Brick paving robot equipment for pedestrian path and construction method

Abstract

The invention provides brick paving robot equipment for a pedestrian path. The brick paving robot equipment comprises a movable chassis, a paving executing mechanism, a feeding assembly, a detecting and positioning module, a control and communication module and an energy assembly. Wherein the movable chassis is used for moving a involved pen on the ground; the paving execution mechanism is used for realizing automatic grabbing, accurate alignment, flat paving and intelligent compaction of bricks; the feeding assembly is used for automatically providing raw materials for the paving executing mechanism. The detection positioning module is used for realizing centimeter-level path positioning and construction environment identification; the control and communication module is used for path planning control, task scheduling, abnormity identification and data transmission; the energy assembly is used for providing energy for work of the equipment. According to the brick paving robot equipment for the pedestrian path and the construction method, the construction automation degree is remarkably improved, and the manual labor intensity is relieved; the invention further provides a specific construction method of the equipment.

Description

Technical Field

[0001] This invention belongs to the field of construction equipment and intelligent control technology in building engineering, and more specifically, it relates to a brick-laying robot device and construction method for pedestrian walkways. Background Technology

[0002] Currently, urban pedestrian walkways and plaza paving projects generally employ manual or semi-mechanized methods. Manual paving suffers from high labor intensity, low construction efficiency, and low paving precision; while existing semi-automatic machinery is mostly used for material handling and lacks precise control over the paving process.

[0003] Furthermore, pedestrian walkways are often characterized by non-linearity, meandering, and varying slopes, making it difficult for traditional construction equipment to achieve automated path planning and stable operation. Additionally, manual construction poses safety hazards in high-temperature, low-temperature, or dusty environments.

[0004] Therefore, there is an urgent need for a brick-laying robot and construction method that can solve the problems of existing technologies, such as reliance on manual labor for pedestrian walkway paving, low construction efficiency, difficulty in guaranteeing paving accuracy, and lack of intelligent monitoring and adaptive adjustment. Summary of the Invention

[0005] This invention provides a brick-laying robot device and construction method for pedestrian walkways. It aims to provide an intelligent robot device that can automatically pick up bricks, align and lay them, level and compact the ground, and autonomously plan the path, so as to realize the automation and intelligence of pedestrian walkway paving. This solves the technical problems of existing technologies, such as reliance on manual labor, low construction efficiency, difficulty in guaranteeing paving accuracy, and lack of intelligent monitoring and adaptive adjustment.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is: to provide a brick-paving robot device for pedestrian walkways, comprising: a mobile chassis, a paving execution mechanism, a material feeding component, a detection and positioning module, a control and communication module, and an energy component; wherein, The mobile chassis is used to move the pen on the ground; The paving execution mechanism is used to realize the automatic gripping, precise alignment, flat laying and intelligent compaction of bricks; The feeding assembly is used to automatically supply raw materials to the paving execution mechanism; The detection and positioning module is used to achieve centimeter-level path positioning and construction environment identification; The control and communication module is used for path planning control, task scheduling, anomaly detection, and data transmission. The energy component is used to provide energy to the device.

[0007] Preferably, the detection and positioning module includes: The GNSS unit is mounted on the top of the mobile chassis; UWB units are deployed at the construction site; The IMU module is installed on the mobile chassis; A lidar is installed on the front and / or top of the robot body. The camera includes a front-facing camera, a close-up camera, and a depth camera; wherein, the front-facing camera is mounted on the paving execution mechanism and is used to identify the paving position; the close-up camera and the depth camera are mounted on the end of the robotic arm of the paving execution mechanism and are used for brick joint identification and alignment.

[0008] Preferably, the paving execution mechanism includes a brick-picking robotic arm, the end of which is provided with a vacuum adsorption unit and / or a mechanical clamping unit to achieve adaptive gripping of bricks of different sizes.

[0009] Preferably, the paving execution mechanism includes a compaction device; the compaction device includes a vibratory compaction plate and a force control actuator, and the compaction device is used to automatically adjust the compaction force according to the hardness of the foundation.

[0010] A construction method for a paving robot for pedestrian walkways, characterized by: based on the paving robot equipment for pedestrian walkways as described in any of the above claims, comprising the following steps: Obtain digital terrain and paving design maps of the construction area, and import the construction tasks into the scheduling system; Before construction, the brick-laying robot performs hardware self-checks, sensor calibrations, and initializes the positioning reference system. Based on the imported digital map, site constraints and construction strategies, the path planning module generates a construction path covering the entire paving area, while decomposing the area into several paving segments or paving pieces, and generating a specific sequence and robotic arm operation posture sequence for each segment. By combining multi-source localization with real-time pose fusion, the robot platform's centimeter-level position and orientation, as well as the relative reference pose of the brick-picking robotic arm's end effector, are output in real time. Brick supply and handling; Combined visual and mechanical alignment correction; After the placement is completed, compact the brick surface according to the preset pressure and time parameters; After the paving is completed, a quality inspection will be conducted on the paved area. Preferably, the step of acquiring the digital terrain and design paving map of the construction area and importing the construction task into the scheduling system includes: collecting data on the terrain of the construction area through laser scanning and / or manual surveying.

[0011] Preferably, the paving robot performs hardware self-checks, sensor calibrations, and initializes the positioning reference system before construction, including the following steps: Select and deploy reference standards; The robot is parked at the predetermined initialization position, the chassis steady-state attitude adjustment is initiated, and the IMU baseline attitude is recorded; the robot scans and processes the deployed ground reference points or control points to obtain their observation coordinates in the robot's local coordinate system; Coordinate system registration; Baseline orientation and altitude calibration.

[0012] Preferably, the combined visual and force-sensory alignment correction includes: after the brick is placed in an approximate position, the visual correction unit identifies the target baseline, the pose of adjacent bricks, and the gap deviation, while the force sensor monitors the contact force; the alignment algorithm, which integrates visual and force-sensory information, performs fine-tuning of the horizontal and angular dimensions, ensuring that the gap between the bricks and the height of the adjacent brick planes meet the preset tolerance.

