Control system and control method of intelligent transfer robot for construction site

Abstract

The invention provides a control system and a control method of an intelligent transfer robot for a construction site, and relates to the technical field of BIM (Building Information Modeling). The method comprises the following steps: performing spatial registration fusion processing on real-time spatial perception data and pre-stored initial building structure information to generate a real-time grid map; generating an initial carrying path by adopting an ant colony path search method fused with a space bionic memory mechanism; performing disturbance sensitivity iterative optimization processing on the initial carrying path according to the real-time disturbance characteristics of the path nodes to generate a target carrying path; and then a speed control sequence in the robot running process is determined according to the spatial relative relation between the target carrying path and the obstacle information, a robot path control instruction is generated, and the running safety and the working efficiency of the robot in the dynamic complex construction environment are improved.

Description

Technical Field

[0001] This invention relates to the field of BIM technology, and in particular to a control system and control method for intelligent handling robots used on construction sites. Background Technology

[0002] With the rapid development of intelligent robot technology, robots are increasingly being widely used in the construction industry to perform tasks such as material transportation, on-site handling, and equipment installation assistance, thereby improving construction efficiency and reducing labor costs. However, construction site environments are typically complex and changeable, with intricate spatial structures, a coexistence of static structures and dynamic obstacles, and constant environmental changes caused by the progress of construction. This poses significant challenges to the precise path planning and safe operation of robots.

[0003] Currently, construction site robot systems typically rely on pre-set fixed paths or manually set navigation routes for material handling tasks, lacking the ability to autonomously adapt to real-time dynamic environments. Furthermore, traditional methods often fail to accurately integrate construction design models with real-time spatial perception data, making it difficult for robots to accurately perceive changes in the site environment, affecting the accuracy and reliability of path planning and control. At the same time, when robots operate in complex environments with dynamic obstacles, the lack of effective disturbance sensitivity assessment and real-time path optimization methods makes it difficult to quickly respond to environmental changes and make timely adjustments, leading to increased safety risks and reduced material handling efficiency.

[0004] Therefore, there is an urgent need to develop a control system and method for intelligent handling robots used on construction sites that can accurately integrate real-time perception data and building structure information, support efficient path search and optimization in dynamic environments, and have precise motion control capabilities, so as to improve the safety, adaptability and operating efficiency of handling robots. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a control system and control method for an intelligent handling robot for construction sites.

[0006] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a control system for an intelligent handling robot used on construction sites, the system comprising: Environment registration module: used to extract the spatial geometric features of the site environment based on real-time spatial perception data, and perform spatial registration and fusion processing in combination with pre-stored initial building structure information to generate a real-time raster map containing static structural information and obstacle information. Path search module: Based on the real-time grid map, and with the static structural information, the current pose data of the transport robot, and the preset transport task position as constraints, it uses an ant colony path search method that integrates spatial biomimetic memory mechanism to generate an initial transport path. Path optimization module: used to perform disturbance sensitivity iterative optimization processing on the initial transport path based on the real-time disturbance characteristics of the nodes in the initial transport path, and generate the target transport path; Motion control module: used to determine the speed control sequence during the operation of the transport robot based on the spatial relative relationship between the target transport path and the obstacle information, and to generate robot path control instructions based on the target transport path and the speed control sequence.

[0007] Furthermore, the spatial registration and fusion process includes: Based on the depth discontinuity features or point cloud spatial distribution features in real-time spatial perception data, determine the spatial boundary location points of the on-site environment. Based on the spatial positional relationship between adjacent spatial boundary points, determine the spatial geometric characteristics of the site environment; Based on the pre-stored initial building structure information, the structural model data is analyzed, and a model feature index corresponding to the site environment is established; Based on the static structural features and model feature indexes corresponding to the initial building structure in the spatial geometric features, the spatial position deviation of the site environment relative to the initial building structure is determined; The spatial coordinates of the real-time spatial perception data are adjusted based on the spatial position deviation, and then fused with the initial building structure information to generate a real-time grid map of the construction site containing static structural information and obstacle information.

[0008] Furthermore, determining the spatial positional deviation of the site environment relative to the initial building structure includes: Extract spatial geometric feature vectors from the spatial geometric features of the site environment, and initially construct a three-dimensional coordinate system of the site features to be registered based on the spatial geometric feature vectors; Retrieve model structural features that match the spatial geometric feature vector from the initial building structure information, and use the successfully matched model structural features as a reference to determine its spatial center point and reference three-dimensional coordinate system; Based on the spatial correspondence between the successfully matched on-site spatial geometric features and model structural features, calculate the spatial translation vector between the center points of the spatial location, and calculate the spatial rotation matrix based on the angles between the three-dimensional coordinate system of the on-site features and the reference three-dimensional coordinate system at each axis. Based on the calculated spatial translation vector and spatial rotation matrix, a six-degree-of-freedom spatial pose transformation matrix between the site environment and the initial building structure is constructed to obtain the spatial position deviation.

[0009] Furthermore, the process of generating the initial transport path is as follows: Based on the passage status of grid nodes mapped in the real-time grid map according to the static structural information, the feasible passage of the transport robot in the construction site is determined. Based on the historical passage status data of the handling robot in the feasible passage, if there is no historical passage status data, the preset baseline passage status data is called to determine the spatial bionic memory factor of each feasible passage. Based on the spatial biomimetic memory factor, the access probability of grid nodes during the ant colony path search process is adjusted; Based on the adjusted grid node access probabilities, an initial transport path is generated from the current pose of the transport robot to the preset transport task location.

