Building robot masonry construction path planning method and control system

Abstract

The application discloses a kind of construction robot masonry construction path planning method and control system, comprising the following steps: 1) narrow masonry operation space is scanned using multi-modal perception method, and space boundary, obstacle information and masonry wall process data are collected, dynamic constraint three-dimensional point cloud model is generated, and is mapped to digital twin scene, and physical-virtual two-way synchronous mapping model is constructed;2) generate multiple candidate masonry paths based on improved RRT algorithm, embed masonry process constraint condition and action redundancy determination standard, simulate mechanical arm full-process masonry action in digital twin scene;3) drive mechanical arm by bionic joint control logic, and masonry operation is executed along the optimal path by combining joint space hierarchical planning strategy;4) dynamically correct subsequent masonry path based on the real-time construction data, adapt to masonry working condition deviation and process parameter change, while incorporating energy consumption optimization mechanism and emergency path mechanism.

Description

Technical Field

[0001] This application relates to the field of intelligent buildings, and in particular to a method and control system for planning the construction path of a building robot. Background Technology

[0002] With the rapid development of intelligent construction technology, construction robots are gradually replacing manual labor in wall construction, especially in confined spaces such as coke ovens and furnaces, and in high-precision masonry scenarios. Leveraging their high efficiency and precision, they have become one of the core pieces of equipment for the intelligent transformation of the construction industry. Path planning in masonry operations, as a core technology for autonomous robot operation, directly determines the robotic arm's motion accuracy, work efficiency, energy consumption, and construction safety. Optimizing this technology is crucial for promoting the large-scale deployment of construction robots.

[0003] Currently, path planning for construction robots is mostly based on traditional path search algorithms. Among these, the Rapid Expanding Random Tree (RRT) algorithm and its improved versions are widely used in obstacle avoidance path planning for robotic arms due to their excellent dynamic spatial adaptability. Existing technologies have addressed the issues of unguided random sampling, high path redundancy, and slow convergence speed in traditional RRT algorithms by optimizing sampling mechanisms, adjusting step size strategies, or introducing smoothing algorithms, thus improving the efficiency and accuracy of path planning to some extent. For example, some solutions construct four random trees for collaborative search by setting fixed sampling points, guiding the tree's expansion direction, reducing invalid expansions, and enhancing the stability and feasibility of paths in complex environments. Other solutions combine pixelated planning space with optimized sampling mechanisms to shorten path search and algorithm convergence time. Meanwhile, the integration of digital twin technology with construction robots has also made progress. By constructing virtual simulation scenarios, the construction process can be rehearsed and the robotic arm's motion control can be achieved, improving the robot's adaptability in confined spaces. Some equipment has achieved a construction accuracy within ±1 mm, far exceeding the level of traditional manual processes.

[0004] However, in actual masonry construction scenarios, especially under constraints such as confined spaces, dynamic working conditions, and complex processes, existing path planning technologies still have many shortcomings, such as:

[0005] The integration of complex environmental perception and process parameters is insufficient. Existing multimodal perception solutions mainly focus on collecting spatial boundary and obstacle information, failing to achieve synchronous perception and fusion storage of spatial information and core masonry process parameters. Furthermore, in construction site environments with varying dust and lighting conditions, single sensors or simple combinations of perception structures are prone to data deviations, leading to a disconnect between path planning and actual process requirements, necessitating frequent manual intervention and adjustments. Simultaneously, existing technologies have poor adaptability to dynamic working conditions, struggling to respond in real time to sudden events such as changes in the state of already constructed walls or obstacle displacement, resulting in weak dynamic path adjustment capabilities.

[0006] Existing improved RRT series algorithms mostly focus on obstacle avoidance and convergence speed optimization, without fully incorporating masonry process constraints and motion redundancy judgment criteria. This can lead to issues such as excessively high robotic arm motion redundancy rates and incompatibility with masonry processes in the generated paths. Furthermore, existing solutions rarely incorporate energy consumption assessment into the core indicators of path selection. During robotic arm operations, unreasonable paths often result in excessive joint energy consumption and motor load, which does not meet the requirements of green construction.

[0007] In confined spaces, the range of motion of robotic arms is limited. Existing control logic often relies on large-scale movements of the entire arm, resulting in high workload occupancy and a significant risk of collisions. While biomimetic control concepts are applied, they lack deep integration with hierarchical planning of joint space, making it difficult to achieve high-precision micro-adjustments and multi-joint coordinated movements. Furthermore, the handling of masonry deviations is crude, lacking a tiered correction strategy. A single adjustment mode is used for deviations of varying degrees, leading to either over-correction affecting work efficiency or under-correction causing potential construction quality issues.