[0013] Preferably, the step of outputting the centimeter-level position and attitude of the robot platform and the relative reference pose of the end effector of the brick-picking robot arm in real time through multi-source positioning and real-time pose fusion includes: continuously collecting multi-source data using GNSS, UWB, IMU, LiDAR and visual odometry during driving and operation; the perception fusion module fuses the data from each sensor; and outputs the centimeter-level position and attitude of the robot platform and the relative reference pose of the end effector of the robot arm in real time.

[0014] Preferably, it also includes: performing real-time distortion correction and illumination compensation on visual data to ensure robust recognition under complex lighting conditions; and / or using a local iterative nearest point or a pose refinement algorithm based on a depth map during alignment correction to output the pose correction amount; and / or in multi-robot collaborative scenarios, the path planning module supports task allocation and work area segmentation to enable multiple robots to work in parallel and avoid conflicts.

[0015] The beneficial effects of the paving robot equipment and construction method for pedestrian walkways provided by this invention are as follows: Compared with the prior art, the paving robot equipment and construction method for pedestrian walkways provided by this invention significantly improve the degree of construction automation and reduce the intensity of manual labor; through multi-source sensor fusion and vision-force alignment, millimeter-level placement accuracy and stable paving quality are achieved; the built-in adaptive control and anomaly recognition mechanism improves on-site response capabilities and reduces rework rates; the data-driven self-learning optimization capability enables the system to continuously improve during long-term use; and it supports multi-machine collaboration and remote visual management, facilitating large-scale engineering promotion and information-based construction management. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A schematic diagram of the structure of a brick-paving robot device for pedestrian walkways provided in an embodiment of the present invention; Figure 2 A schematic diagram of the structure of a brick-picking robotic arm used in a brick-laying robot device for pedestrian walkways, provided in an embodiment of the present invention; Figure 3 A schematic diagram illustrating a construction method for a brick-paving machine used for pedestrian walkways, provided in an embodiment of the present invention; Figure 4 A block diagram illustrating the composition of a paving execution mechanism used in a brick-paving robot device for pedestrian walkways, provided as an embodiment of the present invention; Figure 5 A block diagram illustrating the composition of the control and alignment modules of a brick-laying robot device for pedestrian walkways, provided in an embodiment of the present invention. Figure 6 This is a block diagram of the detection and positioning module used in a brick-paving robot device for pedestrian walkways, provided as an embodiment of the present invention.

[0018] In the diagram: 1. Drive wheel; 2. Brick-picking robotic arm; 3. Vision correction unit; 4. Compaction device; 5. Main control processor; 6. Motion control unit; 7. Wireless communication unit; 8. GNSS unit; 9. UWB unit; 10. IMU module; 11. LiDAR; 12. Camera; 13. Energy component; 14. Material feeding component. Detailed Implementation

[0019] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0020] Please refer to the following: Figures 1 to 6This invention provides a brick-laying robot device for pedestrian walkways. The brick-laying robot device includes: a mobile chassis, a paving execution mechanism, a material feeding component, a detection and positioning module, a control and communication module, and an energy component 13. The paving execution mechanism is used to automatically grasp, precisely align, level, and intelligently compact bricks. The detection and positioning module is used to achieve centimeter-level path positioning and construction environment recognition. The control and communication module is used for path planning control, task scheduling, anomaly recognition, and data transmission. The construction method is used to achieve fully automated operation and remote control. The mobile chassis is used for stable movement of the robot on uneven ground, and the energy component is used to provide energy to the device. The material feeding component is used to automatically supply raw materials to the paving execution mechanism. Specifically, the mobile chassis includes drive wheels 1, a steering mechanism, and a suspension leveling mechanism; the paving execution mechanism includes a brick-grabbing robotic arm 2, a vision correction unit 3, a paving leveling unit, and a compaction device 4; the control and communication module includes a main control processor 5, a motion control unit 6, and a wireless communication unit 7. The detection and positioning module includes a GNSS unit 8, a UWB unit 9, an IMU module 10, a lidar 11, and a camera 12.

[0021] As a specific implementation of the present invention, the mobile chassis preferably adopts differential drive and electronic power steering; the mobile chassis is equipped with working condition sensors (such as tilt sensors and load sensors) for reference by the upper control strategy.

[0022] As one specific implementation of this invention, please refer to the following: Figures 1 to 6The GNSS unit 8 includes an RTK antenna structure and is mounted at the center of the top of the mobile chassis, raised 250mm by a magnetic shield to reduce metal interference. The GNSS unit 8 connects to the main control processor via a CAN or UART interface to provide centimeter-level absolute positioning reference. There are 2-4 UWB units 9. These 2-4 UWB units 9 are deployed on fixed structures or temporary supports at the construction site; one UWB tag is mounted on the robot body, located slightly rear of the chassis center. The UWB units 9 communicate wirelessly with the main control processor to provide relative positioning information. The IMU module 10 is mounted near the center of gravity of the mobile chassis to reduce vibration impact. The IMU module 10 communicates with the main control board via a high-speed SPI interface for real-time attitude compensation and inertial navigation. The LiDAR 11 includes a 2D LiDAR and a 3D LiDAR. The 2D LiDAR is deployed at the front of the robot body for obstacle avoidance. The 3D LiDAR is deployed on the top of the robot body for environmental mapping and contour recognition. The lidar unit 11 is connected to the main control processor via Ethernet or UART. The GNSS unit 8 achieves a positioning accuracy of ≤1 cm in unobstructed scenarios. In obstructed or weak satellite scenarios, centimeter-level positioning accuracy can be maintained through data fusion from the UWB unit 9, IMU module 10, and lidar unit 1. The camera 12 includes a forward-facing camera, a close-up camera, and a depth camera. The forward-facing camera is mounted in the middle of the paving mechanism to identify the paving position; the close-up and depth cameras are mounted at the end of the robotic arm for brick seam recognition and alignment. The camera 12 is connected to the vision processing unit via USB 3.0 or MIPI CSI. The recognition framework of the camera 12 is based on a lightweight convolutional neural network (e.g., a variant of MobileNet) to ensure real-time performance. The camera 12 is used to identify the target placement position, adjacent brick pose features, and seam baseline.