[0010] Furthermore, determining the spatial biomimetic memory factor for each feasible channel includes: The smoothness index of the feasible passage is calculated based on the travel speed and travel time of the handling robot when it has passed through the feasible passage in the past, or based on the preset benchmark speed and benchmark time. Based on the travel smoothness index of the feasible route, the stability factor of the feasible route is determined; Obtain the physical length of the feasible path, and determine the path priority weight of the feasible path based on the ratio of the stability factor to the physical length of the feasible path; The spatial biomimetic memory factor of each feasible channel is determined based on the path priority weight.

[0011] Furthermore, the disturbance sensitivity iterative optimization process includes: The spatial disturbance range of the nodes is determined based on the spatial coordinates of the nodes on the initial transport path in the real-time grid map and the preset safety perception radius. The disturbance density of path nodes is determined based on the number and distribution of obstacle regions within the spatial disturbance range. Based on the perturbation density and its distribution along the path, a sensitivity sorting queue is constructed for the path nodes; The optimal node spatial location is determined based on the sensitivity sorting queue, and the node positions in the initial transport path are updated to obtain the target transport path.

[0012] Furthermore, the sensitivity sorting queue for constructing path nodes includes: Based on the disturbance density of the path nodes, the regional disturbance intensity of the nodes is determined; The sensitivity local gradient of a node is determined based on the difference in regional perturbation intensity between adjacent nodes along the path. Based on the regional disturbance intensity and the sensitivity local gradient, a comprehensive sensitivity evaluation index is calculated for the nodes, and the path nodes are sorted according to the comprehensive sensitivity evaluation index to construct a sensitivity ranking queue for the path nodes.

[0013] Furthermore, determining the optimal node spatial location based on the sensitivity sorting queue includes: Based on the comprehensive sensitivity evaluation index of path nodes in the sensitivity sorting queue, determine the adjustable range of node spatial position; Based on the adjustable range of the node and the motion constraints of the handling robot, determine the candidate adjustment range of the node position; Based on the spatial topological relationship between nodes and obstacles within the candidate adjustment range, the optimal node spatial position is selected.

[0014] Furthermore, determining the speed control sequence during the operation of the handling robot includes: The initial movement speed of each path segment is determined based on the spatial distance between nodes on the target transport path and the disturbance density of obstacle information. Acquire real-time raster maps from multiple consecutive historical time steps and combine them with the real-time raster map at the current time step to construct a real-time raster map sequence; Based on the real-time grid map sequence, obstacles near the path nodes are correlated and tracked in a time sequence, and the relative spatial distance between the obstacles and the path nodes and the moving speed of the obstacles are calculated. Based on the relative spatial distance and the moving speed of the obstacle, a speed adjustment factor is determined at the path node, and the initial movement speed of each path segment is adjusted using this speed adjustment factor to obtain the target passing speed of the path node. Based on the target passing speed of the path nodes, a speed control sequence is generated for the operation of the handling robot.

[0015] Secondly, the present invention provides a control method for an intelligent handling robot used on construction sites, the method comprising: Based on real-time spatial perception data, the spatial geometric features of the site environment are extracted, and spatial registration and fusion processing is performed in combination with the pre-stored initial building structure information to generate a real-time raster map containing static structural information and obstacle information. Based on the real-time grid map, and with the static structural information, the current pose data of the transport robot, and the preset transport task position as constraints, an ant colony path search method that integrates spatial biomimetic memory mechanism is used to generate an initial transport path. Based on the real-time disturbance characteristics of the nodes in the initial transport path, the initial transport path is subjected to disturbance sensitivity iterative optimization processing to generate the target transport path; Based on the spatial relative relationship between the target transport path and the obstacle information, a speed control sequence is determined during the operation of the transport robot, and a robot path control command is generated based on the target transport path and the speed control sequence.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This invention uses spatial registration and fusion processing of real-time spatial perception data and initial building structure information to accurately construct a real-time grid map containing static structural information and obstacle information. This solves the problem of inaccurate robot positioning caused by insufficient environmental perception accuracy at construction sites in existing technologies, and improves the real-time perception capability and operational safety of intelligent handling robots in complex construction site environments.

[0017] This invention generates a target transport path that can dynamically adapt to changes in the construction site environment by integrating an ant colony path search method with a spatial biomimetic memory mechanism and a disturbance-sensitive iterative optimization process. This overcomes the problem that traditional path planning methods cannot effectively cope with environmental disturbances and improves the robot's path planning adaptability and operational stability in complex dynamic environments.

[0018] This invention determines the speed control sequence and generates precise path control commands in real time based on the spatial relative relationship between the obstacle and the transport robot and the obstacle's moving speed. This avoids the problem of insufficient dynamic response capability of existing robot motion control methods to the environment, and realizes efficient, safe and precise motion control of the robot in the complex and ever-changing environment of the construction site, further improving the robot's operating efficiency and the safety of the construction site. Attached Figure Description

[0019] Figure 1 This is an architecture diagram of the control system for an intelligent handling robot used on a construction site, as shown in Example 1.

[0020] Figure 2 This is a flowchart of a control method for an intelligent handling robot used on a construction site, as shown in Example 2. Detailed Implementation

[0021] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example 1

[0022] Please see Figure 1 This invention provides a control system for an intelligent handling robot used on construction sites, comprising: Environment registration module: used to extract the spatial geometric features of the site environment based on real-time spatial perception data, and perform spatial registration and fusion processing in combination with pre-stored initial building structure information to generate a real-time raster map containing static structural information and obstacle information. It should be noted that the real-time spatial perception data specifically refers to three-dimensional point cloud data or depth map data acquired by real-time perception devices such as LiDAR or RGB-D depth cameras. These data reflect the real-time spatial structural state of the site environment. The pre-stored initial building structure information is generally determined by building information model (BIM) data provided during the construction design phase, specifically including the location coordinates, dimensional parameters, and topological relationships of building components.

