Small-space steel structure welding and grinding integrated robot and working method thereof

Abstract

The invention belongs to the technical field of steel structure welding and grinding, and relates to a small-space steel structure welding and grinding integrated robot and a working method thereof. The working method comprises the steps that a welding seam position list of a to-be-operated steel structure, space environment information, the current position of a robot and a remotely-issued operation task instruction are obtained, and a collision-free moving track is planned according to the welding seam position list and the space environment information; controlling the robot to move to a first target welding seam position according to the collision-free moving track; collecting a two-dimensional image of a first target welding seam, extracting a welding seam edge graph, and scanning a welding seam area to generate corresponding three-dimensional point cloud data; identifying the welding seam type based on the welding seam edge graph and the three-dimensional point cloud data, and extracting welding seam size parameters; determining an optimal welding track and welding process parameters corresponding to the optimal welding track, and starting welding operation; in the welding process, the posture of the welding gun is continuously corrected so that the welding gun can be always kept perpendicular to the normal direction of the weld surface, and meanwhile welding process parameters are dynamically adjusted according to the welding state monitored in real time.

Description

Technical Field

[0001] This invention belongs to the field of steel structure welding and grinding technology, and particularly relates to an integrated robot for welding and grinding small-space steel structures and its working method. Background Technology

[0002] Welding and post-weld grinding are common construction procedures within the confined spaces of steel structures such as ship cabins, box girder cavities, and equipment foundation components. However, due to the high degree of enclosure and insufficient ventilation in these structures, high temperatures, fumes, and harmful gases can easily accumulate during operations, posing significant safety hazards to personnel working in this environment for extended periods. Furthermore, the limited operating space restricts the range of motion of welding torches and grinding tools, often forcing workers to perform tasks in uncomfortable postures, affecting operational stability and reducing overall work efficiency.

[0003] Furthermore, the compact layout of components and numerous obstructions within small spaces make it difficult to access weld locations, hindering equipment positioning and clamping, and resulting in significant weld accessibility issues. Under these conditions, welding processes are more prone to defects such as incomplete fusion, slag inclusions, or porosity. Uniform heat input distribution is also difficult to maintain, increasing the risk of weld deformation or excessive residual stress. Simultaneously, spatial constraints also complicate quality inspection. Limited visual inspection angles and the difficulty in implementing conventional flaw detection methods such as ultrasonic testing mean that on-site inspection typically relies on post-weld offline testing, hindering timely problem detection and process adjustments.

[0004] Currently, although robotic equipment has begun to be introduced into some welding operations, existing equipment is mostly used in open environments and is not well adapted to space-constrained scenarios. Larger equipment is difficult to enter narrow sections, and welding and grinding units are often on different platforms, making it impossible to complete the preceding and following processes continuously. Repeated positioning can easily lead to error accumulation, affecting the final weld formation. At the same time, there is still a lack of mature and feasible complete technologies for achieving weld recognition, 3D morphology acquisition, trajectory generation, and process monitoring inside complex structures.

[0005] In summary, how to achieve the integration, stabilization, and automation of welding and grinding processes in confined spaces, and solve problems such as poor equipment accessibility, positioning difficulties, inconvenient quality monitoring, and insufficient forming stability, has become an urgent technical problem to be solved. Summary of the Invention

[0006] This invention addresses the problems of poor equipment accessibility, limited manual operation, difficulties in quality monitoring, and low efficiency in process connection during steel structure welding and grinding operations in confined spaces. It proposes an integrated robot for welding and grinding steel structures in small spaces and its working method. By integrating welding, grinding, and multi-source sensing units on a compact platform, it achieves automated collaboration in weld recognition, three-dimensional shape acquisition, trajectory planning, and attitude control. It can complete high-precision welding and post-weld grinding in confined environments, effectively improving weld formation consistency and work efficiency, and providing reliable support for automated manufacturing and maintenance operations inside complex structures.

[0007] To achieve the above objectives, in a first aspect, the present invention provides an integrated method for welding and grinding steel structures in small spaces, comprising: The system obtains a list of weld locations on the steel structure to be worked, spatial environment information, the current position of the integrated welding and grinding robot, and remotely issued work instructions, and plans a collision-free movement trajectory accordingly. The robot is controlled to move to the first target weld position according to a collision-free movement trajectory. Two-dimensional images of the first target weld are acquired and weld edge maps are extracted. At the same time, the weld area is scanned to generate corresponding three-dimensional point cloud data. Weld type identification and weld size parameters extraction based on weld edge map and 3D point cloud data; Based on the weld type and weld size parameters, determine the optimal welding trajectory and its corresponding welding process parameters, and start the welding operation; During the welding process, the welding torch posture is continuously corrected to keep it perpendicular to the normal of the weld surface, and the welding process parameters are dynamically adjusted according to the welding status monitored in real time. The method for acquiring a two-dimensional image of the first target weld and extracting the weld edge image is as follows: The global vision camera is controlled to acquire a two-dimensional image of the first target weld and input into the intelligent recognition module for feature fusion to generate a weld image with complete edges; Multi-scale feature extraction and stepwise spatial reconstruction are performed on weld seam images to obtain different types of image pixels; The image pixels are divided into weld area, base material area and background area, and the weld boundary contour is extracted to generate weld edge map.

