, and query whether the density of all reachable points within the neighborhood radius of the object point is greater than the minimum number of contained points. If it is greater, create a cluster with the object point as the core point.

[0028] Iterate through all objects reachable from the core point density, and merge the clusters reachable from the core point density;

[0029] Once a cluster is completed, the next point will be selected to start clustering, and the clustering will end when no new object points are added to any cluster.

[0030] In a further embodiment of the first aspect, based on the point cloud cluster, the contour dimensions of the current obstacle are extracted, including:

[0031] The minimum bounding rectangle enclosing the point cloud cluster is constructed by subtracting the maximum and minimum coordinates of the point cloud clusters along the x and y axes to obtain the length and width of the obstacle outline.

[0032] length=|x max -x min |

[0033] width = |y max -y min |

[0034] In the formula, length and width are the length and width of the minimum bounding rectangle, respectively;

[0035] The coordinates of the four vertices of the minimum bounding rectangle of the obstacle point cloud cluster are (x, y ... min y min ), (x max y min ), (x min y max ), (x max y max ).

[0036] In a further embodiment of the first aspect, step S6 further includes:

[0037] Based on the minimum bounding rectangle of the point cloud cluster, three levels of early warning areas are sequentially divided, namely the lidar visibility area, the deceleration warning area, and the braking alarm area.

[0038] During the actual operation of the crane trolley, the sizes of the laser radar's visible area, deceleration warning zone, and braking alarm zone will be dynamically adjusted according to the speed, satisfying the following relationship:

[0039]

[0040]

[0041] In the formula, r1 is the safe distance of the vehicle's braking warning zone; r2 is the safe distance of the vehicle's deceleration warning zone; r 10 The safe distance for the braking warning zone when the vehicle is stationary; r 20 v is the safe distance of the deceleration warning zone when the trolley is stationary; v is the speed of the trolley; t is the braking time of the trolley; v2 is the fixed intermediate speed to which the trolley decelerates when the deceleration command is executed; t2 is the braking time corresponding to the fixed intermediate speed.

[0042] In a further embodiment of the first aspect, step S7 further includes:

[0043] S7-1. Select point cloud clusters that enter the visible area of ​​the lidar;

[0044] S7-2. Determine whether the point cloud clusters selected in step S7-1 are included in the braking alarm area:

[0045] If so, a stop signal is output to control the crane trolley to stop immediately;

[0046] If not, proceed to step S7-3;

[0047] S7-3. Determine whether the point cloud clusters selected in step S7-1 are included in the deceleration warning zone:

[0048] If so, a deceleration signal is output to control the crane trolley to decelerate;

[0049] If not, a normal travel signal is output, and the crane trolley maintains its current operating speed and direction.

[0050] As a second aspect of the present invention, an active collision avoidance system for crane trolleys is proposed, the collision avoidance system comprising a sensing module, a decision-making module, and an execution module.

[0051] The perception module includes several single-line lidar sensors, which are arranged on the crane trolley. The detection range of the multiple single-line lidar sensors can cover the three-dimensional spatial environment on both sides of the crane trolley's movement trajectory. The single-line lidar sensors scan the original point cloud data set of obstacles in the current three-dimensional spatial environment.

[0052] The decision-making module includes a first point cloud data processing unit, a second point cloud data processing unit, a contour extraction unit, and a collision avoidance model construction unit. The first point cloud data processing unit performs a downsampling operation on the original point cloud data set to obtain a downsampled point cloud data set. The second point cloud data processing unit performs a clustering operation on the downsampled point cloud data set to cluster it into different point cloud clusters. The contour extraction unit extracts the contour dimensions of the current obstacle based on the point cloud clusters. The collision avoidance model construction unit establishes an active collision avoidance model based on the contour dimensions of the current obstacle.

[0053] The execution module is used to determine the spatial relationship between the current working space of the crane trolley and the active anti-collision model in real time during the operation of the crane trolley, and to take corresponding anti-collision measures.

[0054] As a third aspect of the invention, a computer-readable storage medium is provided, which stores at least one executable instruction that, when executed on an electronic device, causes the electronic device to perform the crane trolley active collision avoidance method as described in the first aspect.