[0023] Specifically, the spatial registration and fusion process includes: Based on the depth discontinuity features or point cloud spatial distribution features in real-time spatial perception data, determine the spatial boundary location points of the on-site environment. During implementation, the collected real-time 3D point cloud data is first filtered and denoised, and outlier points are removed. Then, the depth difference features of adjacent areas or the point cloud density change features in the spatial data are used to extract the boundary points of the site environment. The boundary points include, but are not limited to, the junction of the wall and the ground, the outline of the structural column, the edge of the temporary construction equipment, and the edge of the area where materials are stacked.

[0024] Based on the spatial positional relationship between adjacent spatial boundary points, determine the spatial geometric characteristics of the site environment; It should be understood that the spatial geometric features are obtained by calculating geometric parameters such as the spatial distance between adjacent spatial boundary points, the relative position angle, and the point set normal vector; specifically, the spatial geometric features include spatial structural elements such as the planar equation of the wall structure, the spatial outline of the columns, and the outer contour line of the edge area of ​​the material stacked on site.

[0025] Based on the pre-stored initial building structure information, the structural model data is analyzed, and a model feature index corresponding to the site environment is established; It should be further noted that the initial building structure information is generally stored in the BIM model. The specific data includes the spatial location, component size, and spatial relationship between components of each structural component during the design phase. In order to achieve the integration between the on-site perception data and the building model data, this embodiment parses the structural model data and establishes a structural feature index. The feature index includes the coordinates of the center point of the spatial location of each structural element, the geometric dimensions of the component, and its spatial orientation information, for subsequent spatial matching.

[0026] Based on the static structural features and model feature indexes corresponding to the initial building structure in the spatial geometric features, the spatial position deviation of the site environment relative to the initial building structure is determined; Understandably, determining the spatial deviation between the site environment and the building structure model is necessary to achieve data fusion under a unified spatial coordinate system, thereby providing the necessary conditions for generating accurate real-time raster maps.

[0027] The determination of the spatial positional deviation of the site environment relative to the initial building structure includes: Extract spatial geometric feature vectors from the spatial geometric features of the site environment, and initially construct a three-dimensional coordinate system of the site features to be registered based on the spatial geometric feature vectors; It should be understood that the spatial geometric feature vector specifically includes a feature set composed of the direction vector, normal vector, or spatial geometric center point of the on-site environment spatial geometric features.

[0028] In practice, the main direction vector and corresponding normal vector of each feature structure (such as wall, column or equipment outline) in space are first determined based on the extracted spatial geometric features. Then, the main direction vector or normal vector is used as a reference to initially establish a three-dimensional coordinate system of the site environment.

[0029] It should be noted that the established three-dimensional coordinate system includes the spatial geometric center point of the site environment as the origin, and the principal direction vector and normal vector as the directions of each coordinate axis, thus forming an initial coordinate reference system for registration.

[0030] Retrieve model structural features that match the spatial geometric feature vector from the initial building structure information, and use the successfully matched model structural features as a reference to determine its spatial center point and reference three-dimensional coordinate system; It should be further noted that the initial building structure information in this embodiment comes from the building information model (BIM) generated during the pre-construction design phase, including the spatial coordinates, dimensions, and structural orientation information of the components.

[0031] In practice, the spatial geometric feature vectors of the site environment are first used to search the model structure database one by one using spatial feature matching methods (such as vector angle calculation or spatial feature similarity measurement methods), and it is determined whether the site feature vectors and the feature vectors in the model meet the pre-set matching threshold conditions. When the matching threshold conditions are met, the feature in the model is taken as the successfully matched model structure feature. Then, the center point of the spatial position corresponding to the successfully matched model structure feature is used as the origin of the reference three-dimensional coordinate system, and the principal direction vector and normal vector of the model structure feature are used to determine the coordinate axis direction of the reference three-dimensional coordinate system.

[0032] Based on the spatial correspondence between the successfully matched on-site spatial geometric features and model structural features, calculate the spatial translation vector between the center points of the spatial location, and calculate the spatial rotation matrix based on the angles between the three-dimensional coordinate system of the on-site features and the reference three-dimensional coordinate system at each axis. In practice, the calculation process for the spatial translation vector between the center points of the spatial location is as follows: First, the coordinates of the spatial center points of the on-site spatial geometric features and the successfully matched model structural features are extracted respectively, where the coordinates of the center point of the on-site spatial geometric features are labeled P. real (x) r , yr ,z r The coordinates of the center point of the model's structural features are labeled P. model (x) m ,y m ,z m ); Subsequently, based on the coordinates of the aforementioned center point, the vector difference is calculated to obtain the spatial translation vector T. The specific calculation formula is as follows: .

[0033] It should be understood that the spatial rotation matrix is ​​used to describe the directional difference between the three-dimensional coordinate system of the site features and the reference three-dimensional coordinate system; in this embodiment, the calculation process of the spatial rotation matrix is ​​as follows: Extract the direction vectors of each axis of the three-dimensional coordinate system of the scene features, denoted as X. r Yr and Zr; and the corresponding axis direction vectors of the reference three-dimensional coordinate system, denoted as X, Yr, and Zr respectively. m Y m and Z m ; The rotation angle of each axis is obtained by calculating the angle between each pair of corresponding axis direction vectors between the on-site feature 3D coordinate system and the reference 3D coordinate system. The specific calculation formula is as follows: , ; In the formula, θ x θ y and θ z These represent the rotation angles of the on-site feature 3D coordinate system relative to the model reference 3D coordinate system in the X, Y, and Z axes, respectively. Based on the rotation angles of each axis mentioned above, a composite rotation is performed in sequence around the coordinate axes Z, Y, and X to obtain the spatial rotation matrix R.