[0008] In some embodiments, the method for scanning the weld area to generate corresponding three-dimensional point cloud data is as follows: A laser beam is projected onto the weld surface using a laser profile sensor to form laser stripes in the weld area; The laser contour sensor is controlled to scan along the extension direction of the weld seam. Its built-in CMOS camera continuously acquires multiple frames of laser stripe images during the scanning process and locates the center point of the stripes in each frame. Based on the principle of triangulation and the pre-calibrated internal and external parameters of the CMOS camera, the spatial coordinates of each stripe center point are solved to obtain the set of three-dimensional sampling point coordinates accumulated along the scanning direction on the weld surface. The coordinates of the three-dimensional sampling points are transformed, registered, and stitched together to generate complete three-dimensional point cloud data covering the weld area.

[0009] In some embodiments, the method for extracting weld size parameters is as follows: Based on the acquired 3D point cloud data of the weld, multiple cross-sectional positions are determined along the length of the weld at a preset fixed interval, and the corresponding point cloud contour is extracted at each cross-sectional position. For each point cloud contour, perform curve fitting, and calculate the geometric dimension parameters of the weld at the corresponding cross section based on the fitting results; The geometric dimensional parameters measured at each cross-sectional location are statistically processed to calculate the average, standard deviation, maximum, and minimum values ​​of each parameter, so as to comprehensively describe the dimensional characteristics of the weld.

[0010] In some embodiments, the method for generating the optimal welding trajectory is as follows: Based on the start coordinates, end coordinates, and edge diagram of the weld, several feature points characterizing the geometric features of the weld are extracted, and control points for describing the geometric shape of the weld centerline are selected from the feature points to form a control point set. A cubic Bézier curve is constructed based on the set of control points, and the spatial position and curve shape of the Bézier curve are determined by the spatial coordinates of each control point. By iteratively adjusting the spatial position of the control points, the Bezier curve is fitted to the actual centerline of the weld, thus obtaining the fitted curve of the weld centerline. The fitted curve is discretized at a preset sampling interval to generate a welding path point sequence, and the spatial coordinates of each path point are recorded. For each welding path point, the tangent direction of the fitted curve at that path point is calculated as the direction of the welding torch forward. Combined with the three-dimensional point cloud data of the weld, the normal vector of the weld surface at that welding path point is solved as a reference for the welding torch attitude, so as to constrain the direction and attitude of the trajectory point. Based on the spatial coordinates of the welding path points, the forward direction of the welding torch, and the welding torch posture reference, the corresponding joint angle parameters are solved through the inverse kinematics of the welding robot arm to map the path points to the robot arm movements. The welding path point sequence is integrated with the corresponding joint angle parameters to form a complete welding trajectory, i.e., the optimal welding trajectory.

[0011] In some embodiments, the integrated welding and grinding method for small-space steel structures further includes: After welding is completed, a dynamic grinding trajectory is generated based on the surface formation of the weld and the preset optimal grinding strategy, and the grinding robot arm is driven to perform grinding operations according to the dynamic grinding trajectory. During the grinding process, based on real-time feedback of grinding pressure and surface flatness, the feed rate of the grinding head is automatically corrected and the dynamic grinding trajectory is optimized until the weld surface meets the quality requirements. After completing the current weld, the robot moves to the next weld position along a collision-free trajectory, and completes the work on all welds in sequence.

[0012] In a second aspect, the present invention provides an integrated robot for welding and grinding steel structures in small spaces, used to implement the integrated welding and grinding method for steel structures in small spaces described in the first aspect of the present invention, comprising: Mobile work platform; The welding robotic arm and the grinding robotic arm are symmetrically arranged on both sides of the top of the mobile work platform; The visual sensing module, function control module, intelligent recognition module, and posture control module are installed on the mobile work platform, wherein: The visual sensing module is used to acquire visual information about the seam area; The intelligent recognition module identifies the weld type and location based on the collected visual information and generates corresponding welding and / or grinding operation trajectories. The functional control module coordinates and controls the motion execution process of the welding robot arm, the grinding robot arm and the mobile work platform based on the welding and / or grinding operation trajectory. The attitude control module is used to correct the working attitude of the mobile work platform and the end effector of the robotic arm in real time based on the feedback results of the vision sensing module and the intelligent recognition module.

[0013] In some embodiments, the visual sensing module includes: At least one global vision camera is used to acquire two-dimensional images of the weld. In addition, at least one laser profile sensor is used to acquire three-dimensional topographic information of the weld.

[0014] In some embodiments, the end effector of the welding robotic arm is configured with: Miniature industrial cameras are used to acquire images of the welding process in real time. In addition, an arc length tracking sensor is used to monitor changes in the welding arc length.

[0015] In some embodiments, the end effector of the grinding robotic arm integrates: Force sensors are used to detect grinding contact force in real time; In addition, a laser displacement sensor is used to detect the flatness of the weld surface after grinding.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: (1) By using robots to replace manual labor in enclosed or semi-enclosed small space environments, operators can be effectively prevented from being exposed to high temperatures, smoke and dust and harmful gases, thus completely eliminating the safety hazards in traditional manual welding and grinding operations and ensuring construction safety from the source.

[0017] (2) The robot has a compact overall size and can move flexibly in a limited space with a width of only about 400mm, making it suitable for work areas that traditional equipment cannot enter or that are difficult for humans to access. The welding robotic arm and grinding robotic arm configured in the solution can work independently or in concert, and can handle welding and post-processing tasks for various types of welds. The system supports both autonomous control and manual intervention modes, and can quickly adjust its strategy in the event of obstructions or sudden changes in weld condition, thereby improving operational reliability and adaptability.