[0023] It should be noted that all the data from the aforementioned sensors are managed using a unified timestamp within the main control processor and fused through the perception fusion module using EKF or graph optimization methods to achieve high-precision pose output. Specifically, the perception fusion unit performs time synchronization and spatial registration on the aforementioned sensor data, uses filtering and optimization algorithms to output real-time pose and local environment map, and provides local position confidence estimation to the outside world.

[0024] As one specific implementation of this invention, please refer to the following: Figures 1 to 6 The brick-picking robotic arm 2 adopts a six-degree-of-freedom structure with a repeatability accuracy of ≤±0.5 mm. The end of the robotic arm 2 is equipped with a vacuum suction unit and a mechanical gripping unit, enabling adaptive gripping of bricks of different sizes. Specifically, the vacuum suction unit is a vacuum suction cup, and the mechanical gripping unit is a gripper. The vacuum suction cup and gripper can generate a force of 50-500N.

[0025] In some feasible embodiments, the brick-removing robotic arm 2 includes joint drives, an encoder, and a torque / force sensor. The end effector of the brick-removing robotic arm 2 can be switched between a vacuum suction cup module and a mechanical gripping module to accommodate bricks of different sizes and surface conditions.

[0026] As one specific implementation of this invention, please refer to the following: Figures 1 to 6 The visual correction unit 3 identifies the brick joint baseline based on the image recognition algorithm and achieves high-precision alignment of the bricks through the electronically controlled fine-tuning mechanism.

[0027] As one specific implementation of this invention, please refer to the following: Figures 1 to 6 The compaction device 4 includes a vibratory compaction plate and a force-controlled actuator, used to automatically adjust the compaction force according to the hardness of the foundation. The vibratory compaction plate is installed at the center of the front end of the actuator and is connected by a set of vertical linear guide rails, allowing it to move up and down. The force-controlled actuator drives a lead screw via a servo motor to press down and raise the compaction plate; the actuator is equipped with a force sensor (such as a piezoresistive or strain gauge type) for real-time monitoring of the compaction force. The compaction device 4 can control the compaction process in real-time using a closed-loop system.

[0028] In some feasible embodiments, the compaction device 4 also includes a local adjustment tool for fine-tuning and repairing when local height differences or misalignments are found.

[0029] In some feasible embodiments, the compaction device 4 can be closed-loop regulated in the range of 0–2000 N and supports a safety threshold for force / displacement combined triggering.

[0030] In some feasible embodiments, the compaction device 4 is connected to the main frame of the actuator via an independent mounting base. A flexible vibration isolation structure is used between the vibratory compaction plate and the force control actuator to prevent vibration from affecting the upper-level control accuracy. The force control actuator is connected to the main control processor via a CAN / EtherCAT bus, and the vibratory motor is driven by a PWM controller to achieve closed-loop control of the compaction force.

[0031] As one specific implementation of this invention, please refer to the following: Figures 1 to 6 The feeding assembly 14 includes a brick stacking bin, an automatic feeding belt / tilting mechanism, and a material level detection sensor. The feeding assembly 14 supports two brick feeding methods: on-site material stacking and modular material bins; the bin has an automatic discharge cycle control, which is linked with the robotic arm's gripping cycle to reduce waiting time and improve continuous operation capability.

[0032] In one specific embodiment of the present invention, the control and communication module includes an industrial-grade main control processor (CPU+GPU or embedded x86 / ARM platform), a motion controller (real-time EtherCAT / CanBus driver), a PLC interface (optional), and a wireless communication unit (4G / 5G / Wi-Fi / dedicated short-range communication), used for path planning, motion control, vision processing, anomaly detection, and data interaction with the cloud platform. The main control unit supports fault tolerance mechanisms, task distribution, and log storage functions.

[0033] As a specific embodiment of the present invention, the energy component 13 includes a high-energy-density lithium battery pack and a power management system (BMS) with overcharge / over-discharge / overheat protection; it is also equipped with an emergency stop, safety fence / audible and visual alarm, collision detection and redundant braking system to ensure on-site safety.

[0034] The brick-laying robot device for pedestrian walkways provided by the present invention also includes a work recording and data reporting module. This module is used to record the grasping time, placement coordinates, pressure intensity, image and sensor data of each brick, and periodically or in real time report them to the cloud platform for engineering quality management and subsequent model training.

[0035] The brick-laying robot equipment for pedestrian walkways provided by this invention also includes a multi-machine collaborative scheduling unit, which is used to perform task allocation, work area division and mutual avoidance strategies when multiple brick-laying robots are deployed on site, and supports collaborative control based on distributed ROS or cloud scheduler.

[0036] The brick-paving robot equipment for pedestrian walkways provided by this invention also includes a construction method software module, implemented on the main control unit, which includes: a path planning module (global planning and local obstacle avoidance), a perception fusion module (EKF / Graph-SLAM), a grasping and placement trajectory generation module, a visual / force alignment correction module, a compaction control module (force closed loop), an anomaly recognition module (rules + deep learning), a task scheduling and remote monitoring module, and a self-learning optimization module (for model training and parameter updating).

[0037] The brick-laying robot device for pedestrian walkways provided by this invention employs a layered strategy in its path planning module: globally, it uses map-based path planning (e.g., path generation based on swimlane segmentation and curve fitting), while locally, it uses dynamic A* or constrained MPC with a cost map for real-time obstacle avoidance and smooth trajectory generation. The perception fusion module uses time-synchronized multi-sensor fusion (IMU pre-integration + LiDAR / visual odometry + UWB / GNSS correction), and automatically switches to LiDAR / VO-dominated positioning mode when GNSS is unreliable. The intelligent decision-making module incorporates multiple anomaly detectors (e.g., detection of grasping failure, detection of excessive placement deviation, and detection of compaction anomalies), and triggers different levels of response based on confidence level (local retry, parameter adaptation, and manual reminder). It boasts advantages such as increasing the brick-laying area per hour by at least 200%-300% (compared to traditional manual teams) on straight, standard walkways, while maintaining brick joint errors ≤ ±2 mm and maximum surface height differences ≤ 3 mm.