[0034] For example, the spatial rotation matrix R is represented as follows: ; The rotation matrix around the X-axis is represented as follows: ; The rotation matrix around the Y-axis is represented as: ; The rotation matrix around the Z-axis is represented as: .

[0035] Based on the calculated spatial translation vector and spatial rotation matrix, a six-degree-of-freedom spatial pose transformation matrix between the site environment and the initial building structure is constructed to obtain the spatial position deviation.

[0036] It should be understood that, in this embodiment, the six-degree-of-freedom spatial pose transformation matrix is ​​used to describe the position and orientation relationship between the coordinate system of the site environment and the coordinate system of the initial building structure model, including two parts: translation and rotation in three-dimensional space; specifically, the spatial pose transformation matrix is ​​represented in the form of a homogeneous transformation matrix: ; In the formula, M represents the six-degree-of-freedom spatial pose transformation matrix of the site environment relative to the initial building structure model; R represents the spatial rotation matrix, used to describe the rotational relationship between the site environment coordinate system and the initial building structure model coordinate system; T represents the spatial translation vector, used to describe the positional displacement between the site environment coordinate system and the initial building structure model coordinate system; r ab This represents the element in the rotation matrix R at row a and column b, and its specific value is determined by the corresponding term obtained after multiplying and combining the rotation matrices of each axis mentioned above; t x t y and t z These represent the component values ​​of the spatial translation vector T in the X, Y, and Z axes, respectively.

[0037] In summary, the spatial position deviation in this embodiment is specifically quantified by the aforementioned spatial pose transformation matrix M, which is used to achieve accurate registration and fusion of on-site perception data and building model data in a unified spatial coordinate system. This provides a unified spatial reference for subsequent real-time grid map construction, ensuring that the handling robot achieves accurate path planning and motion control.

[0038] The spatial coordinates of the real-time spatial perception data are adjusted based on the spatial position deviation, and then fused with the initial building structure information to generate a real-time grid map of the construction site containing static structural information and obstacle information.

[0039] During implementation, after determining the spatial positional deviation, this embodiment uses a spatial transformation method to adjust the coordinates of the real-time spatial perception data, ensuring consistency between the real-time data and the initial building model data in spatial coordinates. Specifically, the coordinate fusion process uses a raster map to represent the fused data. The real-time raster map is spatially divided into regular units, each of which records the spatial and attribute information of the static structure or obstacle at the corresponding location, providing a unified data format.

[0040] It should be noted that, in order to meet the system's subsequent needs for multiple historical time data sequences, each generated real-time raster map will be saved to the system storage device at a set time interval, forming a continuous historical raster map data sequence, which is convenient for the path planning module and motion control module to use when making dynamic path search and speed control decisions.

[0041] Path search module: Based on the real-time grid map, and with the static structural information, the current pose data of the transport robot, and the preset transport task position as constraints, it uses an ant colony path search method that integrates spatial biomimetic memory mechanism to generate an initial transport path. It should be noted that the real-time grid map contains static structural information and obstacle information, and each grid node is marked as passable or impassable; the current pose data of the transport robot specifically includes the robot's position coordinates and posture information in the real-time grid map; the preset transport task location is specifically the spatial coordinates of the transport target location within the construction site.

[0042] Specifically, the ant colony path search method integrating spatial biomimetic memory mechanism includes: Based on the passage status of grid nodes mapped in the real-time grid map according to the static structural information, the feasible passage of the transport robot in the construction site is determined. It should be noted that the grid node accessibility status mentioned in this embodiment specifically refers to the accessibility marker of each grid cell in the real-time grid map, which includes two states: accessible and inaccessible. After the static structural information is mapped into the grid map, the grid cell where the structure is located is marked as inaccessible, while the grid cells in the unobstructed area outside the structure are marked as accessible.

[0043] In practice, based on the current pose coordinates of the transport robot and the preset transport task coordinates, combined with all grid nodes marked as passable in the grid map, all passable areas where the robot can move within the construction site are determined, which are the feasible passages.

[0044] Based on the historical passage status data of the handling robot in the feasible passage, if there is no historical passage status data, the preset baseline passage status data is called to determine the spatial bionic memory factor of each feasible passage. It should be further noted that the historical passage status data specifically refers to the historical data such as the travel speed and passage time recorded by the transport robot during the execution of multiple tasks in the past, which is stored in the system storage device; when the system is started for the first time or when historical data is missing, the system uses the preset benchmark passage status data, including the benchmark travel speed and benchmark passage time.

[0045] In practice, the spatial biomimetic memory factor is used to characterize the degree to which the transport robot remembers the historical passage status characteristics of different feasible channels, so as to simulate the mechanism by which biological individuals select paths based on historical experience.

[0046] The determination of the spatial biomimetic memory factor for each feasible channel includes: The smoothness index of the feasible passage is calculated based on the travel speed and travel time of the handling robot when it has passed through the feasible passage in the past, or based on the preset benchmark speed and benchmark time. In practice, the specific calculation formula for the smoothness of movement index is expressed as follows: ; In the formula, F smooth V represents the smoothness of travel along a feasible route. real V represents the historical travel speed of the transport robot in this channel. base This represents the preset baseline travel speed, t. base t represents the preset baseline travel time. real This indicates the historical passage time of the transport robot in this passage.

[0047] For feasible channels lacking historical data, the above indicators are calculated using the baseline speed and baseline time preset by the system.

[0048] It should be noted that the travel smoothness index F of the feasible channel smooth The larger the value, the smoother the passage; the preset reference speed represents the reference average speed of the transport robot under normal conditions in the construction site, and the reference time specifically represents the reference length of time for the robot to pass through a feasible passage under normal conditions.