[0018] (3) The robot is equipped with a protective shell and necessary heat dissipation and dust removal structures, which can effectively shield the internal electronic and transmission components from the influence of dust, welding slag and high temperature environment, and extend the service life of the whole machine. The vulnerable parts such as grinding head and dust collection box adopt a modular quick disassembly and assembly structure, which can be replaced in a short time, reducing maintenance workload and later use costs. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall structure of the small-space steel structure welding and grinding integrated robot in an embodiment of the present invention; Figure 2 This is a top view of the integrated welding and grinding robot for small-space steel structures in an embodiment of the present invention after the top shell has been removed; Figure 3 This is a flowchart of the integrated welding and grinding method for small-space steel structures in an embodiment of the present invention.

[0020] In the diagram: 1. Motion control module; 2. Function control module; 3. Wireless transmission module; 4. Intelligent recognition module; 5. Attitude control module; 6. Auxiliary system; 7. Vision sensing module; 8. Protective shell; 9. Rotary wheel; 10. Six-axis high-precision structure; 11. Welding device; 12. Miniature industrial camera; 13. Grinding head. Detailed Implementation

[0021] The present invention will now be described in detail through exemplary embodiments. However, it should be understood that, without further description, elements, structures, and features in one embodiment may be advantageously incorporated into other embodiments.

[0022] Welding and grinding of steel structures in confined spaces are limited by spatial dimensions, ventilation conditions, and visibility, often resulting in problems such as equipment inaccessibility, difficulty in weld identification, and restricted working posture. This leads to unstable weld quality, high rework rates, and low overall construction efficiency. Existing methods still rely on manual observation, manual operation, and experience-based judgment, making it difficult to obtain real-time information on weld position and morphology, and even more difficult to dynamically correct the welding and grinding process, easily leading to quality risks such as welding defects and uneven processing. This invention proposes an integrated robot for welding and grinding steel structures in small spaces and its working method. By integrating visual perception, 3D scanning, path planning, and posture control, it achieves accurate acquisition of weld spatial features and fully automated operation of the welding-grinding process. It can generate operation records containing weld type, dimensional parameters, and dynamic adjustment information, significantly improving the consistency and quality stability of weld processing.

[0023] In a broad embodiment of the present invention, a method for integrated welding and grinding of steel structures in small spaces includes: The system obtains a list of weld locations on the steel structure to be worked, spatial environment information, the current position of the integrated welding and grinding robot, and remotely issued work instructions, and plans a collision-free movement trajectory accordingly. The robot is controlled to move to the first target weld position according to a collision-free movement trajectory. Two-dimensional images of the first target weld are acquired and weld edge maps are extracted. At the same time, the weld area is scanned to generate corresponding three-dimensional point cloud data. Weld type identification and weld size parameters extraction based on weld edge map and 3D point cloud data; Based on the weld type and weld size parameters, determine the optimal welding trajectory and its corresponding welding process parameters, and start the welding operation; During the welding process, the welding torch posture is continuously corrected to keep it perpendicular to the normal of the weld surface, and the welding process parameters are dynamically adjusted according to the welding status monitored in real time. The method for acquiring a two-dimensional image of the first target weld and extracting the weld edge image is as follows: The global vision camera is controlled to acquire a two-dimensional image of the first target weld and input into the intelligent recognition module 4 for feature fusion to generate a weld image with complete edges; Multi-scale feature extraction and stepwise spatial reconstruction are performed on weld seam images to obtain different types of image pixels; The image pixels are divided into weld area, base material area and background area, and the weld boundary contour is extracted to generate weld edge map.

[0024] In some embodiments, the method for scanning the weld area to generate corresponding three-dimensional point cloud data is as follows: A laser beam is projected onto the weld surface using a laser profile sensor to form laser stripes in the weld area; The laser contour sensor is controlled to scan along the extension direction of the weld seam. Its built-in CMOS camera continuously acquires multiple frames of laser stripe images during the scanning process and locates the center point of the stripes in each frame. Based on the principle of triangulation and the pre-calibrated internal and external parameters of the CMOS camera, the spatial coordinates of each stripe center point are solved to obtain the set of three-dimensional sampling point coordinates accumulated along the scanning direction on the weld surface. The coordinates of the three-dimensional sampling points are transformed, registered, and stitched together to generate complete three-dimensional point cloud data covering the weld area.

[0025] In some embodiments, the method for extracting weld size parameters is as follows: Based on the acquired 3D point cloud data of the weld, multiple cross-sectional positions are determined along the length of the weld at a preset fixed interval, and the corresponding point cloud contour is extracted at each cross-sectional position. For each point cloud contour, perform curve fitting, and calculate the geometric dimension parameters of the weld at the corresponding cross section based on the fitting results; The geometric dimensional parameters measured at each cross-sectional location are statistically processed to calculate the average, standard deviation, maximum, and minimum values ​​of each parameter, so as to comprehensively describe the dimensional characteristics of the weld.