[0008] Therefore, a method and control system for planning the construction path of a building robot for masonry construction is proposed. Summary of the Invention

[0009] This application aims to at least partially solve one of the technical problems in the aforementioned technologies.

[0010] To achieve the above objectives, the first aspect of this application proposes a method for planning a construction path for a building robot, comprising the following steps:

[0011] 1) A multimodal perception method is used to scan the narrow masonry work space, collect information on the space boundary, obstacles and the process data of the masonry wall, generate a dynamic constrained three-dimensional point cloud model, and map it to the digital twin scene to build a physical-virtual two-way synchronous mapping model.

[0012] 2) Based on the improved RRT algorithm, multiple candidate masonry paths are generated, and masonry process constraints and motion redundancy judgment criteria are embedded. The entire process of robotic arm masonry action is simulated in the digital twin scenario, and the optimal path with no collision, meeting the redundancy rate and process requirements is selected.

[0013] 3) The robotic arm is driven by bionic joint control logic and performs masonry work along the optimal path by combining the joint space hierarchical planning strategy, while simultaneously collecting real-time construction data through the multimodal feedback module;

[0014] 4) Based on the real-time construction data, the subsequent masonry path is dynamically corrected to adapt to the deviation of masonry working conditions and changes in process parameters. At the same time, energy consumption optimization mechanism and emergency path mechanism are integrated to ensure the efficiency, quality and safety of masonry operations.

[0015] In addition, the construction robot masonry path planning method proposed in this application may also have the following additional technical features:

[0016] As a further description of the above technical solution:

[0017] The multimodal perception method described in step 1) adopts a combination structure of monocular vision sensor, lidar and inertial measurement unit to synchronously collect core parameters of masonry process. The core parameters include at least mortar fluidity and brick pressing force data, so as to realize the synchronous perception and fusion storage of spatial information and process parameters.

[0018] As a further description of the above technical solution:

[0019] In step 2), based on the size of the work area and the distribution of obstacles, 3-5 candidate masonry paths are generated. The redundancy judgment standard is set to the action redundancy rate not exceeding 15%. Through full-process simulation of the digital twin scenario, candidate paths with collision risks and excessive redundancy rates are eliminated. The selected optimal path is then synchronously sent to the construction robot execution end and the digital twin module for filing.

[0020] As a further description of the above technical solution:

[0021] The bionic joint control logic described in step 3) is achieved through the collaboration of flexible drive components and high-precision torque detection elements, driving the robotic arm joints to perform micron-level high-precision micro-adjustments. By replacing the large-scale movement of the entire arm with the coordinated movement of multiple joints, the occupation rate and collision risk in confined spaces are significantly reduced.

[0022] As a further description of the above technical solution:

[0023] The deviation classification correction mentioned in step 4) specifically involves dividing the masonry deviation into three levels: slight deviation, moderate deviation, and severe deviation. Slight deviation corresponds to a deviation value ≤1mm, moderate deviation corresponds to a deviation value of 1-2mm, and severe deviation corresponds to a deviation value >2mm. For different levels of deviation, a classification correction strategy is triggered, which includes path fine-tuning, single brick re-laying path planning, and local area path backtracking, to ensure that the deviation is controllable.

[0024] As a further description of the above technical solution:

[0025] The energy consumption optimization mechanism described in step 4) is implemented by adding an energy consumption assessment unit in the digital twin pre-simulation stage. This unit quantifies the joint motion energy consumption, motor load, and operation time of each candidate path, and prioritizes the selection of paths whose energy consumption is reduced by more than 15% compared to the benchmark path and meets the masonry process requirements, thus balancing efficiency and energy saving.

[0026] As a further description of the above technical solution:

[0027] The emergency path mechanism described in step 4) presets emergency path templates for various sudden faults. The sudden faults include at least brick jamming, mortar blockage, and sensor failure. After the fault is triggered, the robotic arm backtracks to the preset safe position along the minimum damage trajectory. The path backtracking time is strictly controlled within 5 seconds, and the audible and visual warning signals are triggered simultaneously and the fault information is uploaded.

