Multifunctional all-terrain ship intelligent welding robot

Abstract

This invention proposes a multifunctional all-terrain intelligent welding robot for ships, comprising a quadruped chassis module, a six-degree-of-freedom robotic arm module, a vision recognition system, and a hierarchical path planning system. The quadruped chassis module employs a Mecanum wheel quadruped chassis, a liftable electromagnetic adsorption structure, and a guide rail electromagnetic foot collaborative structure to form an omnidirectional mobile platform supporting all-terrain and all-attitude operations. The six-degree-of-freedom robotic arm module is fixed above the quadruped chassis module and uses a floating coupling to isolate chassis vibration, achieving high-precision positioning of the robotic arm. The vision recognition system processes image data using a YOLOv8 model improved through Mosaic data augmentation and AFPN feature fusion, and combines OpenCV edge-morphology algorithms to optimize target recognition, developing a dedicated integrated detection algorithm for ship section assembly. The hierarchical path planning system performs three-dimensional path planning using an improved RRT*Connect algorithm, and combines force feedback data to correct the trajectory in real time, completing circumferential, blind-angle-free repair.

Description

Technical Field

[0001] This invention relates to the field of ship welding robot technology, and in particular to a multifunctional all-terrain intelligent ship welding robot. Background Technology

[0002] In shipbuilding, welding of sectional structures is a core process, involving welding operations in numerous confined spaces (such as double bottoms, bulkheads, and areas between longitudinal struts). During the hull section assembly phase, manual labor accounts for over 70% of the total construction time. Welders are exposed to harsh environments such as high temperature, high humidity, noise, and high altitudes, resulting in significant occupational health risks and a short average age of onset for occupational diseases like pneumoconiosis. Furthermore, welding quality is heavily influenced by worker experience, and defects such as incomplete penetration and undercut are prone to occur in confined areas. Currently, the application of automated welding robots in shipbuilding is still limited, mainly concentrated in steel plate cutting and welding. For example, the ship welding equipment in patent CN117506288A welds small components using a movable platform, and the automatic welding equipment for ship production in patent CN207642535U uses a motor to move a fixed column to the appropriate position and automatically welds the seams. However, these inventions are insufficient to meet the complex environmental requirements of sectional welding in shipbuilding. To achieve full automation and all-terrain adaptation, this invention uses image recognition and path planning technology to build a dedicated platform for complex environments with segmented ship hulls, which has the advantages of high work efficiency, strong flexibility, and low human intervention. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent robot that can autonomously complete multiple tasks such as welding and painting in the complex space of ship sections, so as to replace manual work in high-risk environments and ensure quality stability.

[0004] To achieve the above objectives, this invention proposes a multifunctional all-terrain intelligent welding robot for ships, comprising: The quadruped chassis module includes a chassis and four independently controllable wheel-leg mechanisms. The four wheel-leg mechanisms are evenly distributed and fixed under the chassis. Each wheel-leg mechanism is controlled by a first servo motor and a second servo motor to rotate in the xy plane and yz plane, respectively. Mecanum wheels are installed at the ends of the wheel-legs. A liftable electromagnetic chuck is provided in the center of the chassis via a bracket. Electromagnetic feet with guide rails are symmetrically arranged on both sides of the bracket. The electromagnetic chuck and electromagnetic feet work together to enable the robot to suspend from the side wall and perform ceiling operations. Six-degree-of-freedom robotic arm module: Fixed above the chassis, with a welding gun, spray gun or gripper tool detachably installed at the end. Its movement is driven by a stepper motor to achieve the first degree of freedom of rotation of the turntable bearing, and the movement of each joint is controlled by a servo motor group and belt drive mechanism. Visual recognition system: including an intelligent recognition camera fixed to the end of the robotic arm, configured to process image data using a YOLOv8 model improved by Mosaic hybrid data augmentation and AFPN feature fusion, and integrating OpenCV edge detection and morphological algorithms to optimize target recognition; A hierarchical path planning system includes: A global path planner based on the ROS framework uses the A* algorithm to optimize the path topology structure based on a pre-stored environment topology map. The local obstacle avoidance module uses the Dynamic Window Method (DWA) to construct a cost map in response to dynamic obstacles. The robotic arm path planning method generates curvature-continuous B-spline interpolation trajectories in multi-scale maps by improving the RRT*Connect algorithm.