[0026] In some embodiments, the method for generating the optimal welding trajectory is as follows: Based on the start coordinates, end coordinates, and edge diagram of the weld, several feature points characterizing the geometric features of the weld are extracted, and control points for describing the geometric shape of the weld centerline are selected from the feature points to form a control point set. A cubic Bézier curve is constructed based on the set of control points, and the spatial position and curve shape of the Bézier curve are determined by the spatial coordinates of each control point. By iteratively adjusting the spatial position of the control points, the Bezier curve is fitted to the actual centerline of the weld, thus obtaining the fitted curve of the weld centerline. The fitted curve is discretized at a preset sampling interval to generate a welding path point sequence, and the spatial coordinates of each path point are recorded. For each welding path point, the tangent direction of the fitted curve at that path point is calculated as the direction of the welding torch forward. Combined with the three-dimensional point cloud data of the weld, the normal vector of the weld surface at that welding path point is solved as a reference for the welding torch attitude, so as to constrain the direction and attitude of the trajectory point. Based on the spatial coordinates of the welding path points, the forward direction of the welding torch, and the welding torch posture reference, the corresponding joint angle parameters are solved through the inverse kinematics of the welding robot arm to map the path points to the robot arm movements. The welding path point sequence is integrated with the corresponding joint angle parameters to form a complete welding trajectory, i.e., the optimal welding trajectory.

[0027] In some embodiments, the integrated welding and grinding method for small-space steel structures further includes: After welding is completed, a dynamic grinding trajectory is generated based on the surface formation of the weld and the preset optimal grinding strategy, and the grinding robot arm is driven to perform grinding operations according to the dynamic grinding trajectory. During the grinding process, based on real-time feedback of grinding pressure and surface flatness, the feed rate of the grinding head 13 is automatically corrected and the dynamic grinding trajectory is optimized until the weld surface meets the quality requirements. After completing the current weld, the robot moves to the next weld position along a collision-free trajectory, and completes the work on all welds in sequence.

[0028] This invention also provides an integrated robot for welding and grinding steel structures in small spaces, used to realize the above-mentioned integrated welding and grinding method for steel structures in small spaces, including: Mobile work platform; The welding robotic arm and the grinding robotic arm are symmetrically arranged on both sides of the top of the mobile work platform; The visual sensing module 7, function control module 2, intelligent recognition module 4, and posture control module 5 are installed on the mobile work platform, wherein: The visual sensing module 7 is used to acquire visual information about the seam area; The intelligent recognition module 4 identifies the weld type and location based on the collected visual information and generates corresponding welding and / or grinding operation trajectories. The functional control module 2 coordinates and controls the motion execution process of the welding robot arm, the grinding robot arm and the mobile work platform based on the welding and / or grinding operation trajectory. The posture control module 5 is used to correct the working posture of the mobile work platform and the end effector of the robotic arm in real time based on the feedback results of the vision sensing module 7 and the intelligent recognition module 4.

[0029] In some embodiments, the visual sensing module 7 includes: At least one global vision camera is used to acquire two-dimensional images of the weld. In addition, at least one laser profile sensor is used to acquire three-dimensional topographic information of the weld.

[0030] In some embodiments, the end effector of the welding robotic arm is configured with: Miniature industrial camera 12 is used to acquire images of the weld seam welding process in real time; In addition, an arc length tracking sensor is used to monitor changes in the welding arc length.

[0031] In some embodiments, the end effector of the grinding robotic arm integrates: Force sensors are used to detect grinding contact force in real time; In addition, a laser displacement sensor is used to detect the flatness of the weld surface after grinding.

[0032] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0033] See Figure 1 and Figure 2 According to a first aspect of the present invention, a small-space steel structure welding and grinding integrated robot is provided. To meet the requirements of small space accessibility, its overall dimensions are designed to be approximately length × width × height = 800mm × 400mm × 500mm. It includes a mobile work platform, a welding robotic arm, and a grinding robotic arm. The welding robotic arm and the grinding robotic arm are symmetrically installed on both sides of the top of the mobile work platform and fixed to the internal steel structure of the mobile work platform by high-strength bolts. The bases of the grinding robotic arm and the welding robotic arm both have built-in rotary joints, which can realize 360° horizontal rotation, thereby expanding the working range.

[0034] The mobile work platform includes a visual sensing module 7, a function control module 2, an intelligent recognition module 4, and a posture control module 5, wherein: The vision sensing module 7 is used to acquire visual information about the weld area, including two high-definition global vision cameras and a laser contour sensor. The global vision cameras are used to capture two-dimensional images of the weld, and the laser contour sensor is used to scan the three-dimensional shape information of the weld. The acquired two-dimensional images and three-dimensional shape data are transmitted to the intelligent recognition module 4 for subsequent weld recognition and path planning.

[0035] Functional control module 2 coordinates and controls the movements of the welding robot arm, grinding robot arm, and mobile work platform based on the welding and / or grinding operation trajectories. It employs a PLC controller with sixteen digital input / output interfaces to receive manual commands or signals from the intelligent recognition module 4, and outputs control signals to the welding device 11 and grinding tool system to achieve work mode switching. Simultaneously, this module coordinates the robot arm's posture and the movement of the mobile work platform to ensure coordinated operation of all components.

[0036] The intelligent recognition module 4 identifies the weld type and location based on the collected visual information and generates corresponding welding and / or grinding operation trajectories. It incorporates a weld recognition algorithm and path planning algorithm based on a deep learning model. After receiving the two-dimensional image and three-dimensional shape data from the visual sensing module 7, the deep learning model identifies the weld type (e.g., fillet weld, butt weld) and dimensions (e.g., weld leg height, weld width), then plans a collision-free trajectory, and finally plans the optimal welding / grinding trajectory.