[0049] Based on the travel smoothness index of the feasible route, the stability factor of the feasible route is determined; In practical implementation, the stability factor of the feasible channel is specifically calculated using the following formula: ; In the formula, F stable The stability factor represents the feasible path.

[0050] It should be noted that the stability factor F stable This value is used to characterize the stability of the transport robot in traversing the feasible path throughout history; a higher value indicates that the path has historically had better travel stability.

[0051] Obtain the physical length of the feasible path, and determine the path priority weight of the feasible path based on the ratio of the stability factor to the physical length of the feasible path; In practical implementation, the priority weight of feasible pathway paths can be expressed as: ; In the formula, W priority L represents the path priority weight of a feasible route. path This indicates the physical length of the feasible passage on the real-time raster map.

[0052] It should be noted that the path priority weight is used to characterize the priority of the handling robot when selecting different feasible channels. A larger value indicates that the feasible channel has higher historical traffic stability and shorter physical length, and is therefore given priority in the path search process.

[0053] The spatial biomimetic memory factor of each feasible channel is determined based on the path priority weight.

[0054] In practice, the path priority weights of all feasible channels are normalized, specifically as follows: ; In the formula, M i W represents the spatial biomimetic memory factor of the i-th feasible channel. priority (i) represents the path priority weight corresponding to the i-th feasible path. This represents the sum of priority weights for all feasible routes, where n represents the total number of feasible routes in the real-time raster map.

[0055] It should be understood that the spatial biomimetic memory factor is used to dynamically adjust the access probability of the transport robot to the grid node during the subsequent ant colony path search process; the larger the memory factor value of each channel, the more stable and shorter the passage state of the transport robot in that channel in the past, thus increasing the access probability of the corresponding channel grid node in the ant colony algorithm.

[0056] Based on the spatial biomimetic memory factor, the access probability of grid nodes during the ant colony path search process is adjusted; It is understandable that in this embodiment, during the initial transport path search process using the ant colony path search algorithm, each ant visits the grid nodes in a probabilistic manner to find a path; the role of the spatial biomimetic memory factor is to dynamically adjust the access probability of grid nodes in each feasible channel, increase the access probability of path nodes with better historical access status, and decrease the access probability of path nodes with poor historical access status, thereby affecting the path search behavior of the ant colony.

[0057] Based on the adjusted grid node access probabilities, an initial transport path is generated from the current pose of the transport robot to the preset transport task location.

[0058] It should be noted that the adjusted grid node access probability is specifically used as the selection rule for each node in the ant colony path search process. That is, at the current grid node, the ant selects the next node based on the access probability of all neighboring passable grid nodes. By repeating the above process, a path is gradually formed from the current position of the transport robot to the target position of the task. After multiple iterations, the path that obtains the optimal fitness evaluation value is selected as the initial transport path.

[0059] Path optimization module: used to perform disturbance sensitivity iterative optimization processing on the initial transport path based on the real-time disturbance characteristics of the nodes in the initial transport path, and generate the target transport path; It should be noted that the real-time disturbance features specifically refer to the dynamic changes of obstacles in space in the real-time grid map. These changes may have a certain impact on the initial transport path; therefore, it is necessary to conduct disturbance sensitivity assessment and position optimization for each path node.

[0060] Specifically, the disturbance sensitivity iterative optimization process includes: The spatial disturbance range of the nodes is determined based on the spatial coordinates of the nodes on the initial transport path in the real-time grid map and the preset safety perception radius. It should be noted that the safety perception radius is a pre-set safety parameter, and the specific value is set to 0.5m for example; the spatial disturbance range specifically refers to the circular area formed on the real-time grid map with the spatial location coordinates of each path node as the center and the safety perception radius as the radius.

[0061] The disturbance density of path nodes is determined based on the number and distribution of obstacle regions within the spatial disturbance range. In practice, the specific formula for calculating the disturbance density is as follows: ; In the formula, D density (i) represents the disturbance density of the i-th node on the initial transport path, N obstacle(i) represents the number of obstacle regions within the spatial disturbance range of the i-th node, A range This represents the area of ​​the spatial disturbance range of the node.

[0062] It should be noted that the number of obstacle areas refers to the number of grid cells marked as impassable within the spatial disturbance range, while the obstacle distribution refers to the specific distribution of these grid cells in space.

[0063] Based on the perturbation density and its distribution along the path, a sensitivity sorting queue is constructed for the path nodes; In implementation, the sensitivity sorting queue for constructing path nodes includes: Based on the disturbance density of the path nodes, the regional disturbance intensity of the nodes is determined; In practical applications, the regional disturbance intensity of the node is specifically expressed as follows: ; In the formula, F intensity (i) represents the regional disturbance intensity of the i-th node on the initial transport path.

[0064] It is understandable that the regional disturbance intensity is directly quantified using the numerical value of disturbance density to characterize the degree of spatial disturbance at path nodes.

[0065] The sensitivity local gradient of a node is determined based on the difference in regional perturbation intensity between adjacent nodes along the path. In practical implementation, the formula for calculating the comprehensive sensitivity evaluation index in this embodiment is as follows: ; In the formula, G local (i) represents the local sensitivity gradient of the i-th node, F intensity (i+1) represents the regional disturbance intensity of the (i+1)th node, F intensity (i-1) represents the regional disturbance intensity of the (i-1)th node.

[0066] It should be noted that the sensitive local gradient is used to describe the degree of change in the intensity of disturbance in the region before and after a path node.

[0067] Based on the regional disturbance intensity and the sensitivity local gradient, a comprehensive sensitivity evaluation index is calculated for the nodes, and the path nodes are sorted according to the comprehensive sensitivity evaluation index to construct a sensitivity ranking queue for the path nodes.