[0038] This invention also provides a construction method for a brick-paving robot device used in pedestrian walkways, please refer to it as well. Figures 1 to 6 This includes the following steps: Step S1: On-site mapping and task import. Specifically, digital terrain and paving design maps of the construction area are acquired through manual methods or laser scanning / drone aerial surveying, and the construction tasks are imported into the scheduling system.

[0039] The paving design includes brick type, joints, slope, and corners. The construction tasks include the start point, end point, priority areas, brick type and specifications, joint tolerances, and compaction pressure.

[0040] Step S1 includes the following sub-steps: Step S1.1: Topographic data collection of the construction area. This can be achieved through laser scanning and / or supplementary manual surveying.

[0041] The laser scanning method utilizes ground-based mobile LiDAR, 3D scanners, or SLAM LiDAR mounted on the robot to acquire 3D point cloud data of the construction area. The operation process is as follows: Equipment deployment and coordinate benchmark establishment: Several fixed reflective target points are deployed in the construction area, or control points measured by a total station are used as a unified coordinate benchmark. The scanning equipment performs external parameter calibration to ensure that multiple scans belong to the same coordinate system. High-density point cloud data acquisition: The LiDAR completes a 360° scan at a frequency of 5–15Hz to acquire the spatial point cloud of the construction surface, with an optimal point cloud density of 200–800 points / m². Point cloud denoising and filtering: Noise points are filtered using statistical outlier removal or radius filtering methods to remove temporary point cloud interference from on-site workers, vehicles, tools, etc. Ground point extraction: Surface points are extracted from the point cloud using RANSAC plane detection or progressive morphological filtering (PMF) methods to generate original terrain elevation data.

[0042] For key areas with significant changes in edges, corners, and slopes, supplementary mapping can be conducted manually, including the following steps: measuring the elevation of feature points using a level; mapping the boundary lines of the construction area and local brick joint direction reference lines using a total station; and merging the manually mapped data with the laser point cloud through ICP registration to improve accuracy.

[0043] Step S1.2: Construction of the digital terrain model Based on the collected data, the system automatically constructs a digital terrain model (DTM / DEM) of the construction area, including: a) Elevation surface fitting: using the Triangulated Network Integrity (TIN) algorithm or raster interpolation method (IDW / Kriging) to convert the point cloud into a continuous terrain surface model. b) Slope and elevation difference analysis: the system automatically calculates the overall slope value, local slope abrupt change points, and areas with abnormal elevation differences. This data is used for subsequent automatic adjustment of paving strategies. c) Obstacle identification: using geometric features to identify and label road manhole covers, curbs, raised barriers or temporary material stockpiles, and avoidance areas, and record them as "non-construction areas".

[0044] Step S1.3: Digital processing of paving design drawings The paving design can be derived from CAD, BIM, or digital copies of paper drawings. The parsing process includes: a) Drawing format parsing and coordinate unification: automatically identifying brick specifications, extracting paving direction lines and joint direction lines, and converting the drawing coordinates to an engineering coordinate system consistent with the site coordinates. b) Establishing a parametric model of brick joints: the system extracts brick joint width, staggered joint methods, brick shape variations in corner areas, and edge finishing rules for non-rectangular areas, all stored in a structured parametric manner for easy retrieval by the robot.

[0045] Step S1.4, Terrain Adaptation Analysis: The theoretical elevation of the bricks after laying is automatically calculated based on the DEM model, and special areas such as inclined laying areas, cutting areas, and areas requiring manual adjustment of the subbase are identified.

[0046] Step S1.5: Importing Construction Task Parameters The processed construction drawings and terrain model are imported into the scheduling system to form the task initialization configuration, including: setting the path start and end points, marking priority construction areas, importing brick type / brick material parameters, setting quality and tolerance, and setting pressure parameters.

[0047] Step S1.6: Generation of Digital Task Package The system ultimately generates a "paving task package" that can be directly called by the robot, including: a complete 3D terrain model and obstacle model, the theoretical placement position and posture of each brick, the robot's drivable area and drivability constraints, brick joint and staggered joint rule parameters, auxiliary control parameters, construction priority and safety boundary.

[0048] This task package will be used for path planning and execution strategy generation in step S3.

[0049] Step S2: Equipment Self-Check and Initialization. Before construction, the paving robot performs a hardware self-check, sensor calibration, and initializes the positioning reference system (establishing a local coordinate system and acquiring GNSS / base station differential or relative reference points). The hardware self-check includes checking the power, drive, robotic arm joints, suction cups / grippers, sensors, communication, and power supply.

[0050] The self-testing process of this invention is based on existing conventional self-testing methods for industrial robots and mobile robots, but it has been functionally expanded and improved in the following aspects to meet the special requirements of construction sites: Existing application key points: power supply and BMS health checks, drive unit and motor encoder self-tests, communication link tests, sensor connectivity checks, and mechanical component limit position judgments. Improvements of this invention: Environmental adaptive self-test: Added self-testing of the effects of light, dust, temperature and humidity on vision and electronic equipment; Load simulation self-test: Performed a simulated action with typical grasping torque on the robotic arm, measured the deviation of joint current and torque curves from the nominal model, and judged actuator wear or grasping mechanism misalignment; Sensor joint self-test: Detected whether the time synchronization and spatial calibration of multiple sensors were consistent through short-time data fusion. If the error exceeded the limit, an automatic calibration process was executed or manual calibration was prompted; Safety system loop test: Simulated collision / emergency stop triggering to verify the response time and reliability of the brake and emergency power failure system.

[0051] To ensure the accuracy of subsequent path planning and alignment correction, the steps for initializing the positioning reference frame are as follows: Step S2.1: Select and deploy reference benchmarks Select or arrange at least three reasonably distributed ground control points in the construction area. It is recommended that the spacing between the points cover the characteristic scale of the entire construction area. Use a high-precision RTK / GNSS or total station to measure these control points and record their global coordinates into the system.