[0055] Beneficial Effects: This invention proposes an active collision avoidance method and system for crane trolleys. The system uses high-precision, non-contact laser scanning technology to capture surrounding environmental information in real time. Advanced data processing algorithms are then used to preprocess the acquired point cloud data, performing coordinate transformation, downsampling, and cluster analysis to accurately identify the outline size and coordinates of obstacles. Based on this, the system can intelligently set and dynamically adjust the size of different levels of protection zones according to the actual operating status of the crane's traveling mechanism, ensuring a rapid response under any working condition and effectively preventing various collision accidents. This significantly improves the safety level and overall work efficiency of the crane trolley during operation. Attached Figure Description

[0056] Figure 1 This is a schematic diagram of the installation of a single-line lidar for a crane trolley.

[0057] Figure 2 This is a schematic diagram of the hardware architecture of a crane trolley active collision avoidance system.

[0058] Figure 3 This is a flowchart of the control algorithm for the active collision avoidance system of a crane trolley.

[0059] Figure 4 The flowchart shows the DBSCAN algorithm based on the improved KD-Tree.

[0060] Figure 5 This is a schematic diagram of a three-level early warning active safety collision avoidance model.

[0061] Figure 6 This is a flowchart of the active collision avoidance algorithm for crane trolleys. Detailed Implementation

[0062] In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention can be practiced without one or more of these details. In other instances, certain technical features well-known in the art have not been described in order to avoid obscuring the invention.

[0063] Research has revealed the following drawbacks in existing crane trolley anti-collision systems:

[0064] (1) Limited detection range: Traditional collision avoidance solutions, such as the combination of limit switches and single-point laser sensors, as well as ultrasonic ranging solutions, can only detect obstacles in a straight line or a small area due to their working principles and hardware performance limitations. They cannot fully cover the three-dimensional spatial environment. This limitation may lead to a large number of potential collision risks going undetected in the complex and ever-changing working environment of ports, thus causing safety accidents.

[0065] (2) Significant impact on equipment: Conventional mechanical safety collision avoidance systems often take emergency stop measures when they detect obstacles, which will cause the crane to bear a large impact force, thus affecting its service life and stability.

[0066] (3) Slow response speed and low intelligence: Existing collision avoidance systems generally have a slow response speed and cannot quickly adapt to changing working conditions. In addition, they often cannot intelligently adjust the size of the protection zone according to the actual operating status of the crane, that is, they do not have sufficient intelligence to adapt to the collision avoidance needs under different conditions.

[0067] This invention aims to solve the problem of collisions that may occur between cranes, between cranes and personnel, or between cranes and obstacles during crane operation. Traditional collision avoidance solutions, such as combinations of limit switches and single-point laser sensors, and ultrasonic ranging solutions, exhibit significant shortcomings in practical applications: limited detection range, unable to fully cover the complex and ever-changing three-dimensional spatial environment; detection accuracy limited by hardware performance and environmental interference, prone to false alarms or missed alarms; slow response speed, unable to make timely and accurate judgments and effective preventive measures; and low level of intelligence, lacking the ability to dynamically adjust the protection area according to actual conditions. Therefore, to solve the above technical problems, this invention relates to an active collision avoidance method and system for crane trolleys. Through efficient point cloud data preprocessing technology, including coordinate transformation, downsampling, and clustering algorithms, effective obstacle information is extracted in real time to construct an active collision avoidance model suitable for cranes, enabling real-time dynamic adjustment of the protection area and achieving a higher level of collision warning and control. The technical solution of this invention is specifically described below through several embodiments.

[0068] Figure 1 This diagram illustrates the installation of a single-line lidar sensor on a crane trolley. One single-line lidar sensor is installed on each of the crane trolley's traveling trolleys to ensure coverage of the three-dimensional spatial environment on both sides of the trolley's movement trajectory.

[0069] Figure 2This is the hardware architecture of a crane trolley active collision avoidance system. The system mainly consists of a perception module, a decision-making module, and an execution module. The perception module primarily comprises four single-line LiDAR sensors, used to scan and detect obstacles during crane operation. The decision-making module acquires the point cloud information from the LiDAR sensors, processes it through algorithms such as coordinate transformation, noise reduction, and clustering to obtain effective obstacle information, forms an active collision avoidance model, and sends the final result to the PLC. The execution module mainly receives the results from the collision avoidance algorithm and controls the traveling mechanism to decelerate or stop.

[0070] Figure 3 This is a flowchart of the control algorithm for the active collision avoidance system of a crane trolley. The control algorithm is mainly used for transforming sensor point cloud coordinates, downsampling, clustering, extracting obstacle information, establishing collision avoidance models, and outputting early warning signals.