[0037] The attitude control module 5 is used to correct the working posture of the mobile work platform and the end effector of the robotic arm in real time based on the feedback results of the vision sensing module 7 and the intelligent recognition module 4. The attitude control module 5 includes a six-axis gyroscope and a servo drive unit. The gyroscope collects the robot's tilt angle data in real time. If the tilt exceeds the preset threshold, it outputs a signal to the height adjustment mechanism of the rotating wheel 9 (built-in electric push rod, adjustment stroke 0-50mm) to adjust it to horizontal. At the same time, this module controls the joint rotation angle of the robotic arm according to the three-dimensional contour of the weld seam to ensure that the welding / grinding tool remains perpendicular to the weld seam surface.

[0038] The welding robotic arm adopts a six-axis high-precision structure 10, with each axis using a harmonic reducer to meet the posture adjustment requirements of complex welds. A miniature industrial camera 12 is installed at the end of the welding robotic arm (near the welding torch) to collect images of the weld process in real time and transmit them to a remote control terminal. The welding device 11 adopts a gas metal arc welding (MIG) system, which can automatically match welding parameters according to the weld thickness (2-10mm). At the same time, it has a built-in arc length tracking sensor for monitoring changes in the welding arc length, which can adjust the welding torch height in real time to avoid welding defects caused by changes in weld height.

[0039] The grinding robotic arm also employs a six-axis high-precision structure 10 to meet the grinding trajectory tracking requirements under complex weld morphology. The end effector of the grinding robotic arm integrates a grinding tool system, which consists of a replaceable grinding head 13, a high-speed drive motor, a feed adjustment mechanism, and a status detection unit. The grinding head 13 adopts a modular structure, allowing for quick switching between grinding wheels (for rough grinding) and flap wheels (for fine grinding) according to process requirements. The high-speed motor provides a stable output speed for the grinding head 13, and the feed amount is precisely adjusted by a micro-feed mechanism driven by a servo motor to adapt to different weld heights and surface unevenness. A force sensor is installed between the grinding head 13 and the end effector of the robotic arm to detect the grinding contact force in real time and feed it back to the functional control module 2, achieving closed-loop adjustment of the feed amount of the grinding head 13 to avoid over- or under-grinding. Simultaneously, the grinding tool system also integrates a laser displacement sensor to detect the flatness of the weld surface after grinding, providing data support for subsequent quality assessment.

[0040] The aforementioned integrated welding and grinding robot for steel structures in small spaces utilizes a compact platform design, dual robotic arm collaborative operation, and multi-source visual perception to achieve automatic weld seam identification and full-process processing of welding and grinding. This transforms traditional operations relying on human experience into precise, data-driven control. The robot can flexibly navigate confined spaces of approximately 400 mm in width, replacing manual labor in high-temperature, dusty, and hazardous gas environments, effectively eliminating the safety risks of working in confined spaces. Deep learning-based recognition algorithms and real-time posture control mechanisms ensure high consistency in welding formation and grinding smoothness under complex weld seam conditions, significantly reducing rework and improving quality stability. Modular grinding heads, easily replaceable wear parts, and external protective structures further reduce maintenance costs and ensure long-term stable operation under complex conditions such as enclosed spaces, dust, and high temperatures.

[0041] See Figure 1 and Figure 2 According to a second aspect of the present invention, a small-space steel structure welding and grinding integrated robot is provided. The mobile work platform further includes a mobile control module 1, a wireless transmission module 3, an auxiliary system 6, a protective shell 8, and four rotating wheels 9 for movement, wherein: The core chip of the motion control module 1 incorporates a PID algorithm with a response time of ≤100ms, used to control the robot's movement trajectory. This module receives spatial coordinate data and, in conjunction with a preset route, adjusts the rotation speed of wheel 9 using the PID algorithm to compensate for movement errors and achieve precise positioning.

[0042] The wireless transmission module 3 adopts 5G / 4G dual-mode communication with a transmission rate of ≥100Mbps and a latency of ≤50m. This module is used to transmit visual data from the visual sensing module 7, pressure data from the force sensor, and real-time temperature data from the temperature sensor to the remote control terminal (computer / tablet). At the same time, it receives control commands from the remote control terminal (such as route adjustment and parameter modification), realizing two-way data interaction between the robot and the remote control terminal.

[0043] The auxiliary system 6 includes a cooling fan, a temperature sensor, and a dust removal fan. The temperature sensor is used to monitor the internal temperature of the platform. When the temperature is >60℃, the cooling fan will start automatically. During the welding and grinding process, the dust removal fan runs continuously and sucks dust and welding slag into the built-in dust collection box through the pipe. At the same time, a cooling water pipeline is set near the welding device 11 to cool the welding torch nozzle in real time.

[0044] The protective shell 8 is made of 304 stainless steel with a thickness of 3mm and is coated with a high-temperature resistant coating. The shell is a sealed structure with sealing strips in key parts (such as wiring ports and motors). Dustproof nets are only installed at the cooling fan and dust removal fan. Telescopic dustproof covers are used at the connection points between the welding robot arm, the grinding robot arm and the platform to prevent welding slag from entering the joint gaps.