[0068] In practical implementation, the formula for calculating the comprehensive sensitivity evaluation index in this embodiment is as follows: ; In the formula, E sensitivity(i) represents the comprehensive sensitivity evaluation index of the i-th node, and α and β represent the weight coefficients of regional disturbance intensity and local sensitivity gradient, respectively. It should be noted that the values ​​of the weighting coefficients α and β were determined based on multiple experiments and the actual field scenario. In this embodiment, the regional disturbance intensity weighting coefficient is 0.6, and the sensitive local gradient weighting coefficient β is 0.4.

[0069] Based on the magnitude of the comprehensive sensitivity evaluation index values, the path nodes are sorted from largest to smallest to obtain the sensitivity ranking queue.

[0070] It should be understood that nodes ranked higher have a higher priority in subsequent position optimization processes.

[0071] The optimal node spatial location is determined based on the sensitivity sorting queue, and the node positions in the initial transport path are updated to obtain the target transport path.

[0072] The step of determining the optimal node spatial location based on the sensitivity sorting queue includes: Based on the comprehensive sensitivity evaluation index of path nodes in the sensitivity sorting queue, determine the adjustable range of node spatial position; In practical implementation, the adjustable range of the point's spatial position is determined using the following formula: ; In the formula, L adjust (i) represents the adjustable range of the spatial position of the i-th path node in the sensitivity sorting queue, L max E represents the maximum allowable adjustment range of the preset node position. max This represents the maximum value of the comprehensive sensitivity evaluation index among all nodes.

[0073] It should be noted that the node sensitivity index L adjust (i) The larger the value, the larger the adjustable range; the maximum allowable adjustment range L max Based on the actual application scenario, the value is set to 1.0m in this embodiment.

[0074] Based on the adjustable range of the node and the motion constraints of the handling robot, determine the candidate adjustment range of the node position; It should be noted that the motion constraints of the transport robot specifically include constraint parameters such as the robot's turning radius, maximum lateral and longitudinal movement distance; in this embodiment, the robot's minimum turning radius is 0.6m, the maximum lateral movement distance in a single step is 0.5m, and the maximum longitudinal movement distance in a single step is within the range of 0.8m.

[0075] It should also be noted that the candidate adjustment range is specifically determined by taking the initial path node position as the center, the adjustable range as the basis, and combining the above-mentioned robot motion constraints. For example, a rectangular or circular candidate region can be constructed based on the adjustable range and motion constraints.

[0076] Based on the spatial topological relationship between nodes and obstacles within the candidate adjustment range, the optimal node spatial position is selected.

[0077] It should be further noted that this embodiment selects the optimal location by calculating the spatial topological relationships between multiple locations within the candidate adjustment range and obstacles, specifically including minimum distance and feasibility status; the location optimization calculation method uses the following formula: ; In the formula, P optimal (i) represents the optimal spatial location of the i-th path node after optimization, S(i) represents the set of all possible location points within the candidate adjustment range of the i-th node, and d(p) represents the minimum spatial distance from the candidate location point p to the nearest obstacle within the candidate adjustment range.

[0078] After optimizing node by node through the above steps, the iterative update from the initial transport path to the target transport path is completed, and the target transport path is obtained.

[0079] Motion control module: used to determine the speed control sequence during the operation of the transport robot based on the spatial relative relationship between the target transport path and the obstacle information, and to generate robot path control instructions based on the target transport path and the speed control sequence.

[0080] It should be noted that the motion control module in this embodiment is used to determine the speed control sequence during the operation of the transport robot based on the spatial relative relationship between the target transport path and the obstacle information, and to generate robot path control instructions based on the target transport path and the speed control sequence, so as to ensure the safe and stable operation of the transport robot at the construction site.

[0081] Specifically, determining the speed control sequence during the operation of the handling robot includes: The initial movement speed of each path segment is determined based on the spatial distance between nodes on the target transport path and the disturbance density of obstacle information. In implementation, the target transport path is composed of multiple path nodes connected sequentially. Each path segment specifically refers to the path portion between two adjacent nodes. The specific formula for calculating the initial movement speed of each path segment is as follows: ; In the formula, V initial(k) represents the initial velocity of the Kth path segment, V max V represents the robot's maximum permissible operating speed. base L represents the preset baseline speed of the handling robot. segment (k) represents the spatial length of the Kth path segment in the target transport path, D density (k) represents the obstacle disturbance density value corresponding to the Kth path segment, and λ and μ represent the preset model adjustment coefficients; It should be noted that the maximum permissible operating speed V max Reference speed V base The preset model adjustment coefficients λ and μ are set according to the actual scene. In this embodiment, the maximum allowable operating speed V max The value is taken as 1.5 m / s, and the reference velocity V is... base The value is 0.5 m / s, the adjustment coefficient λ is 1.0, and the adjustment coefficient μ is 0.5; the obstacle disturbance density value D density (k) is obtained by the ratio of the number of grid nodes marked as obstacles near the path segment in the real-time grid map to the length of the path segment.

[0082] It should be understood that when the obstacle disturbance density value is large, the initial movement speed will be reduced accordingly to ensure safety.

[0083] Acquire real-time raster maps from multiple consecutive historical time steps and combine them with the real-time raster map at the current time step to construct a real-time raster map sequence; It should be further noted that this embodiment uses a fixed sampling time interval to continuously collect real-time grid maps of multiple historical time steps; for example, it sets one frame to be collected every 2 seconds, and collects a total of 5 consecutive frames of historical grid maps, which are then combined with the current real-time grid map to form a real-time grid map sequence for the temporal analysis and motion state recognition of obstacles in the next step.