[0028] The second aspect of this application proposes a construction robot masonry construction path planning and control system, including:

[0029] The sensing module is used to perform multimodal sensing scanning to collect information on confined spaces, masonry process parameters, and real-time construction data.

[0030] The digital twin module is used to build a physical-virtual synchronous mapping model to complete the virtual pre-playing, screening and dynamic updating of candidate paths;

[0031] The path planning module is used to run the improved RRT algorithm to generate candidate paths, and combine redundancy determination and energy consumption assessment to complete the selection and distribution of the optimal path;

[0032] The biomimetic control module is used to drive the coordinated movement of the robotic arm based on biomimetic joint control logic, and to perform joint space hierarchical trajectory planning and work posture adaptation.

[0033] The feedback correction module is used to achieve dynamic path adjustment, graded correction of masonry deviation, and emergency path triggering based on multimodal real-time data.

[0034] As a further description of the above technical solution:

[0035] The bionic control module has multiple core masonry posture template libraries built in, which can automatically match and adapt the working posture to the current confined space conditions based on the pre-simulation results of the digital twin module.

[0036] As a further description of the above technical solution:

[0037] It also includes a range management unit that automatically plans the optimal stopping path from the masonry break point to the charging point when the battery level drops to a preset warning threshold. The stopping path must simultaneously meet three conditions: not hindering subsequent construction, facilitating the reproduction of the break point path, and avoiding obstacles at the shortest distance. After stopping, it automatically records the masonry status, break point location, and process parameters to provide data support for the continuation of subsequent operations.

[0038] Advantages of this invention:

[0039] According to the present application, a construction robot masonry construction path planning method and control system can be adapted to the high-precision automated masonry construction requirements in confined spaces. It adopts a multimodal perception structure combining a monocular vision sensor, LiDAR and IMU to simultaneously collect spatial boundary, obstacle information and process parameters such as mortar flowability and brick pressing force. It integrates and generates a dynamic constrained three-dimensional point cloud model and maps it to a digital twin scene to construct a physical-virtual two-way synchronous mapping, which solves the problems of perception and process disconnect and insufficient dynamic working condition adaptation in the existing technology, and provides accurate data support for path optimization.

[0040] Based on the improved RRT algorithm, 3-5 candidate paths are generated, and masonry process constraints and action redundancy judgment criteria (redundancy rate ≤15%) are embedded. Combined with digital twin pre-simulation, the collision-free optimal path is screened. At the same time, an energy consumption assessment unit is added to prioritize the selection of paths with energy consumption reduction of more than 15%, so as to achieve a multi-objective balance of obstacle avoidance, process adaptation and energy consumption optimization.

[0041] Bionic joint control is achieved through flexible drive and torque detection, combined with layered joint space planning, multi-joint collaborative fine adjustment replaces large-scale movement of the whole arm, reducing the occupation rate in confined spaces. Furthermore, a three-level deviation correction strategy (corresponding to ≤1mm, 1-2mm, and >2mm deviations) is used to accurately match different deviation scenarios, balancing operational accuracy and efficiency.

[0042] The solution features a pre-set emergency path template. In the event of a failure, the robotic arm will backtrack to a safe position along the minimum damage trajectory within 5 seconds and issue an early warning. All modules form a closed-loop control link, coupled with endurance management and breakpoint resume functions, to ensure the continuity and safety of operations. In addition, the bionic control module has a built-in multi-condition posture template library, which can automatically adapt to the working posture without manual programming. It integrates the functions of the entire process to avoid fragmentation, greatly reduces reliance on manual labor, and helps large-scale automated construction, comprehensively improving the accuracy, stability and economy of masonry operations.

[0043] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0044] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0045] Figure 1 This is a schematic block diagram of a construction robot masonry construction path planning method and control system according to an embodiment of this application;

[0046] Figure 2 This is a schematic flowchart of a construction robot masonry construction path planning method and control system according to an embodiment of this application. Detailed Implementation

[0047] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0048] The following describes, with reference to the accompanying drawings, a construction robot masonry path planning method and control system according to embodiments of this application.

[0049] like Figure 2 As shown in Embodiment 1 of this application, a construction robot masonry path planning method includes the following steps:

[0050] 1) A multimodal perception method is used to scan the narrow masonry work space, collect information on the space boundary, obstacles and the process data of the masonry wall, generate a dynamic constraint 3D point cloud model, and map it to the digital twin scene to build a physical-virtual two-way synchronous mapping model (by building a dynamic constraint model through multimodal perception, the synchronous collection of spatial and process data is realized, providing an accurate basis for subsequent planning).