[0045] The aforementioned integrated welding and grinding robot for small-space steel structures achieves stable support and efficient collaboration for welding and grinding operations within confined spaces through precise trajectory adjustment by the motion control module 1, high-speed data interaction by the wireless transmission module 3, and intelligent temperature control and dust removal by the auxiliary system 6. PID algorithm-driven motion control significantly improves positioning accuracy, enabling the robot to maintain reliable maneuverability in narrow environments; 5G / 4G communication ensures real-time transmission of visual and sensor data, enhancing remote scheduling capabilities; and a comprehensive heat dissipation, dust removal, and cooling structure prevents performance degradation due to temperature rise or dust accumulation. The sealed protective shell 8, combined with a telescopic dust cover, effectively prevents welding slag and dust from entering critical parts, improving overall machine durability. The overall structure provides the robot with high stability, high reliability, and low maintenance requirements in integrated welding and grinding operations, offering an efficient and safe automated solution for small-space steel structure processing.

[0046] See Figure 3 According to a third aspect of the present invention, a method for integrated welding and grinding of steel structures in small spaces is provided, which is applied to the integrated welding and grinding robot for steel structures in small spaces described in the second aspect of the present invention, and includes the following steps: Step 1: Obtain the list of weld locations on the steel structure to be worked, spatial environment information, the current position of the integrated welding and grinding robot, and the remotely issued work task instructions, and plan a collision-free movement trajectory accordingly.

[0047] Specifically, the method for obtaining the list of weld locations on the steel structure to be worked on is as follows: Several spatial positioning reference points with unique codes are set up in the area to be worked on. Measure the spatial distance between reference points and generate a field coordinate system that describes the actual structure on site; Spatial registration of the digital model of the steel structure is performed based on the on-site coordinate system to ensure that the coordinates of the digital model are consistent with the on-site coordinate reference. The theoretical coordinates of each weld are extracted from the registered digital model of the steel structure and converted into the corresponding field coordinates. Welds are prioritized based on their spatial location and pre-defined operational strategies, thereby generating a list of weld locations that includes their location and attribute information.

[0048] Specifically, the method for obtaining space environment information is as follows: The mobile work platform is controlled to move along a preset path within the work area. A laser contour sensor is used to continuously scan steel structure components, support frames, pipeline equipment, and other obstacles that affect the operation, thereby acquiring spatial environmental data of the work area. During the scanning process, visual texture images of the work area are simultaneously acquired by a global vision camera to supplement the surface features and spatial boundary information of obstacles; Spatial environmental data and visual texture images are spatially registered and fused to construct a three-dimensional environmental map that includes geometric structure, surface texture and spatial relationships; The location and shape of static obstacles are identified and marked in the 3D environment map. At the same time, the safety boundary of the danger zone is delineated according to the operation rules, the minimum width of the passage path is evaluated, and the motion degree of freedom constraints of the robotic arm are determined at different positions.

[0049] Step 2: Control the robot to move to the first target weld position according to the collision-free movement trajectory.

[0050] Step 3: Acquire a two-dimensional image of the first target weld and extract the weld edge map. At the same time, scan the weld area to generate corresponding three-dimensional point cloud data.

[0051] Specifically, the method for acquiring a two-dimensional image of the first target weld and extracting the weld edge image is as follows: The global vision camera is controlled to acquire a two-dimensional image of the first target weld and input into the intelligent recognition module 4 for feature fusion to generate a weld image with complete edges; Multi-scale feature extraction and stepwise spatial reconstruction are performed on weld seam images to obtain different types of image pixels; The image pixels are divided into weld area, base material area and background area, and the weld boundary contour is extracted to generate weld edge map.

[0052] Specifically, the method for scanning the weld area to generate corresponding 3D point cloud data is as follows: A laser beam is projected onto the weld surface using a laser profile sensor to form laser stripes in the weld area; The laser contour sensor is controlled to scan along the extension direction of the weld seam. Its built-in CMOS camera continuously acquires multiple frames of laser stripe images during the scanning process and locates the center point of the stripes in each frame. Based on the principle of triangulation and the pre-calibrated internal and external parameters of the CMOS camera, the spatial coordinates of each stripe center point are solved to obtain the set of three-dimensional sampling point coordinates accumulated along the scanning direction on the weld surface. The coordinates of the three-dimensional sampling points are transformed, registered, and stitched together to generate complete three-dimensional point cloud data covering the weld area.

[0053] Step 4: Identify the weld type and extract weld size parameters based on the weld edge map and 3D point cloud data.

[0054] Specifically, the method for extracting weld size parameters is as follows: Based on the acquired 3D point cloud data of the weld, multiple cross-sectional positions are determined along the length of the weld at a preset fixed interval, and the corresponding point cloud contour is extracted at each cross-sectional position. For each point cloud contour, perform curve fitting, and calculate the geometric dimension parameters of the weld at the corresponding cross section based on the fitting results; The geometric dimensional parameters measured at each cross-sectional location are statistically processed to calculate the average, standard deviation, maximum, and minimum values ​​of each parameter, so as to comprehensively describe the dimensional characteristics of the weld.