[0084] Based on the real-time grid map sequence, obstacles near the path nodes are correlated and tracked in a time sequence, and the relative spatial distance between the obstacles and the path nodes and the moving speed of the obstacles are calculated. In the specific implementation process, the temporal association and tracking of obstacles in this embodiment refers to identifying the spatial position changes of obstacles between adjacent time steps based on multiple consecutive frames of grid maps, and determining the motion state of obstacles accordingly.

[0085] The formula for calculating the relative spatial distance between obstacles and path nodes is as follows: ; In the formula, D rel (k) represents the relative spatial distance between the k-th path node and the nearest obstacle, Pnode (k) represents the spatial coordinates of the k-th path node, P obs (k) represents the spatial location of the obstacle near the k-th path node.

[0086] The specific formula for calculating the moving speed of an obstacle is as follows: ; In the formula, V obs (k) represents the moving speed of the obstacle near the k-th path node, P obs (k, t) represents the spatial position of the obstacle at the current time step, P obs (k, t-1) represents the spatial position of the obstacle in the previous time step, and Δt represents the time interval between two consecutive historical time steps.

[0087] Based on the relative spatial distance and the moving speed of the obstacle, a speed adjustment factor is determined at the path node, and the initial movement speed of each path segment is adjusted using this speed adjustment factor to obtain the target passing speed of the path node. In the specific implementation process, the formula for calculating the speed adjustment factor at path nodes is as follows: ; In the formula, f adjust (k) represents the speed adjustment factor at the k-th path node, D safe ω represents the preset safety braking buffer distance constant, and ω represents the weight of the obstacle speed influence.

[0088] It should be noted that the safety braking buffer distance constant D safe The weight ω for the influence of obstacle speed is set according to the on-site construction requirements. In this embodiment, the safety braking buffer distance constant D is... safe The value is 2.0m, and the obstacle speed influence weight ω is 1.0.

[0089] Understandably, the magnitude of the speed adjustment factor represents the degree of influence of obstacles near the path node on the robot's running speed; when the obstacle is close to the node or the obstacle moves at a high speed, the value of the speed adjustment factor decreases, thereby further reducing the robot's speed when passing through the node.

[0090] The specific formula for calculating the target passage speed of a path node is as follows: ; In the formula, V target (k) represents the target speed passing through the k-th path node; Based on the target passing speed of the path nodes, a speed control sequence is generated for the operation of the handling robot.

[0091] It should be noted that by calculating the target passing speed of each node on the target transport path one by one, the complete speed control sequence of the transport robot during operation is obtained.

[0092] During implementation, based on the determined target transport path and speed control sequence, the motion control module of this embodiment generates robot path control instructions. Specifically, the path control instructions include: the spatial position coordinate sequence of the target transport path nodes, the target passing speed value that the robot should adopt at each path node, and specific motion control instruction signals (such as motor speed, torque setpoint, etc.) generated based on the above speed values.

[0093] The path control command is output to the motion control execution unit of the handling robot, thereby guiding the robot to safely and accurately perform the handling task based on the real-time obstacle status. Example 2

[0094] like Figure 2 As shown in the example, the parts not detailed in this embodiment are as shown in Example 1. This embodiment discloses a control method for an intelligent handling robot used on construction sites, including: Based on real-time spatial perception data, the spatial geometric features of the site environment are extracted, and spatial registration and fusion processing is performed in combination with the pre-stored initial building structure information to generate a real-time raster map containing static structural information and obstacle information. Based on the real-time grid map, and with the static structural information, the current pose data of the transport robot, and the preset transport task position as constraints, an ant colony path search method that integrates spatial biomimetic memory mechanism is used to generate an initial transport path. Based on the real-time disturbance characteristics of the nodes in the initial transport path, the initial transport path is subjected to disturbance sensitivity iterative optimization processing to generate the target transport path; Based on the spatial relative relationship between the target transport path and the obstacle information, a speed control sequence is determined during the operation of the transport robot, and a robot path control command is generated based on the target transport path and the speed control sequence.

[0095] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0096] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A control system for an intelligent carrying robot for a construction site, characterized in that, The system comprises: An environment registration module: for extracting the spatial geometric features of the site environment according to real-time spatial perception data, and performing spatial registration fusion processing in combination with pre-stored initial building structure information, to generate a real-time grid map containing static structure information and obstacle information; A path search module: for generating an initial carrying path based on the real-time grid map, taking the static structure information, the current pose data of the carrying robot, and the preset carrying task position as constraint conditions, and using an ant colony path search method that integrates a spatial biomimetic memory mechanism; A path optimization module: for performing disturbance sensitivity iterative optimization processing on the initial carrying path according to the real-time disturbance features of the nodes in the initial carrying path, to generate a target carrying path; A motion control module: for determining a speed control sequence in the running process of the carrying robot according to the spatial relative relationship between the target carrying path and the obstacle information, and generating a robot path control instruction according to the target carrying path and the speed control sequence.

2. The control system of the intelligent carrying robot for a construction site according to claim 1, wherein The spatial registration fusion processing comprises: Determining the spatial boundary position points of the site environment according to the depth discontinuity features or the point cloud spatial distribution features in the real-time spatial perception data; Determining the spatial geometric features of the site environment according to the spatial positional relationship between adjacent spatial boundary position points; Analyzing the structure model data according to the pre-stored initial building structure information, and establishing a model feature index corresponding to the site environment; Determining the spatial position deviation of the site environment relative to the initial building structure according to the static structure features in the spatial geometric features corresponding to the initial building structure and the model feature index; Adjusting the spatial coordinate position of the real-time spatial perception data according to the spatial position deviation, and performing coordinate fusion with the initial building structure information, to generate a real-time grid map of the construction site containing static structure information and obstacle information.