[0005] Furthermore, the wheel-leg mechanism includes a linkage structure, consisting of a leg housing, a wheel assembly, a linkage shaft, and a bracket, which enables obstacle-crossing functionality by independently adjusting the height of the wheel legs.

[0006] Furthermore, in the wheel-leg mechanism, two connecting rod shafts are fixed to the bracket by two universal joints, and the two connecting rod shafts and the two universal joints form a parallelogram connecting rod structure.

[0007] Furthermore, the six-degree-of-freedom robotic arm module includes a first robotic arm, a second robotic arm, and a third robotic arm: The six-degree-of-freedom robotic arm module is fixed to the robotic arm base via a turntable bearing. The robotic arm base is fixed to the chassis and has a base stepper motor installed inside. The base stepper motor is connected to the turntable bearing through a drive gear, which drives the turntable bearing to rotate, giving the robotic arm its first degree of freedom. The turntable bearing is mounted on the base of the robotic arm and connected to the first robotic arm via the base of motor A. The base of motor A houses a servo motor A for controlling the first robotic arm. The servo motor A controls the first robotic arm to rotate around the Z-axis via gear transmission.

[0008] The first robotic arm and the second robotic arm are connected and a B servo motor is installed at the joint; the B servo motor uses the first robotic arm as a base to control the second robotic arm to rotate around the output axis of the B servo motor.

[0009] The end of the second robotic arm is equipped with a C motor turntable, on which a C servo motor is mounted; The third robotic arm is connected to a C servo motor, which is connected to a belt drive via a belt-driven gear.

[0010] Furthermore, the robotic arm module employs a floating coupling to isolate the chassis from vibration during movement; the joint connection ends of the robotic arm are coated with a high-temperature and arc-resistant protective layer; and the electromagnetic chuck is combined with a pressure sensor in a closed loop to control the amplitude of vibration during operation.

[0011] Furthermore, the method for path planning of the robotic arm is as follows: through multi-scale map construction and pose input, an initial path is first generated in the coarse-scale map using a pre-set tree strategy, then mapped to the fine-scale map and finely planned using an improved RRT algorithm; during the planning process, joint rotation risk is automatically detected, and if necessary, constraints are reconstructed and iteratively optimized by adding virtual spherical obstacles; for safe paths, redundant nodes are pruned, and finally, curvature continuity is achieved through curve interpolation, outputting an executable trajectory.

[0012] This invention also proposes a method for assembling ship sections, using the robot to assemble ship sections, the method being as follows: 1) Real-time acquisition of welding scene images through a visual recognition system to identify targets and terrain features; 2) Based on the optimal path framework generated by the global path planner, the Mecanum wheels are driven to achieve omnidirectional movement, and the wheel leg height is adjusted and the electromagnetic adsorption force is switched to cross obstacles. 3) The robotic arm trajectory is corrected in real time through force feedback data, and full circumferential welding / repair operations are completed within DNV classification society standard pipe fittings.

[0013] Furthermore, during the welding process, the vision recognition system monitors the weld morphology in real time and dynamically adjusts the welding parameters and the position of the robotic arm to control straightness errors.

[0014] Compared with the prior art, the advantages of the present invention are: 1. This invention adopts a four-legged chassis and electromagnetic adsorption collaborative structure design (Mecanum wheels + liftable electromagnetic chuck + guide rail electromagnetic feet) to form an omnidirectional mobile platform with all-terrain and all-attitude operation capabilities. It supports lateral and longitudinal movement, as well as turning on the spot. The working movement path is shortened, and the efficiency is greatly improved. It breaks through the limitation of traditional welding robots that are limited to ground operation, and realizes ground movement, 90° side wall suspension and ceiling inversion operation in narrow space of ship sections. It completely solves the problem of manual operation in complex areas such as double bottom and compartment.

[0015] 2. This invention significantly improves the accuracy and robustness of target recognition in harsh ship environments such as corrosion, shadows, and structural occlusion by combining an improved YOLOv8 visual model (integrating Mosaic data augmentation and AFPN feature fusion) with the OpenCV edge-morphology optimization algorithm, overcoming the failure risk of existing equipment in low-light and high-interference scenarios.