[0051] 2) Based on the improved RRT algorithm, multiple candidate masonry paths are generated, and masonry process constraints and motion redundancy judgment criteria are embedded. The entire process of robotic arm masonry action is simulated in the digital twin scenario, and the optimal path with no collision, meeting the redundancy rate and process requirements is selected (by using the improved RRT algorithm in combination with digital twin pre-simulation, the obstacle avoidance and convergence advantages of the algorithm are retained, and the optimal path is selected through process constraints and redundancy judgment to avoid invalid movement).

[0052] 3) The robotic arm is driven by bionic joint control logic and performs masonry work along the optimal path by combining the joint space hierarchical planning strategy. At the same time, real-time construction data is collected through the multimodal feedback module (bionic joint control and hierarchical planning are adopted to adapt to the characteristics of narrow space operation and reduce the risk of collision).

[0053] 4) Based on the real-time construction data, the subsequent masonry path is dynamically corrected to adapt to the deviation of masonry working conditions and changes in process parameters. At the same time, energy consumption optimization mechanism and emergency path mechanism are integrated to ensure the efficiency, quality and safety of masonry operations (the fourth step uses dynamic correction, energy consumption optimization and emergency mechanism to deal with working condition fluctuations and sudden failures to ensure the stability of the entire process).

[0054] like Figure 2 As shown:

[0055] The multimodal perception method described in step 1) adopts a combination structure of monocular vision sensor, lidar and inertial measurement unit to synchronously collect core parameters of masonry process. The core parameters include at least mortar fluidity and brick pressing force data, so as to realize the synchronous perception and fusion storage of spatial information and process parameters.

[0056] like Figure 2 As shown:

[0057] In step 2), based on the size of the work area and the distribution of obstacles, 3-5 candidate masonry paths are generated. The redundancy judgment standard is set to the action redundancy rate not exceeding 15%. Through full-process simulation of the digital twin scenario, candidate paths with collision risks and excessive redundancy rates are eliminated. The selected optimal path is simultaneously sent to the construction robot execution end and the digital twin module for filing.

[0058] In the above process, a monocular vision sensor can quickly collect visual information such as wall texture and brick position, while lidar has high-precision spatial ranging capabilities and can accurately capture spatial boundaries and obstacle outlines. An inertial measurement unit can monitor the posture and motion trajectory of the robotic arm in real time. The combination of the three forms a three-dimensional perception system of vision, distance, and posture, which breaks through the limitation of existing perception that only focuses on spatial information. It simultaneously collects process parameters such as mortar fluidity and brick pressing force, realizing the fusion storage and synchronous transmission of spatial data and process data. This ensures that the path planning not only adapts to the spatial environment but also conforms to the requirements of the masonry process, reducing the deviation between the path and the actual operation from the source.

[0059] like Figure 2 As shown:

[0060] The bionic joint control logic described in step 3) is achieved through the collaboration of flexible drive components and high-precision torque detection elements, driving the robotic arm joints to perform micron-level high-precision micro-adjustments. By replacing the large-scale movement of the entire arm with the coordinated movement of multiple joints, the occupation rate and collision risk in confined spaces are significantly reduced.

[0061] In the above process, considering the working range and obstacle distribution characteristics of masonry in confined spaces, the number of candidate paths is limited to 3-5. This avoids both insufficient paths to select the optimal solution and excessive paths that increase the computational load and rehearsal time of the algorithm. Setting a 15% motion redundancy rate threshold is based on the joint motion characteristics of the robotic arm and the requirements of the masonry process. Too low a redundancy rate will result in insufficient motion flexibility and difficulty in coping with minor changes in working conditions, while too high a redundancy rate will cause ineffective motion and energy waste. Through full-process simulation of the digital twin scenario, the collision risk and redundancy rate of each path can be intuitively detected, and substandard paths can be eliminated to ensure that the paths issued to the execution end are safe, efficient, and adaptable to the process.

[0062] like Figure 2 As shown:

[0063] The deviation classification correction mentioned in step 4) specifically involves dividing the masonry deviation into three levels: slight deviation, moderate deviation, and severe deviation. Slight deviation corresponds to a deviation value ≤1mm, moderate deviation corresponds to a deviation value of 1-2mm, and severe deviation corresponds to a deviation value >2mm. For different levels of deviation, a classification correction strategy is triggered, which includes path fine-tuning, single-brick re-laying path planning, and local area path backtracking, to ensure that the deviation is controllable.