[0052] Step S2.2: Static positioning and distance measurement of the robot. The robot is parked at the predetermined initialization position, the chassis steady-state attitude adjustment is initiated, and the IMU baseline attitude is recorded. The robot obtains its observation coordinates in the robot's local coordinate system by scanning the deployed ground reference points or control points with an onboard camera or LiDAR.

[0053] Step S2.3, Coordinate System Registration If there are ≥3 matching control points, then solve for rotation R and translation t through rigid body transformation (using least squares / Umeyama algorithm) to achieve least squares registration between the robot's observation point and the known control points: ; If GNSS / RTK is available, the global coordinates of the robot chassis can be provided through RTK to directly establish the transformation relationship; otherwise, visual matching (ICP fine registration between the robot endpoint cloud and the construction point cloud) is used to obtain the transformation matrix.

[0054] Verify the registration results; if they exceed the threshold, trigger automatic retesting or prompt for manual review.

[0055] Step S2.4, Baseline Orientation and Height Calibration Use a positioning antenna or azimuth reference to determine the robot's orientation and align the local X-axis to the north or design direction of the design coordinate system; use a fixed-height antenna / total station or optional laser height measurement module to calibrate the elevation of the robot's worktable to ensure the Z-direction elevation accuracy.

[0056] Step S2.5, Calibration Data Saving and Dynamic Update Strategy Save the registration transformation matrix as the working matrix of the current job session, and automatically trigger micro-calibration periodically during the job or when system drift (increased IMU residual or UWB difference) is detected; to prevent single point failure, it is recommended to reserve several backup reference points on site and implement automatic switching.

[0057] Step S3: Path Planning and Segmented Task Generation. Based on the imported digital map, site constraints, and construction strategy, the path planning module generates the optimal construction path covering the entire paving area. Simultaneously, it decomposes the area into several paving segments or paving pieces, and generates a specific sequence (row / column order, staggered joint order) and robotic arm pose sequence for each segment. This step transforms the imported design drawings and terrain model into a segmented paving task sequence executable by the robot, encompassing path planning, segment division, brick sequence generation within a single segment, and robotic arm end-effector pose calculation. The specific implementation steps are as follows: Step S3.1: Decompose the coverage area (region rasterization / cell division) Project the paved area onto a plane and rasterize it. The grid size can be based on the brick size (such as half the brick length or the brick length as the unit) to facilitate subsequent staggered and aligned strategies. Tagging the grid based on design textures (such as vertical tiling, horizontal tiling, and diagonal tiling) generates tiling templates or "tiling matrices".

[0058] Step S3.2, Tile Block / Strip Strategy Semantic maps are used to identify partition boundaries and apply a segmentation strategy: stripe segmentation is preferred for long, narrow areas; two-dimensional tile blocks are used for complex obstacles or areas with multiple curves. Each block generates a subtask (including sub-start point, end point, brick type, and row priority). When dividing the blocks, try to meet the constraints of the robotic arm's working radius and chassis accessibility to avoid exceeding the robotic arm's workspace.

[0059] Step S3.3: Generation of the laying sequence within a single block (rows / columns and staggered seams) Generate a sequence of individual bricks according to design requirements: for example, the standard staggered rule is half the brick length offset between adjacent rows; this can be described by a parametric formula: the lateral offset of the brick in the r-th row and c-th column. .

[0060] Considering seam tolerances Combined with the boundary trimming strategy (whether to cut bricks), generate cutting paths or arrange gap-retention schemes for boundary bricks.

[0061] Step S3.4, Path Cost Function and Optimization Define the overall cost function J (the objective is to minimize the total operation time, the amount of robot arm movement, and the number of repositioning attempts, while also constraining the paving quality): ; in, To estimate the total operation time, This refers to the change in pose at the end of the robotic arm (distance in joint space or Cartesian space). The weights α, β, and γ represent the number of times repositioning is required and can be adjusted according to the operating conditions.

[0062] Within a single block, an initial trajectory is generated using a heuristic path (lawnmower / boustrophedon). Then, based on the aforementioned cost function, sequential fine-tuning is performed using local search (2-opt, simulated annealing, or local optimization) to reduce the movement of the robotic arm during its non-working period and improve parallel efficiency.

[0063] Step S3.5: Solving the end-effector pose sequence and inverse kinematics (IK) Generate the end-capture pose P for each brick. grab (x,y,z,ϕ,θ,ψ) and placement pose P place (Including fine-tuning of the pose before placement). The corresponding joint angle sequence q=f is obtained by solving using inverse kinematics (IK). IK If IK has no solution or is close to a singular pose, the placement position is automatically adjusted (e.g., offset outward by 20–50 mm on the same plane) and recalculated until the joint limit and obstacle avoidance constraints are met.

[0064] Step S3.6: Trajectory Smoothing and Time Parameterization Cubic spline / Quintic polynomial interpolation is applied to the joint angle sequence to achieve a smooth trajectory, satisfying velocity / acceleration constraints and reducing vibration effects. Time-scaling is applied to generate velocity profiles, ensuring that the end effector grasps / places within a controllable velocity range and is synchronized with the suction cup / gripper movement.

[0065] Step S3.7, Collision Detection and Obstacle Avoidance During the planning phase, volume-based collision detection (using bounding box / convex hull approximation) is performed based on the 3D site model to eliminate or locally replan trajectories with potential risks. Local obstacle avoidance employs a dynamic cost map (occupancy grid) and a local planner (DWA / MPC) to avoid sudden obstacles in real time, while maintaining consistency with the global task.

[0066] Step S3.8: Generate task instructions and issue format Each task segment is encapsulated as a work order (including block ID, brick sequence, end pose list, compaction parameters, time window, task priority, etc.) and uploaded to the scheduler. The scheduler assigns tasks to the current robot instance or multi-robot collaborative system according to priority and supports task rollback and retry strategies (failure count threshold, retry delay, manual intervention).