[0016] 3. This invention establishes a hierarchical collaborative path planning system, constructing a three-level planning architecture that includes global A* algorithm path planning, local DWA obstacle avoidance, and robotic arm RRT*Connect algorithm path planning. This enables the pre-stored global topology map to shorten planning time, improves the real-time obstacle avoidance response speed of dynamic obstacles, automatically optimizes the robotic arm trajectory to resolve joint rotation risks, and achieves curvature continuity through B-spline interpolation, thus solving the welding quality problem caused by abrupt trajectory changes in existing technologies.

[0017] 4. This invention enhances the suction force and isolates the transmission of vibration from chassis movement by using a dual-mode shock absorption system of mechanical leg and electromagnetic chuck combined with a floating coupling design. This reduces the amplitude of vibration during operation. Combined with real-time feedback from pressure sensors, the robotic arm maintains ultra-high repeatability positioning accuracy even in bumpy or suspended states. The straightness error of the weld seam is significantly reduced compared to manual welding, overcoming the shortcomings of traditional mobile welding platforms that cannot balance flexibility and precision.

[0018] 5. This invention improves YOLOv8 by using Mosaic hybrid data augmentation and introducing AFPN feature fusion, and trains a visual detection technology model, including dataset collection and annotation, noise and shadow addition, and using OpenCV to perform edge and morphological algorithm operations on the recognized images, and designs a dedicated integrated detection algorithm for the ship section assembly environment.

[0019] 6. This invention employs an improved RRT*Connect algorithm for three-dimensional path planning and combines force feedback data to correct the trajectory in real time, enabling full-circumferential, blind-angle-free maintenance within DNV classification society standard pipe fittings. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall structure of the robot according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the omnidirectional wheel and leg mechanism of the robot according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the overall structure of the electromagnetic chuck mechanism of the robot according to an embodiment of the present invention; Figure 4 This is a side view of the six-degree-of-freedom robotic arm module structure of the robot according to an embodiment of the present invention; Figure 5 This is a front view of the six-degree-of-freedom robotic arm module structure of the robot according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the belt drive mechanism of the third robotic arm in the six-degree-of-freedom robotic arm module of the robot according to an embodiment of the present invention; Figure 7 This is a general hardware design diagram of the visual recognition system in the robot according to an embodiment of the present invention; Figure 8This is a technical roadmap of the visual recognition system in the robot according to an embodiment of the present invention; Figure 9 This is a flowchart of the platform movement path planning process of the hierarchical path planning system in the robot according to an embodiment of the present invention.

[0021] Figures 1-6 middle, 1. Leg mechanism, 1-1-1 Chassis, 1-1-2 Motor bracket, 1-1-3 Leg housing, 1-1-4 Wheel assembly, 1-1-5 Mecanum wheel, 1-1-6 Parallel connecting rod shaft, 1-1-7 Fixed steering wheel, 1-1-8 Second servo motor, 1-1-9 Universal joint, 1-1-10 Electromagnetic chuck 2. Electromagnetic chuck mechanism, 2-2-1 bracket, 2-2-2 guide rail, 2-2-3 electromagnet feet, 2-2-4 liftable electromagnetic chuck. 3. Six-DOF robotic arm module, 3-1-1 belt, 3-1-2B servo motor, 3-1-3A servo motor, 3-1-4D servo motor gear, 3-1-5 belt driven gear, 3-1-6 third robotic arm, 3-1-7C motor turntable, 3-1-8 second robotic arm, 3-1-9 first robotic arm, 3-1-10A motor base, 3-1-11 robotic arm base, 3-1-12 turntable bearing, 3-1-13 base stepper motor, 3-1-14 turntable bearing drive gear, 3-1-15D servo motor. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be further described below.

[0023] This embodiment proposes a multifunctional all-terrain intelligent welding robot for ships, such as... Figures 1-6 As shown, the robot includes a quadruped chassis module, a six-degree-of-freedom robotic arm module 3, a visual recognition system, and a hierarchical path planning system.

[0024] In this embodiment, the quadruped chassis module includes a chassis 1-1-1, four independently controllable wheel leg mechanisms 1, and an electromagnetic chuck mechanism 2, as follows: Figure 2 As shown, the wheel-leg mechanism 1 includes a linkage structure consisting of a leg housing 1-1-3, a wheel assembly (i.e., a universal wheel bracket) 1-1-4, a parallel connecting rod shaft 1-1-6, and a chassis central bracket 2-2-1. The four evenly distributed wheel-leg mechanisms can independently adjust the height of the wheel legs, allowing each wheel leg to be raised individually to overcome obstacles and achieve the obstacle-crossing function.