[0055] Specifically, the method for generating the optimal welding trajectory is as follows: Based on the start coordinates, end coordinates, and edge diagram of the weld, several feature points characterizing the geometric features of the weld are extracted, and control points for describing the geometric shape of the weld centerline are selected from the feature points to form a control point set. A cubic Bézier curve is constructed based on the set of control points, and the spatial position and curve shape of the Bézier curve are determined by the spatial coordinates of each control point. By iteratively adjusting the spatial position of the control points, the Bezier curve is fitted to the actual centerline of the weld, thus obtaining the fitted curve of the weld centerline. The fitted curve is discretized at a preset sampling interval to generate a welding path point sequence, and the spatial coordinates of each path point are recorded. For each welding path point, the tangent direction of the fitted curve at that path point is calculated as the direction of the welding torch forward. Combined with the three-dimensional point cloud data of the weld, the normal vector of the weld surface at that welding path point is solved as a reference for the welding torch attitude, so as to constrain the direction and attitude of the trajectory point. Based on the spatial coordinates of the welding path points, the forward direction of the welding torch, and the welding torch posture reference, the corresponding joint angle parameters are solved through the inverse kinematics of the welding robot arm to map the path points to the robot arm movements. The welding path point sequence is integrated with the corresponding joint angle parameters to form a complete welding trajectory, i.e., the optimal welding trajectory.

[0056] Step 5: Determine the optimal welding trajectory and its corresponding welding process parameters based on the weld type and weld size parameters, and start the welding operation.

[0057] Step 6: During the welding process, continuously correct the welding torch posture to keep it perpendicular to the normal of the weld surface, and dynamically adjust the welding process parameters according to the real-time monitoring of the welding status.

[0058] Step 7: After welding is completed, a dynamic grinding trajectory is generated based on the weld surface formation state and the preset optimal grinding strategy, and the grinding robot arm is driven to perform grinding operations according to the dynamic grinding trajectory.

[0059] Step 8: During the grinding process, based on the real-time feedback of grinding pressure and surface flatness, the feed amount of the grinding head 13 is automatically corrected and the dynamic grinding trajectory is optimized until the weld surface meets the quality requirements.

[0060] Specifically, the method for automatically correcting the feed rate of the grinding head 13 based on real-time feedback of grinding pressure and surface flatness is as follows: The normal pressure of the grinding head 13 in the direction perpendicular to the workpiece surface and the tangential friction force along the workpiece surface are collected in real time, and the actual contact pressure during the grinding process is obtained accordingly. Synchronously acquire the contour information of the surface of the area in front of the grinding head 13 in the direction of movement, so as to obtain the surface height change trend of the area that the grinding head 13 is about to reach; By fusing the actual contact pressure with the surface height change trend, comprehensive grinding status information for pressure stability determination is obtained. Based on the comprehensive grinding status information, pressure deviation is determined. When the actual contact pressure deviates from the preset pressure value, the grinding robot arm is controlled to make fine adjustments to the 13th position of the grinding head or the feed amount to compensate for the pressure deviation.

[0061] Specifically, the method for determining whether the weld surface meets the quality requirements is as follows: A second scan was performed on the surface of the weld after grinding to obtain the three-dimensional topographic point cloud data. The three-dimensional morphology data after grinding is compared and analyzed with the morphology data before grinding to calculate the actual amount of weld reinforcement removed, surface flatness deviation, and whether there are any depression areas caused by over-grinding. High-resolution grayscale images of the weld surface are acquired, and the orientation, uniformity, and scratch depth parameters of the surface texture are extracted by the intelligent recognition module 4. The measured geometric and surface quality parameters are automatically compared with the preset quality acceptance standards: when all test indicators are within the qualified range, the surface grinding quality of the weld is deemed qualified; if any indicator exceeds the tolerance range, a rework instruction containing the defect location and type is automatically generated, and the grinding robot arm is guided to perform targeted supplementary grinding operations.

[0062] Step 9: After completing the current weld seam operation, control the robot to move to the next weld seam position according to the collision-free movement trajectory, and complete the operation of all weld seams in sequence.

[0063] The integrated welding and grinding method for small-space steel structures described in this invention achieves a high degree of automation and precise control throughout the welding and grinding process through weld location list extraction, 3D environment modeling, collision-free path planning, weld feature recognition, optimal welding trajectory generation, and real-time posture correction. The weld recognition mechanism based on the fusion of 3D point cloud and visual features significantly improves the accuracy of weld positioning and dimension extraction; Bézier curve fitting combined with inverse kinematics solving ensures the continuity and consistency of the welding trajectory; and the real-time feedback control mechanism during the welding and grinding stages effectively suppresses process instability caused by posture deviations, pressure fluctuations, or surface height changes. Combined with the high mobility of robots in confined spaces, this method significantly reduces human intervention while ensuring weld formation quality, improving operational efficiency and stability, and providing a reliable technical path for the automated construction of complex steel structures.

[0064] The above embodiments are used to explain the present invention, but not to limit the present invention. Any modifications and changes made to the present invention within the spirit and scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A small space steel structure welding and polishing integrated working method, characterized in that, The method comprises the following steps: acquiring a welding seam position list of a steel structure to be worked, spatial environment information, a current position of a welding and grinding integrated robot, and a remote issued work task instruction, and planning a collision-free moving track based on the above information; controlling the robot to move to a first target welding seam position according to the collision-free moving track; collecting a two-dimensional image of the first target welding seam and extracting a welding seam edge map, and scanning a welding seam area to generate corresponding three-dimensional point cloud data; identifying a welding seam type and extracting welding seam size parameters based on the welding seam edge map and the three-dimensional point cloud data; determining an optimal welding track and corresponding welding process parameters according to the welding seam type and the welding seam size parameters, and starting welding work; during the welding process, continuously correcting a welding torch posture to keep it always perpendicular to a welding seam surface normal, and dynamically adjusting welding process parameters according to a real-time monitored welding state; wherein the method for collecting a two-dimensional image of the first target welding seam and extracting a welding seam edge map comprises the following steps: controlling a global vision camera to acquire a two-dimensional image of the first target welding seam and inputting the image into an intelligent recognition module for feature fusion to generate a welding seam image with complete edges; performing multi-scale feature extraction and step-by-step spatial recovery on the welding seam image to obtain different types of image pixels; dividing the image pixels into a welding seam area, a base material area and a background area, and extracting a welding seam boundary contour to generate a welding seam edge map.