[0067] Step S3.9, Dynamic Reprogramming Strategy If environmental changes (obstacles, base differences, changes in silo location) are detected during construction, local replanning is triggered: only affected blocks or affected brick sequences are replanned to minimize the amount of redoing; continuous replanning and parallel execution are supported (the current work segment does not affect the completed segment).

[0068] Step S4: Multi-source localization and real-time pose fusion. During the robot's movement and operation, it continuously acquires multi-source data using GNSS (differential or RTK), UWB, IMU, LiDAR, and visual odometry (VO). The perception fusion module fuses the data from each sensor through extended Kalman filtering (EKF) or graph optimization (Graph-SLAM) to output the robot platform's centimeter-level position and attitude, as well as the relative reference pose of the robotic arm's end effector in real time.

[0069] Step S5: Brick Supply and Grabbing. The feeding component delivers bricks to the robotic arm's picking position according to a preset strategy (on-site stacking or automated hopper). The robotic arm (preferably with six degrees of freedom) uses a combination of vacuum adsorption and mechanical clamping, or a switchable gripping strategy, based on the gripping action sequence, in conjunction with end effector force / tactile sensors and vision detection, to quickly and stably grasp bricks of different sizes and surface conditions.

[0070] Step S6: Visual and force-sensing joint alignment correction. After the robotic arm places the brick in an approximate position, the visual correction unit (binocular / structured light depth camera or monocular + laser rangefinder) identifies the target baseline, the pose of adjacent bricks, and the gap deviation, while the force sensor monitors the contact force. The alignment algorithm, which integrates visual and force information (e.g., constrained least squares alignment or deep learning regression), completes millimeter-level horizontal and angular fine adjustments to ensure that the gap between the bricks and the height of the adjacent brick planes meet the preset tolerance (preferably within ±2 mm).

[0071] Step S7: Placement and Compaction. After placement, the compaction unit (including a vibratory compaction plate, a linear servo actuator, and force control feedback) compacts the brick surface according to preset compaction force and time parameters, monitors the compaction force and displacement in real time, and performs secondary compaction or local correction if necessary.

[0072] Step S8: Online Quality Inspection and Recording. After paving is completed, the vision system and flatness sensors (such as laser profilometry) perform quality inspection on the paved area. The inspection items include brick joint width, flatness, height difference, and chamfer position. If the inspection results exceed the preset threshold, the intelligent decision-making module triggers a repair action or a manual intervention command. All construction data (pose, capture log, compaction force, image recording, and anomaly log) are uploaded to the construction cloud platform for traceability and model training.

[0073] Step S9: Anomaly Identification and Adaptive Adjustment. The system identifies construction anomalies (such as brick slippage, jamming, base settlement, and material depletion) based on deep learning or a rule engine, and automatically adjusts the path, captures parameters or pressure according to the anomaly type, or issues remote alarms and downloads on-site operation suggestions.

[0074] Step S10: Work Completion and Self-Learning. After a single or batch of work is completed, the self-learning optimization module updates the control parameters (grabbing strategy, placement trajectory, pressure intensity) and path planning weights based on the construction data to improve the efficiency and quality of subsequent work.

[0075] Preferably, the implementation process of steps S4 to S8 further includes the following refinement steps: Real-time distortion correction and illumination compensation are performed on visual data to ensure robust recognition under complex lighting conditions. The specific steps are as follows: ① Camera distortion correction Cameras typically exhibit radial and tangential distortion, requiring model correction based on intrinsic parameters.

[0076] The commonly used Brown-Conrady model is as follows: Radial distortion:

[0077]

[0078] Tangential distortion:

[0079] The final correction point is: ,

[0080] After correction, the image can be restored to its true proportions, which can be used for subsequent brick joint detection and line extraction.

[0081] ② Illumination compensation and brightness normalization Employing either the Retinex enhancement algorithm or the Adaptive Histogram Equalization (CLAHE) method: For scenes with uneven brightness, CLAHE is used to normalize local brightness and improve edge recognition rate; for shadow scenes, multi-scale Retinex (MSR) is used to separate reflection components and enhance texture details; for nighttime lighting scenes, Gamma correction is used to improve information in low-brightness areas.

[0082] The final output image features uniform brightness and enhanced line details, improving the stability of brick seam recognition.

[0083] (2) For millimeter-level alignment correction, Local Iterative Closest Point (ICP) or a depth map-based pose refinement algorithm is used to output the pose correction amount. The specific details are as follows: Once the brick is moved to its pre-positioned location by the robotic arm, its precise pose relative to the ground and adjacent bricks needs to be aligned to the millimeter level. Specifically, this is achieved as follows: ① Fine-tuning of ICP (Iterative Closest Point) based on local point clouds Local point clouds of the bottom of the current brick and adjacent brick areas are acquired using a depth camera / structured light sensor; the template point cloud of the brick design model is used as a reference; and the pose increment is solved using a point-to-plane ICP model. Error function: ;in: = Brick template dot cloud; = Actual observed point cloud; = The normal vector of the corresponding point.

[0084] The Gauss-Newton iteration is used to solve for ΔR and Δt until the following conditions are met:

[0085] Output:

[0086] Used for final fine-tuning of the end effector of the robotic arm.

[0087] ② Pose Regression Based on Depth Maps (Deep Learning Method) When lighting is insufficient or local geometry is unclear, a lightweight CNN / Transformer model is used to regress pose deviations from the depth map:

[0088] Where D is the depth map; the output is the 6D pose deviation. The model is trained with a large number of real construction scene samples to improve its robustness to dust and shadows.

[0089] (3) Force closed-loop control (such as PI / PID or more advanced adaptive control) is used during the compaction process to ensure that the compaction energy is consistent with the compacted base layer, so as to avoid brick displacement or damage to the base layer; (4) In multi-robot collaborative scenarios, the path planning module supports task allocation and work area division to enable multiple robots to work in parallel and avoid conflicts. The process is as follows: ① Spatial Partitioning The overall construction area is divided into multiple sub-areas, allowing different robots to operate in independent areas and avoiding trajectory overlap.