3. The control system of the intelligent carrying robot for a construction site according to claim 2, wherein The determination of the spatial position deviation of the site environment relative to the initial building structure comprises: Extracting a spatial geometric feature vector from the spatial geometric features of the site environment, and initially constructing a site feature three-dimensional coordinate system according to the spatial geometric feature vector; Retrieving a model structure feature matching the spatial geometric feature vector from the initial building structure information, and taking the model structure feature that has successfully matched as a reference to determine its spatial position center point and reference three-dimensional coordinate system; According to the spatial correspondence relationship between the successfully matched site spatial geometric feature and the model structure feature, calculating a spatial translation vector between the spatial position center points, and calculating a spatial rotation matrix according to the axial angles of the site feature three-dimensional coordinate system and the reference three-dimensional coordinate system; According to the calculated spatial translation vector and spatial rotation matrix, constructing a six-degree-of-freedom spatial pose transformation matrix between the site environment and the initial building structure to obtain the spatial position deviation.

4. The control system of the intelligent carrying robot for a construction site according to claim 1, wherein The generation process of the initial carrying path is as follows: Determining the feasible passageway of the carrying robot in the construction site based on the grid node passable state of the static structure information mapped in the real-time grid map; According to historical passing state data of the carrying robot in the feasible passage, if there is no historical passing state data, preset reference passing state data is called, and a spatial biomemorization factor of each feasible passage is determined; According to the spatial biomemorization factor, an access probability of a grid node in an ant colony path search process is adjusted; According to the adjusted access probability of the grid node, an initial carrying path from a current pose of the carrying robot to a preset carrying task position is generated.

5. The control system of the intelligent carrying robot for a construction site according to claim 4, wherein The determination of the spatial biomemorization factor of each feasible passage comprises: According to a travel speed and a passing time of the carrying robot when passing through the feasible passage in the past, or according to a preset reference speed and a reference time, a travel fluency index of the feasible passage is calculated; According to the travel fluency index of the feasible passage, a stability factor of the feasible passage is determined; A physical length of the feasible passage is obtained, and a path priority value of the feasible passage is determined according to a ratio of the stability factor to the physical length of the feasible passage; The spatial biomemorization factor of each feasible passage is determined according to the path priority value.

6. The control system of the intelligent carrying robot for construction sites according to claim 1, characterized in that, The disturbance sensitivity iterative optimization process comprises: According to a spatial position coordinate of a node on the initial carrying path in a real-time grid map and a preset safety perception radius, a spatial disturbance range of the node is determined; According to a number and distribution of obstacle regions in the spatial disturbance range, a disturbance intensity of the path node is determined; According to the disturbance intensity and its distribution along the path, a sensitivity sorting queue of the path node is constructed; An optimal node spatial position is determined according to the sensitivity sorting queue, and a node position on the initial carrying path is updated to obtain a target carrying path.

7. The control system of the intelligent carrying robot for a construction site according to claim 6, wherein The construction of the sensitivity sorting queue of the path node comprises: According to the disturbance intensity of the path node, a regional disturbance intensity of the node is determined; According to a difference in the regional disturbance intensity between adjacent nodes of the path node, a local sensitivity gradient of the node is determined; A comprehensive sensitivity evaluation index of the node is calculated based on the regional disturbance intensity and the local sensitivity gradient, and the path node is sorted according to the comprehensive sensitivity evaluation index to construct the sensitivity sorting queue of the path node.

8. The control system of the intelligent carrying robot for a construction site according to claim 7, wherein The determination of the optimal node spatial position according to the sensitivity sorting queue comprises: According to the comprehensive sensitivity evaluation index of the path node in the sensitivity sorting queue, an adjustable amplitude of the node spatial position is determined; According to the adjustable amplitude of the node and a motion constraint condition of the carrying robot, a candidate adjustment range of the node position is determined; According to a spatial position topological relation between the node and obstacle information in the candidate adjustment range, an optimal node spatial position is selected.

9. The control system of the intelligent carrying robot for construction sites according to claim 1, characterized in that, The determination of the speed control sequence in the carrying robot running process comprises: According to a spatial distance between nodes on the target carrying path and a disturbance intensity of obstacle information, an initial motion speed of each path segment is determined; Real-time grid maps of a plurality of historical time steps are obtained and combined with a real-time grid map of a current time step to construct a real-time grid map sequence; Based on the real-time grid map sequence, obstacles near the path node are time-series associated and tracked, a relative spatial distance between the obstacles and the path node and a moving speed of the obstacles are calculated; and According to the relative spatial distance and the moving speed of the obstacle, a speed adjustment factor at the path node is determined, and the initial motion speed of each path segment is adjusted by using the speed adjustment factor to obtain a target passing speed of the path node; According to the target passing speed of the path node, a speed control sequence in the running process of the carrying robot is generated.

10. A control method of an intelligent carrying robot for a construction site, characterized by, The method comprises: According to the real-time spatial perception data, spatial geometric features of the field environment are extracted, and spatial registration fusion processing is performed in combination with pre-stored initial building structure information to generate a real-time grid map containing static structure information and obstacle information; Based on the real-time grid map, the static structure information, the current pose data of the carrying robot, and the preset carrying task position are taken as constraint conditions, and an ant colony path search method with a fusion spatial biomemetic memory mechanism is used to generate an initial carrying path; According to the real-time disturbance features of the nodes in the initial carrying path, disturbance sensitivity iterative optimization processing is performed on the initial carrying path to generate a target carrying path; According to the spatial relative relationship between the target carrying path and the obstacle information, a speed control sequence in the running process of the carrying robot is determined, and a robot path control instruction is generated according to the target carrying path and the speed control sequence.

Images