[0064] In the above process, drawing on the logic of human joint coordinated movement, flexible drive components are used to replace traditional rigid drives, reducing the impact and vibration during the movement of the robotic arm. With the addition of high-precision torque detection elements, the force on the joints can be fed back in real time, enabling precise control of joint movement. To address the problem of limited range of motion in confined spaces, multi-joint coordinated fine-tuning is used instead of whole-arm movement. This reduces the space occupied by the robotic arm and improves the accuracy of masonry posture adjustment. It is especially suitable for masonry operations in complex locations such as around door and window openings and corners. At the same time, it reduces wear on the robotic arm joints and extends the service life of the equipment.

[0065] like Figure 1 As shown:

[0066] The energy consumption optimization mechanism described in step 4) is achieved by adding an energy consumption evaluation unit in the digital twin pre-simulation stage. It quantifies the joint movement energy consumption, motor load and operation time of each candidate path, and prioritizes the selection of paths whose energy consumption is reduced by more than 15% compared with the benchmark path and meets the masonry process requirements, taking into account both efficiency and energy saving.

[0067] In the above process, based on the accuracy requirements of masonry work (the industry standard for high precision is ±2mm), the deviation is divided into three levels: slight deviation ≤1mm, which can be corrected by minor path adjustments without interrupting the work; moderate deviation 1-2mm (excluding 1mm and including 2mm), which requires planning a path for re-laying individual bricks to avoid the accumulation of deviation; and severe deviation >2mm, which requires backtracking of the path in the local area to thoroughly correct the deviation to ensure the overall quality of the wall. The graded correction strategy can be flexibly adjusted according to the severity of the deviation, avoiding overcorrection of slight deviations that would lead to a decrease in work efficiency, and also preventing insufficient correction of severe deviations that would leave quality hazards, thus achieving a dynamic balance between accuracy and efficiency.

[0068] like Figure 2 As shown:

[0069] The emergency path mechanism described in step 4) pre-sets emergency path templates for various sudden faults (including equipment faults and process faults). The sudden faults include at least brick jamming, mortar blockage, and sensor faults. After the fault is triggered, the robotic arm backtracks to the preset safe position along the minimum damage trajectory. The path backtracking time is strictly controlled within 5 seconds. At the same time, an audible and visual warning signal is triggered and the fault information is uploaded.

[0070] An energy consumption assessment unit is added to the digital twin pre-simulation stage. By simulating the process of the robotic arm moving along different candidate paths, the joint motion energy consumption, motor load and operation time of each path are quantitatively calculated. A dual assessment standard of energy consumption and process adaptability is established. A threshold of more than 15% reduction in energy consumption is set. This is based on the energy consumption benchmark of existing masonry robots and the optimization space is determined by combining the improved path algorithm. This can achieve significant energy-saving effect without sacrificing the accuracy of masonry process and operation efficiency in order to excessively pursue energy consumption reduction.

[0071] like Figure 1 As shown, Example 2:

[0072] The construction robot masonry construction path planning and control system includes:

[0073] The sensing module is used to perform multimodal sensing scanning to collect information on confined spaces, masonry process parameters, and real-time construction data.

[0074] The digital twin module is used to build a physical-virtual synchronous mapping model to complete the virtual pre-playing, screening and dynamic updating of candidate paths;

[0075] The path planning module is used to run the improved RRT algorithm to generate candidate paths, and combine redundancy determination and energy consumption assessment to complete the selection and distribution of the optimal path;

[0076] The biomimetic control module is used to drive the coordinated movement of the robotic arm based on biomimetic joint control logic, and to perform joint space hierarchical trajectory planning and work posture adaptation.

[0077] The feedback correction module is used to realize dynamic path adjustment, graded correction of masonry deviation, and emergency path triggering based on multimodal real-time data.

[0078] In the above modules, the perception module provides input data to the system, covering three core types of information: space, process, and real-time operation; the digital twin module builds a virtual simulation carrier to realize path pre-simulation and dynamic updates, replacing physical trial and error; the path planning module, as the core computing unit, runs improved algorithms and integrates multi-objective selection logic; the bionic control module, as the execution terminal driver, ensures accurate path implementation; the feedback correction module realizes dynamic regulation, corrects deviations and triggers emergency response. The modules form a closed loop through signal interaction, ensuring that the data of each link can back-optimize the preceding links. At the same time, the integration of full-function modules avoids manual intervention and realizes fully automated control from data acquisition to operation execution.