[0090] Several commonly used strategies: 1. Based on Voronoi cell division: The robot's initial position is used as a seed point to automatically form the nearest neighbor region.

[0091] 2. Based on regional load balancing: The task is divided equally by weighting the number of bricks, the paving area, or the complexity.

[0092] 3. Based on minimizing distance cost: The optimization objective is:

[0093] in Let be the set of bricks assigned to the i-th robot.

[0094] ② Multi-robot task scheduling Scheduling algorithms such as Hungarian matching algorithm, Distributed Auction algorithm, and Centralized Task Allocator are used to allocate paving tasks according to work area priority and distance cost.

[0095] ③ Conflict detection and avoidance Global planning + local obstacle avoidance: Global: Path generation based on A*, RRT*, PRM, etc. Local: Real-time obstacle avoidance based on DWA (Dynamic Window Algorithm) or MPC If the distance between the two robots is less than a threshold, an avoidance mechanism is triggered, and the nearest robot briefly stops or goes around the obstacle. The system ensures a collision-free and highly efficient collaborative construction process.

[0096] Compared with the prior art, the brick-laying method of the present invention has the following beneficial effects: By combining path planning and segmented paving strategies with multi-source high-precision positioning, accurate coverage of complex curves, slopes and irregular construction surfaces is achieved, significantly improving construction efficiency. The combined effect of path planning and segmented paving strategies achieves: minimum robotic arm movement, shortest chassis movement path, and highest coverage efficiency. This is achieved by transforming the paving area into a "coverage path problem," the brick sequence into a "Traveling Worker Problem (TSP)" for optimization, using strip or tile block segmentation for neater geometry and smoother paths, and minimizing operation time and energy consumption through a cost function.

[0097] Ultimately, this increases construction efficiency by 200% to 300%.

[0098] A combination of visual and tactile alignment and pressure force closed-loop control is adopted to ensure that the gap between brick joints and the flatness of the surface are within strict tolerance range, resulting in stable and excellent paving quality. The benefits of combining multi-source information include: vision is responsible for geometric alignment (detecting the center line of brick joints, calculating horizontal and angular deviations, with alignment accuracy up to ±1–2mm); force sensing is responsible for contact state judgment (detecting minute force changes when a brick touches the bottom, judging whether there are height differences, hollow areas, sharp points, and preventing bricks from being hit or cracked). Combining the two can achieve the following:

[0099] The optimal placement posture is obtained.

[0100] The closed-loop control of compaction intensity ensures constant compaction energy, resulting in uniform compaction, no loosening of bricks, and no damage to the base layer.

[0101] Intelligent anomaly identification and adaptive adjustment reduce the number of manual interventions, enable real-time response to complex construction conditions on-site, and improve construction safety; The data-driven self-learning optimization mechanism can continuously improve the success rate of data capture, compaction parameters, and path planning strategies as construction progresses, achieving long-term improvements in construction efficiency. The self-learning module uses historical construction data to improve the parameter model, and the execution process is as follows: ① Data acquisition and feature extraction: Data acquisition (success rate of grasping, placement deviation, pressure-displacement curve, brick joint accuracy statistics, robotic arm time and energy consumption) is used to generate a training dataset through feature extraction.

[0102] ② Online learning / offline batch training: Can employ reinforcement learning (RL), model predictive control (MPC) parameter self-calibration, and regression model (XGBoost / MLP) parameter optimization; Updates include: optimal gripping force / clamping force, optimal pressure intensity, optimal path weights, and optimal visual thresholds (edge ​​detection parameters). ③ Parameter updates and strategy adjustments The updated parameters will be written into the next cycle of construction tasks, making the system "more accurate with use" and continuously improving construction efficiency.

[0103] In this construction method, the visual correction unit identifies the brick joint baseline based on an image recognition algorithm. The specific implementation steps are as follows: (1) Image acquisition and preprocessing Ground images are acquired using an end-view vision system, and then illumination compensation, image grayscale conversion, and edge enhancement are performed.

[0104] (2) Edge extraction The Canny algorithm is used to extract edge features from the processed image to obtain the brick seam outline segments.

[0105] (3) Line detection The Hough transform is used to detect the straight lines containing the brick joints; for irregular paving, the least squares line fitting algorithm is used. ; Used to determine the center line of the brick joint.

[0106] (4) Analysis of gap direction and offset The angle is calculated by comparing the detected brick joint line segment direction vector v with the preset design direction v0:

[0107] To identify the deflection angle; The offset is obtained by calculating the distance between the actual center line of the brick joint and the theoretical center line.

[0108] (5) Generation of position adjustment amount Based on the offset and angle error, generate adjustment commands for Δx, Δy, and Δθ; The position compensation of the end effector is performed before placement using an electronically controlled fine-tuning mechanism.

[0109] This process ensures that the brick alignment accuracy reaches the millimeter level. This brick-laying robot for pedestrian walkways integrates multi-sensor high-precision positioning, a six-degree-of-freedom brick-picking and placement actuator, a vision and force-sensing joint alignment correction, a compaction and quality inspection unit, and an intelligent control platform based on multi-level control algorithms and self-learning optimization. It achieves fully automated operation from path planning, brick picking, alignment, laying to compaction and quality assessment, significantly improving paving efficiency and quality, reducing labor intensity, and enhancing construction safety.

[0110] This invention provides a brick-laying robot device and construction method for pedestrian walkways, which can automatically pick up bricks, align and lay them, level and compact the ground, and autonomously plan the path. Compared with the prior art, through mechanical structure optimization and multi-sensor fusion technology, it achieves high-precision, unmanned and efficient construction of the walkway paving process, significantly reduces the intensity of manual labor, and improves the paving quality and construction safety.