[0071] The implementation steps of the control algorithm are as follows:

[0072] (1) Coordinate transformation

[0073] LiDAR typically acquires point cloud data by scanning and discretely sampling within a local plane. Polar coordinates are the most suitable coordinate system for describing this data acquisition method. However, since subsequent clustering algorithms require data input in a Cartesian coordinate system, the first preprocessing step for the LiDAR point cloud data is to convert the polar coordinates to Cartesian coordinates. Taking the LiDAR sensor itself as the origin of the polar coordinate system, the point cloud data set at time t can be represented as:

[0074]

[0075] In the formula, point i Represents the i-th sampling line; ρ i θ represents the distance value returned by the sampling point i. i This indicates the sampling angle corresponding to the current distance value; This indicates the maximum scanning angle, which is 270°.

[0076] The coordinate expression in the rectangular coordinate system at time t is:

[0077] x point =ρ i cosθ i

[0078] y point =ρ i sinθ i

[0079] In the formula, x point The x-coordinate in a rectangular coordinate system; y point This represents the ordinate in a rectangular coordinate system.

[0080] (2) Point cloud downsampling

[0081] Due to the large amount of raw point cloud data collected, directly processing the raw point cloud computing is costly and inefficient. In order to save algorithm time and improve the efficiency of system algorithm operation and the real-time nature of information acquisition, this system adopts the voxelized grid method to downsample redundant point cloud information, that is, to reduce the amount of point cloud dataset as much as possible without affecting the system function.

[0082] The voxelization mesh method generates a 3D voxel grid based on a given point cloud, and approximates all points in the grid by using the center point of all points within each grid. The formula for calculating the grid center point is:

[0083]

[0084] In the formula, x, y, and z are the coordinate values ​​of the center point in each dimension, N is the number of points contained in the voxel cell, and x i y i z i Let be the coordinates of the i-th point in each dimension.

[0085] (3) Point cloud clustering

[0086] This system uses the DBSCAN clustering algorithm, which is a density-based clustering algorithm that does not require pre-specifying the number of clusters and can find clusters of arbitrary shapes in noisy data, and has good anti-interference ability.

[0087] The core idea of ​​the DBSCAN algorithm is as follows: Given a dataset D, randomly select an object point X. Query whether all density-reachable points within X's Eps neighborhood radius are greater than MinPts. If they are, create a cluster with X as the core point. Then, iteratively aggregate all objects density-reachable from the core point. This process merges clusters of density-reachable core points. During algorithm execution, some density-reachable objects from the core point may be queried repeatedly. After one cluster is completed, the next point is selected for clustering, continuing until no new points are added to any cluster, at which point the clustering ends.

[0088] The DBSCAN algorithm requires frequent data searches, necessitating an efficient data organization method for retrieval. To meet the real-time requirements of collision avoidance systems, a KD-Tree is used to establish topological relationships between irregular point cloud data from LiDAR, enabling rapid identification of nearby points or clusters and fulfilling the real-time clustering requirements. A KD-Tree is a multi-dimensional data structure similar to a binary tree, providing a topological structure for k-dimensional data, facilitating efficient data lookup and search. The dataset is constructed using a KD-Tree. Before clustering, a neighborhood set for each data point is generated. Each data point is traversed, and a nearest neighbor search is performed using the constructed KD-Tree to find the neighborhood sets of all data objects, including core points, boundary points, and noise points. The identified core points are then used in subsequent clustering steps. Improving the DBSCAN algorithm using KD-Trees reduces the time complexity from O(n^2) to O(n^2). 2 The time complexity is reduced to O(nlog₂n), and the algorithm flow is as follows: Figure 4 As shown.

[0089] (4) Obstacle information extraction

[0090] After clustering the point cloud using the DBSCAN algorithm based on KD-Tree, the LiDAR point cloud information is divided into multiple point cloud clusters. The system then calculates obstacle information, extracting information such as the obstacle's outline size and coordinates to facilitate the subsequent establishment of an active collision avoidance model.

[0091] For the extraction of obstacle size information, the length and width of the obstacle outline are obtained by subtracting the maximum and minimum values ​​of the point cloud coordinates in the x and y directions, and then constructing the minimum bounding rectangle that surrounds the point cloud cluster.

[0092] length=|x max -x min |

[0093] width = |y max -y min |

[0094] In the formula, length and width are the length and width of the minimum bounding rectangle, respectively.

[0095] The coordinates of the four vertices of the minimum bounding rectangle of the obstacle point cloud cluster are: (x...) min y min ), (x max y min ), (x min y max ), (x max y max ).