[0079] like Figure 2 As shown:

[0080] The bionic control module has multiple built-in core masonry posture template libraries. Based on the pre-simulation results of the digital twin module, it can automatically match and adapt the working posture to the current confined space conditions. The template library covers the core working postures of typical confined space conditions such as corner masonry, wall patching, and masonry around door and window openings. All of these postures have been determined through extensive testing and simulation optimization, and have high precision and high adaptability. During the pre-simulation of the digital twin module, it will simultaneously analyze the current working conditions (such as space size, obstacle position, and masonry progress) and automatically match the optimal working posture from the template library. This eliminates the need for manual programming and adjustment of the robotic arm posture by the staff, greatly reducing the complexity of operation, improving the equipment's ability to quickly adapt to different working conditions, and meeting the needs of large-scale automated construction.

[0081] like Figure 1 As shown:

[0082] The system also includes a battery life management unit. When the battery level drops to a preset warning threshold, it automatically plans the optimal stopping path from the masonry break point to the charging point. The stopping path must simultaneously meet three conditions: not hindering subsequent construction, facilitating the reproduction of the break point path, and avoiding obstacles at the shortest distance. After stopping, it automatically records the masonry status, break point location, and process parameters to provide data support for the continuation of subsequent operations.

[0083] The battery management unit monitors the battery level in real time, sets a warning threshold (not less than 20% of the total battery level), and reserves sufficient power for the robotic arm to plan its docking path and move to the charging point. This ensures that the robotic arm does not obstruct subsequent construction (avoiding core masonry areas and passages), facilitates path reproduction (the docking location has distinctive spatial features, making it easy to quickly locate the breakpoint after charging), and minimizes obstacle avoidance (reducing energy consumption and time consumption during docking). After docking, the robotic arm automatically records the masonry status (including breakpoint location, masonry progress, and process parameters). After charging is completed, the robotic arm can directly reproduce the work status based on the records without replanning the path or adjusting process parameters, ensuring continuous and uninterrupted operation.

[0084] Example 3, illustrated below with a specific case:

[0085] Taking the construction of interior partition walls in residential projects as an application scenario, the construction area is a small space with a length of 3m, a width of 2.5m, a height of 2.8m, and a wall thickness of 240mm. Autoclaved aerated concrete blocks with dimensions of 600mm×240mm×200mm and special masonry mortar with a strength grade of M7.5 are used. The masonry accuracy is required to be ±1.5mm. The energy consumption of the operation is reduced by more than 15% compared with traditional robots, and the fault response time is ≤5 seconds.

[0086] First, a 6-axis bricklaying robot is used, equipped with the various modules of the control system proposed in this solution. The perception module consists of a monocular vision sensor (OV7725), a lidar (YDLIDAR G4), and an IMU (MPU6050). The bionic control module is equipped with a flexible drive component (model FD-20) and a torque detection element (accuracy ±0.01N·m). The battery life unit is a 48V / 20Ah lithium battery, with a charging time of 2 hours and a full charge life of 8 hours. The multimodal perception module is jointly calibrated, and the external parameter calibration error of the lidar and vision sensor is ≤0.5mm, and the calibration deviation of the IMU and the joint posture of the robotic arm is ≤0.1°. The CAD model of the work space is imported into the digital twin scene, and process parameters such as block size, mortar flowability (preset 180±5mm), and pressing force (0.3-0.5MPa) are entered to construct a physical-virtual synchronous mapping model with a scene rendering accuracy of 0.1mm.

[0087] Then proceed with the specific operations:

[0088] 1) The perception module is activated to scan the confined working space. The monocular vision sensor collects the baseline of the wall construction and the texture information of the block stacking position. The lidar quickly captures the spatial boundary and the outline of the beam and column obstacles. The obstacle dimensions are 200mm×300mm×2800mm and 150mm×250mm×2800mm, respectively. The IMU records the robot base posture and the initial position of the robotic arm at the same time. The measured value of mortar fluidity is 178mm and the benchmark value of brick pressing force is 0.4MPa. The dynamic constraint three-dimensional point cloud model is generated by fusion. The point cloud density is 100 points / mm². After mapping to the digital twin scene, the positional deviation between the virtual model and the physical space is ≤0.3mm, realizing the synchronous association between spatial information and process parameters.