2. The method according to claim 1, wherein The method for scanning the welding seam area to generate corresponding three-dimensional point cloud data comprises the following steps: projecting a laser beam onto a welding seam surface by using a laser profile sensor to form a laser stripe in the welding seam area; controlling the laser profile sensor to scan along the extension direction of the welding seam, and continuously collecting multiple frames of laser stripe images by a built-in CMOS camera during the scanning process, and positioning stripe center points in each frame of image; solving the spatial coordinates of each stripe center point based on the principle of triangulation and pre-calibrated internal and external parameters of the CMOS camera, to obtain a three-dimensional sampling point coordinate set accumulated along the scanning direction of the welding seam surface; performing coordinate conversion, registration and splicing processing on the three-dimensional sampling point coordinate set to generate complete three-dimensional point cloud data covering the welding seam area.

3. The method according to claim 1, wherein the method is characterized by, The method for extracting welding seam size parameters comprises the following steps: based on the acquired welding seam three-dimensional point cloud data, determining multiple cross-section positions along the length direction of the welding seam at a preset fixed interval, and extracting corresponding point cloud contours at each cross-section position; performing curve fitting for each point cloud contour, and calculating the geometric size parameters of the welding seam at the corresponding cross-section based on the fitting results; statistically processing the geometric size parameters measured at each cross-section position to calculate the average value, standard deviation, maximum value and minimum value of each parameter to comprehensively describe the size characteristics of the welding seam.

4. The method of claim 1, wherein the method further comprises: The method for generating an optimal welding track comprises the following steps: extracting a plurality of feature points representing the geometric characteristics of the welding seam based on the start point coordinates, end point coordinates and welding seam edge map of the welding seam, and selecting control points from the feature points to describe the geometric shape of the welding seam center line to form a control point set; constructing a cubic Bezier curve based on the control point set, and determining the spatial position and curve shape of the Bezier curve through the spatial coordinates of each control point; The spatial positions of the control points are iteratively adjusted to make the Bezier curve fit the actual center line of the weld, thereby obtaining a fitting curve of the weld center line; The fitting curve is discretized at a preset sampling interval to generate a sequence of welding path points, and the spatial coordinates of each path point are recorded; For each welding path point, the tangent direction of the fitting curve at the path point is calculated as the advancing direction of the welding torch, and the normal vector of the weld surface at the path point is solved based on the three-dimensional point cloud data of the weld, serving as a reference for the welding torch posture to constrain the direction and posture of the path point; Based on the spatial coordinates of the welding path points, the advancing direction of the welding torch, and the reference for the welding torch posture, the corresponding joint angle parameters are solved through inverse kinematics of the welding robot to map the path points to the robot motion; The sequence of welding path points and the corresponding joint angle parameters are integrated to form a complete welding trajectory, i.e., the optimal welding trajectory.

5. The method according to any one of claims 1 to 4, wherein the method is characterized in that, Further comprising: After welding is completed, a dynamic polishing trajectory is generated based on the forming state of the weld surface and a preset optimal polishing strategy, and a polishing robot is driven to perform polishing work according to the dynamic polishing trajectory; During polishing, the polishing feed rate is automatically adjusted and the dynamic polishing trajectory is optimized based on real-time feedback of the polishing pressure and surface flatness until the weld surface meets the quality requirements; After completing the current weld work, the robot is controlled to move to the next weld position along a collision-free movement trajectory to complete the work of all welds in turn.

6. A small-space steel structure welding and polishing integrated robot, used for realizing the small-space steel structure welding and polishing integrated working method according to claim 5, characterized in that, Comprising: a mobile work platform; welding robots and polishing robots symmetrically arranged on both sides of the top of the mobile work platform; a vision sensing module, a function control module, an intelligent recognition module, and a posture control module arranged on the mobile work platform, wherein: the vision sensing module is used to acquire visual information of the weld area; the intelligent recognition module identifies the type and position of the weld based on the acquired visual information and generates corresponding welding or / and polishing work trajectories; the function control module coordinates and controls the motion execution process of the welding robots, polishing robots, and mobile work platform based on the welding or / and polishing work trajectories; the posture control module corrects the work posture of the mobile work platform and the end effector of the robots in real time based on the feedback results of the vision sensing module and the intelligent recognition module.

7. The small space steel structure welding and polishing integrated robot according to claim 6, characterized in that, The vision sensing module comprises: at least one global vision camera for acquiring two-dimensional images of the weld; and at least one laser profile sensor for acquiring three-dimensional topographic information of the weld.

8. The small space steel structure welding and polishing integrated robot according to claim 7, characterized in that, The end of the welding robot is equipped with: a miniature industrial camera for real-time acquisition of weld images during welding; and an arc length tracking sensor for monitoring changes in the length of the welding arc.

9. The small space steel structure welding and polishing integrated robot according to claim 7, characterized in that, The end of the polishing robot is integrated with: a force sensor for real-time detection of polishing contact force; and a laser displacement sensor for detecting the flatness of the weld surface after polishing.

Images