[0111] Specifically, this brick-laying robot for pedestrian walkways significantly improves the level of construction automation and reduces the intensity of manual labor; through multi-source sensor fusion and vision-force alignment, it achieves millimeter-level placement accuracy and stable paving quality; the built-in adaptive control and anomaly recognition mechanism improves on-site response capabilities and reduces rework rates; the data-driven self-learning optimization capability enables the system to continuously improve during long-term use; and it supports multi-machine collaboration and remote visual management, facilitating large-scale engineering promotion and information-based construction management.

[0112] The present invention also provides an electronic device, which includes a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor. The transceiver, the memory, and the processor are connected via the bus. When the computer program is executed by the processor, it implements the steps in the brick-laying robot operation method described above (including but not limited to path planning, perception fusion, grasping / placement action generation, visual / force alignment correction, compaction control, anomaly recognition, and self-learning optimization).

[0113] The present invention also provides a computer-readable storage medium having stored thereon a computer program executable by a processor, wherein the computer program, when executed by the processor, implements the steps in the brick-laying robot operation method described in any of the preceding claims. The storage medium can be used as a robot controller, a cloud training server, or an edge computing node to support offline training and online inference.

[0114] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A paving robot for pedestrian walkways, characterized in that, include: Mobile chassis, paving actuator, material supply assembly, detection and positioning module, control and communication module, and energy assembly; among which, The mobile chassis is used to move the pen on the ground; The paving execution mechanism is used to realize the automatic gripping, precise alignment, flat laying and intelligent compaction of bricks; The feeding assembly is used to automatically supply raw materials to the paving execution mechanism; The detection and positioning module is used to achieve centimeter-level path positioning and construction environment identification; The control and communication module is used for path planning control, task scheduling, anomaly detection, and data transmission. The energy component is used to provide energy to the device.

2. The brick-paving robot equipment for pedestrian walkways as described in claim 1, characterized in that: The detection and positioning module includes: The GNSS unit is mounted on the top of the mobile chassis; UWB units are deployed at the construction site; The IMU module is installed on the mobile chassis; A lidar is installed on the front and / or top of the robot body. The camera includes a front-facing camera, a close-up camera, and a depth camera; wherein, the front-facing camera is mounted on the paving execution mechanism and is used to identify the paving position; the close-up camera and the depth camera are mounted on the end of the robotic arm of the paving execution mechanism and are used for brick joint identification and alignment.

3. The brick-paving robot equipment for pedestrian walkways as described in claim 2, characterized in that: The paving execution mechanism includes a brick-picking robotic arm, the end of which is equipped with a vacuum adsorption unit and / or a mechanical clamping unit to achieve adaptive gripping of bricks of different sizes.

4. The brick-paving robot equipment for pedestrian walkways as described in claim 3, characterized in that: The paving execution mechanism includes a compaction device; the compaction device includes a vibratory compaction plate and a force control actuator, and the compaction device is used to automatically adjust the compaction force according to the hardness of the foundation.

5. A construction method for a brick-paving robot used in pedestrian walkways, characterized in that: Based on the paving robot equipment for pedestrian walkways as described in any one of claims 1-4, the process includes the following steps: Obtain digital terrain and paving design maps of the construction area, and import the construction tasks into the scheduling system; Before construction, the brick-laying robot performs hardware self-checks, sensor calibrations, and initializes the positioning reference system. Based on the imported digital map, site constraints and construction strategies, the path planning module generates a construction path covering the entire paving area, while decomposing the area into several paving segments or paving pieces, and generating a specific sequence and robotic arm operation posture sequence for each segment. By combining multi-source localization with real-time pose fusion, the robot platform's centimeter-level position and orientation, as well as the relative reference pose of the brick-picking robotic arm's end effector, are output in real time. Brick supply and handling; Combined visual and mechanical alignment correction; After the placement is completed, compact the brick surface according to the preset pressure and time parameters; After the paving is completed, the paved area will be inspected for quality.

6. The construction method of the paving robot for pedestrian walkways as described in claim 5, characterized in that: The process of acquiring digital terrain and paving design maps of the construction area and importing the construction task into the scheduling system includes: collecting data on the terrain of the construction area through laser scanning and / or manual surveying.

7. The construction method of the brick-paving robot equipment for pedestrian walkways as described in claim 6, characterized in that: Before construction, the paving robot performs hardware self-checks, sensor calibrations, and initializes the positioning reference system, including the following steps: Select and deploy reference standards; The robot is parked at the predetermined initialization position, the chassis steady-state attitude adjustment is initiated, and the IMU baseline attitude is recorded; the robot scans and processes the deployed ground reference points or control points to obtain their observation coordinates in the robot's local coordinate system; Coordinate system registration; Baseline orientation and altitude calibration.

8. The construction method of the brick-paving robot equipment for pedestrian walkways as described in claim 7, characterized in that: The combined visual and force-sensory alignment correction includes: after the brick is placed in an approximate position, the visual correction unit identifies the target baseline, the pose of adjacent bricks, and the gap deviation, while the force sensor monitors the contact force; the alignment algorithm, which integrates visual and force-sensory information, completes the horizontal and angular fine-tuning to ensure that the gap between the bricks and the height of the adjacent brick plane meet the preset tolerance.

9. The construction method of the brick-paving robot equipment for pedestrian walkways as described in claim 8, characterized in that: The method of real-time output of the centimeter-level position and attitude of the robot platform and the relative reference pose of the end effector of the brick-picking robot arm through multi-source positioning and real-time pose fusion includes: continuously using GNSS, UWB, IMU, LiDAR and visual odometry to collect multi-source data during driving and operation; the perception fusion module fuses the data from each sensor to output the centimeter-level position and attitude of the robot platform and the relative reference pose of the end effector of the robot arm in real time.

10. The construction method of the brick-paving robot equipment for pedestrian walkways as described in claim 9, characterized in that, Also includes: Real-time distortion correction and illumination compensation are performed on visual data to ensure robust recognition under complex lighting conditions; and / or during alignment correction, local iterative nearest point or depth map-based pose refinement algorithms are used to output pose correction amounts; and / or in multi-robot collaborative scenarios, the path planning module supports task allocation and work area segmentation to enable multiple robots to work in parallel and avoid conflicts.

Images