[0089] 2) The path planning module runs an improved RRT algorithm with a sampling step size of 0.05m and a target bias factor of 0.8, generating four candidate masonry paths (path lengths of 4.2m, 4.5m, 3.9m, and 4.1m, respectively). It embeds process constraints and motion redundancy judgment criteria (redundancy rate ≤15%), simulating the entire process of the robotic arm operation in a digital twin scenario. The pre-simulation time for a single path is 8 seconds. The energy consumption assessment unit quantifies the energy consumption of each path: Path 1 joint energy consumption 120Wh, Path 2 135Wh, Path 3 102Wh, and Path 4 118Wh. Redundancy rate detection results: Path 1 13%, Path 2 16% (exceeding the standard), Path 3 12%, and Path 4 14%. (Combined with collision detection), Path 3 is ultimately selected as the optimal path, with its energy consumption reduced by 24.4% compared to the baseline path (Path 2), meeting the energy consumption optimization requirements. This is then simultaneously sent to the robot execution end.

[0090] 3) The bionic control module automatically matches the masonry posture for tight spaces and corners from the built-in posture template library (containing 20 typical working condition postures) based on the optimal path. No manual programming is required. The robot arm is controlled in collaboration with flexible drive components and torque detection elements. A joint space hierarchical planning strategy is adopted. The bottom joints (axis 1-3) are responsible for coarse positioning with a motion error of ≤0.2mm. The top joints (axis 4-6) are fine-tuned at the micrometer level with an adjustment accuracy of 0.01mm. Multi-joint collaborative motion replaces large-scale movement of the whole arm, reducing the working space occupancy rate by 30% compared with traditional rigid control.

[0091] During the masonry process, the robotic arm operates according to the process of picking up bricks, dipping in mortar, positioning, and pressing. The pressing force is fed back in real time and stabilized at 0.4±0.02MPa. The mortar thickness is controlled at 11±0.5mm. The time taken to lay a single brick is 12 seconds, and the work efficiency reaches 300 bricks / hour.

[0092] 4) The multimodal feedback module collects real-time data every 2 seconds and monitors 3 types of deviations. After the 50th brick is laid, the deviation value is 0.6mm (slight deviation). The system automatically triggers a path micro-adjustment with an adjustment amount of 0.08mm without interrupting the operation. Due to fluctuations in mortar fluidity, the deviation value of the 120th brick is 1.6mm (moderate deviation). A path for re-laying a single brick is planned, which takes 20 seconds. After the deviation is corrected, it is reduced to 0.4mm. When laying the 200th brick, sensor data anomalies occur (slight malfunction). The emergency path mechanism is triggered. The robotic arm backtracks to a safe position along the minimum damage trajectory, which takes 4.2 seconds. At the same time, an audible and visual warning is triggered. The staff found that the sensor was blocked by dust. After cleaning, the system reproduces the operation based on the records. The deviation of the breakpoint reconnection is ≤0.5mm.

[0093] 5) After 6.4 hours of operation, the battery level drops to the 20% warning threshold. The endurance management unit automatically plans a docking path (2.8m in length, avoiding the core masonry area, with the docking point 1.5m from the charging point). The robotic arm takes 15 seconds to move to the docking position. It automatically records the breakpoint location (wall height 1.6m, 8th column from the left), the masonry progress (208 bricks, 65% complete), and the current process parameters. After 2 hours of charging, the system quickly reproduces the breakpoint state with a path reproduction accuracy of ±0.3mm. The masonry continues until completion with no obvious joint deviation.

[0094] In the description of this specification, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0095] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0096] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for planning a construction path for a building robot, characterized in that, Includes the following steps: 1) A multimodal perception method is used to scan the narrow masonry work space, collect information on the space boundary, obstacles and the process data of the masonry wall, generate a dynamic constrained three-dimensional point cloud model, and map it to the digital twin scene to build a physical-virtual two-way synchronous mapping model. 2) Based on the improved RRT algorithm, multiple candidate masonry paths are generated, and masonry process constraints and motion redundancy judgment criteria are embedded. The entire process of robotic arm masonry action is simulated in the digital twin scenario, and the optimal path with no collision, meeting the redundancy rate and process requirements is selected. 3) The robotic arm is driven by bionic joint control logic and performs masonry work along the optimal path by combining the joint space hierarchical planning strategy, while simultaneously collecting real-time construction data through the multimodal feedback module; 4) Based on the real-time construction data, the subsequent masonry path is dynamically corrected to adapt to the deviation of masonry working conditions and changes in process parameters. At the same time, energy consumption optimization mechanism and emergency path mechanism are integrated to ensure the efficiency, quality and safety of masonry operations.