[0096] (5) Establishment of active collision avoidance model

[0097] Based on the actual operating parameters of the crane's traveling mechanism, such as speed and direction, a flexible and adjustable three-level early warning active safety collision avoidance model is constructed, such as... Figure 5 As shown, the red area is the first-level braking alarm zone, the yellow area is the second-level deceleration warning zone, and the green area is the third-level lidar visible area. Among them, x_range1 and x_range2 are the track widths of the detection area, which can be set according to the actual situation. r1 is the safety distance of the trolley braking alarm zone, r2 is the safety distance of the trolley deceleration warning zone, and r3 is the radar visible range of the trolley.

[0098] Based on the principles of kinematic analysis during braking, the safe distances r1 and r2 of a large vehicle during its movement are related to its speed, as shown in the following formula: the faster the vehicle travels, the larger the collision protection zone, and vice versa. Therefore, in actual operation, the size of the protection zone of the large vehicle will be dynamically adjusted according to its speed, thus better adapting to actual operating conditions.

[0099]

[0100] In the formula, r 10 r is the safe distance for the braking warning zone when the vehicle is stationary. 20 v is the safe distance of the deceleration warning zone when the trolley is stationary, v is the speed of the trolley, t is the braking time of the trolley, v2 is the fixed intermediate speed to which the trolley decelerates when the deceleration command is executed, and t2 is the braking time corresponding to the fixed intermediate speed.

[0101] When obstacle point cloud clusters enter different warning zones, the system intelligently adjusts its output control signals, including deceleration and stop signals, to ensure timely implementation of appropriate collision avoidance measures in dangerous situations. Its algorithmic logic architecture is as follows: Figure 6 As shown.

[0102] Based on the above embodiments, the present invention significantly improves the technical defects of the crane trolley operation safety protection system by introducing two-dimensional laser scanning technology and advanced point cloud data processing algorithms.

[0103] First, compared with the combination of limit switches and single-point laser sensors and ultrasonic ranging solutions, the collision avoidance system of the present invention has all-round detection capabilities, which can cover complex three-dimensional spatial environments, effectively solving the limitation of existing technologies in detection range and avoiding collision accidents caused by blind spots.

[0104] Secondly, the present invention achieves smooth deceleration control, which can effectively reduce the impact of emergency braking on the equipment, help extend the service life of the crane, and greatly reduce the risk of collision caused by human error or sudden situations.

[0105] Finally, this invention utilizes advanced point cloud data processing technologies (such as coordinate transformation, downsampling, and cluster analysis) to quickly and accurately extract key information about obstacles. Combined with the operating parameters of the crane's traveling mechanism, it constructs an active collision avoidance model with real-time performance and adaptability. This improvement not only enhances the overall system response speed but also makes the collision avoidance strategy more intelligent, meeting the safety protection needs under different working conditions.

[0106] In summary, this invention, with its superior detection performance, rapid and accurate response capabilities, and intelligent collision avoidance strategy, not only improves the safety of crane trolley operation but also optimizes overall work efficiency, providing a strong guarantee for the safe and efficient operation of ports.

[0107] As described above, although the invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the invention itself. Various changes in form and detail may be made without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. A method for active collision avoidance of a crane trolley, characterized in that, Includes the following steps: S1. Install one single-line lidar sensor on each of the crane trolleys. The detection range of the multiple single-line lidar sensors can cover the three-dimensional spatial environment on both sides of the crane trolley's movement trajectory. S2. The original point cloud data set of obstacles in the current three-dimensional spatial environment is scanned by multiple single-line lidar sensors; S3. Perform a downsampling operation on the original point cloud data set to obtain a downsampled point cloud data set, specifically including: A three-dimensional voxel grid is formed based on a given set of original point cloud data, and all points in the grid are approximated by the center point of all points in each grid. By retaining the center point of each grid cell, a set of downsampled point cloud data is obtained; S4. Perform a clustering operation on the downsampled point cloud data set to cluster it into different point cloud clusters; S5. Based on the point cloud cluster, extract the contour dimensions of the current obstacle, specifically including: The minimum bounding rectangle enclosing the point cloud cluster is constructed by subtracting the maximum and minimum coordinates of the point cloud clusters along the x and y axes to obtain the length and width of the obstacle outline. In the formula, length and width are the length and width of the minimum bounding rectangle, respectively; The coordinates of the four vertices of the minimum bounding rectangle of the obstacle point cloud cluster are (x, y ... min y min ), (x max y min ), (x min y max ), (x max y max ); S6. Based on the current obstacle's outline dimensions, establish an active collision avoidance model, specifically including: Based on the minimum bounding rectangle of the point cloud cluster, three levels of early warning areas are sequentially divided, namely the lidar visibility area, the deceleration warning area, and the braking alarm area. During the actual operation of the crane trolley, the sizes of the laser radar's visible area, deceleration warning zone, and braking alarm zone will be dynamically adjusted according to the speed, satisfying the following relationship: In the formula, r1 is the safe distance of the vehicle's braking warning zone; r2 is the safe distance of the vehicle's deceleration warning zone; r 10 The safe distance for the braking warning zone when the vehicle is stationary; r 20 v is the safe distance of the deceleration warning zone when the trolley is stationary; v is the speed of the trolley; t is the braking time of the trolley; v2 is the fixed intermediate speed to which the trolley decelerates when the deceleration command is executed; t2 is the braking time corresponding to the fixed intermediate speed. S7. During the operation of the crane trolley, the spatial relationship between the current working space of the crane trolley and the active anti-collision model is determined in real time, and corresponding anti-collision measures are taken, specifically including: S7-1. Select point cloud clusters that enter the visible area of ​​the lidar; S7-2. Determine whether the point cloud clusters selected in step S7-1 are included in the braking alarm area: If so, a stop signal is output to control the crane trolley to stop immediately; If not, proceed to step S7-3; S7-3. Determine whether the point cloud clusters selected in step S7-1 are included in the deceleration warning zone: If so, a deceleration signal is output to control the crane trolley to decelerate; If not, a normal travel signal is output, and the crane trolley maintains its current operating speed and direction.