2. The construction path planning method for building robots as described in claim 1, characterized in that, The multimodal perception method described in step 1) adopts a combination structure of monocular vision sensor, lidar and inertial measurement unit to synchronously collect core parameters of masonry process. The core parameters include at least mortar fluidity and brick pressing force data, so as to realize the synchronous perception and fusion storage of spatial information and process parameters.

3. The construction path planning method for building robots as described in claim 1, characterized in that, In step 2), based on the size of the work area and the distribution of obstacles, 3-5 candidate masonry paths are generated, and the redundancy judgment criterion is set to an action redundancy rate of no more than 15%. By simulating the entire process of a digital twin scenario, candidate paths with collision risks and excessive redundancy are eliminated, and the selected optimal path is simultaneously sent to the construction robot execution end and the digital twin module for filing.

4. The construction path planning method for building robots as described in claim 1, characterized in that, The bionic joint control logic described in step 3) is achieved through the collaboration of flexible drive components and high-precision torque detection elements, driving the robotic arm joints to perform micron-level high-precision micro-adjustments. By replacing the large-scale movement of the entire arm with the coordinated movement of multiple joints, the occupation rate and collision risk in confined spaces are significantly reduced.

5. The construction path planning method for building robots as described in claim 1, characterized in that, The deviation classification correction mentioned in step 4) specifically involves dividing the masonry deviation into three levels: slight deviation, moderate deviation, and severe deviation. Slight deviation corresponds to a deviation value ≤1mm, moderate deviation corresponds to a deviation value of 1-2mm, and severe deviation corresponds to a deviation value >2mm. For different levels of deviation, a classification correction strategy is triggered, which includes path fine-tuning, single brick re-laying path planning, and local area path backtracking, to ensure that the deviation is controllable.

6. The construction path planning method for building robots as described in claim 1, characterized in that, The energy consumption optimization mechanism described in step 4) is implemented by adding an energy consumption assessment unit in the digital twin pre-simulation stage. This unit quantifies the joint motion energy consumption, motor load, and operation time of each candidate path, and prioritizes the selection of paths whose energy consumption is reduced by more than 15% compared to the benchmark path and meets the masonry process requirements, thus balancing efficiency and energy saving.

7. The construction path planning method for building robots as described in claim 1, characterized in that, The emergency path mechanism described in step 4) pre-sets emergency path templates for various sudden faults (including equipment faults and process faults). The sudden faults include at least brick jamming, mortar blockage, and sensor faults. After the fault is triggered, the robotic arm backtracks to the preset safe position along the minimum damage trajectory. The path backtracking time is strictly controlled within 5 seconds. At the same time, an audible and visual warning signal is triggered and the fault information is uploaded.

8. The construction robot masonry path planning and control system according to claim 1, characterized in that, include: The sensing module is used to perform multimodal sensing scanning to collect information on confined spaces, masonry process parameters, and real-time construction data. The digital twin module is used to build a physical-virtual synchronous mapping model to complete the virtual pre-playing, screening and dynamic updating of candidate paths; The path planning module is used to run the improved RRT algorithm to generate candidate paths, and combine redundancy determination and energy consumption assessment to complete the selection and distribution of the optimal path; The biomimetic control module is used to drive the coordinated movement of the robotic arm based on biomimetic joint control logic, and to perform joint space hierarchical trajectory planning and work posture adaptation. The feedback correction module is used to achieve dynamic path adjustment, graded correction of masonry deviation, and emergency path triggering based on multimodal real-time data.

9. The construction robot masonry construction path planning and control system according to claim 8, characterized in that, The bionic control module has multiple core masonry posture template libraries built in, which can automatically match and adapt the working posture to the current confined space conditions based on the pre-simulation results of the digital twin module.

10. The construction robot masonry path planning and control system according to claim 8, characterized in that, It also includes a range management unit that automatically plans the optimal stopping path from the masonry break point to the charging point when the battery level drops to a preset warning threshold. The stopping path must simultaneously meet three conditions: not hindering subsequent construction, facilitating the reproduction of the break point path, and avoiding obstacles at the shortest distance. After stopping, it automatically records the masonry status, break point location, and process parameters to provide data support for the continuation of subsequent operations.

Images