2. The active collision avoidance method for crane trolleys according to claim 1, characterized in that, The original point cloud data set of obstacles in the current three-dimensional spatial environment is scanned by multiple single-line lidar sensors. The original point cloud data set is in polar coordinates, and the original point cloud data at time t is represented as: In the formula, Indicates the i-th sampling line; This represents the distance value returned by the sampling point at angle i; This indicates the sampling angle corresponding to the current distance value; This indicates the maximum scanning angle.

3. The active collision avoidance method for crane trolleys according to claim 2, characterized in that, Step S2 also includes converting the original point cloud data set in polar coordinates to the original point cloud data set in rectangular coordinates. The coordinate expression in rectangular coordinates at time t is: In the formula, Represents the x-coordinate in a rectangular coordinate system; This represents the ordinate in a rectangular coordinate system.

4. The active collision avoidance method for crane trolleys according to claim 1, characterized in that, The formula for calculating the center point of a grid is: In the formula, x, y, and z are the coordinate values ​​of the center point in each dimension; N is the number of points contained in the voxel cell; x i y i z i Let be the coordinates of the i-th point in each dimension.

5. The active collision avoidance method for crane trolleys according to claim 4, characterized in that, Perform clustering operations on the downsampled point cloud data set to cluster it into different point cloud clusters, including: Randomly select an object point from the downsampled point cloud data set, and query whether the density of all reachable points within the neighborhood radius of the object point is greater than the minimum number of contained points. If it is greater, create a cluster with the object point as the core point. Iterate through all objects reachable from the core point density, and merge the clusters reachable from the core point density; Once a cluster is completed, the next point will be selected to start clustering, and the clustering will end when no new object points are added to any cluster.

6. A crane trolley active collision avoidance system, used to execute the crane trolley active collision avoidance method according to any one of claims 1 to 5, characterized in that, include: Sensing module; The perception module includes several single-line lidar sensors, which are arranged on the crane trolley. The detection range of the multiple single-line lidar sensors can cover the three-dimensional spatial environment on both sides of the crane trolley's movement trajectory. The single-line lidar sensors scan the original point cloud data set of obstacles in the current three-dimensional spatial environment. The decision module includes a first point cloud data processing unit, a second point cloud data processing unit, a contour extraction unit, and a collision avoidance model construction unit. The first point cloud data processing unit performs a downsampling operation on the original point cloud data set to obtain a downsampled point cloud data set. The second point cloud data processing unit performs a clustering operation on the downsampled point cloud data set to cluster it into different point cloud clusters. The contour extraction unit extracts the contour dimensions of the current obstacle based on the point cloud clusters. The collision avoidance model building unit establishes an active collision avoidance model based on the current obstacle's outline dimensions; Execution module; The execution module is used to determine the spatial relationship between the current working space of the crane trolley and the active anti-collision model in real time during the operation of the crane trolley, and to take corresponding anti-collision measures.

7. A computer-readable storage medium, characterized in that, The storage medium stores at least one executable instruction, which, when executed on an electronic device, causes the electronic device to perform the crane trolley active collision avoidance method as described in any one of claims 1 to 5.

Images