[0025] like Figure 2As shown, four first servo motors are evenly installed below the chassis 1-1-1. A motor bracket 1-1-2 is installed below the first servo motor. A second servo motor 1-1-8 is installed in the motor bracket 1-1-2 through a bracket slot. The motor bracket 1-1-2 is fixedly connected to the leg housing 1-1-3 through a fixed rudder disk 1-1-7. A wheel assembly 1-1-4 is fixed to the end of the leg housing 1-1-3. A Mecanum wheel 1-1-5 is installed on the wheel assembly 1-1-4. At the same time, the bottom end of the wheel assembly 1-1-4 is connected to the parallel connecting rod shaft 1-1-6 to the central support of the chassis through a universal joint 1-1-9, so that the leg housing 1-1-3, the wheel assembly 1-1-4, the parallel connecting rod shaft 1-1-6 and the central support of the chassis form a linkage structure. In addition, the parallel connecting rod shaft 1-1-6 consists of two parallel connecting rod shafts. The two connecting rod shafts are fixed to the central support of the chassis through two universal joints 1-1-9 on the left and right sides. The two connecting rod shafts and the two universal joints 1-1-9 constitute a parallelogram connecting rod structure.

[0026] Four wheel-leg mechanisms 1 are evenly distributed and fixed below the chassis 1-1-1. Each wheel-leg mechanism is controlled by a first servo motor and a second servo motor to rotate in the xy-plane and yz-plane, respectively. Specifically, the first servo motor controls the movement of the wheel-leg in the xy-plane by controlling the motor bracket 1-1-2 of the second servo motor 1-1-8. At the same time, the motor bracket 1-1-2 of the second servo motor is connected to the leg housing 1-1-3 through the rudder disk 1-1-7, transmitting torque to the leg housing 1-1-3, thereby controlling the rotation of the wheel-leg in the yz-plane.

[0027] like Figure 3 The schematic diagram of the electromagnetic chuck mechanism 2 shown indicates that a liftable electromagnetic chuck 2-2-4 is fixed to the center of the chassis 1-1-1 via a bracket 2-2-1. Symmetrically arranged on both sides of the bracket are electromagnetic feet 2-2-3 with guide rails, which are slidably connected to the guide rails 2-2-2. The electromagnetic chuck 2-2-4 and the two electromagnetic feet 2-2-3 work together to suspend and move the robot platform, enabling sidewall suspension and ceiling operation.

[0028] In this embodiment, the six-degree-of-freedom robotic arm module is fixed above the chassis 1-1-1 via the robotic arm base 3-1-11. The robotic arm adopts a series linkage structure and has six degrees of freedom of motion. Its end is equipped with a standard tool flange, which can quickly disassemble and install various end effectors such as welding guns, spray guns, or grippers. Its motion is driven by the stepper motor 3-1-13 to drive the turntable bearing 3-1-12 to achieve the first degree of freedom of rotation, and the movement of each joint is controlled by the servo motor group (servo motor A and servo motor B) and belt drive mechanism.

[0029] like Figures 4-5As shown, the six-degree-of-freedom robotic arm module 3 includes a first robotic arm 3-1-9, a second robotic arm 3-1-8, and a third robotic arm 3-1-6. The six-degree-of-freedom robotic arm module is fixed to the robotic arm base 3-1-11 via a turntable bearing 3-1-12. The robotic arm base 3-1-11 is fixed on the chassis 1-1-1, and a base stepper motor 3-1-13 is installed inside the base 3-1-11. The base stepper motor 3-1-13 is connected to the turntable bearing 3-1-12 through the turntable bearing drive gear 3-1-14, which drives the turntable bearing 3-1-12 to rotate, giving the robotic arm its first degree of freedom.

[0030] As shown in the figure Figures 4-5 As shown, the turntable bearing 3-1-12 is mounted on the robotic arm base 3-1-11 and connected to the first robotic arm 3-1-9 via the A motor base (A motor turntable) 3-1-10. The A motor base 3-1-10 houses the A servo motor 3-1-3, which drives the first robotic arm 3-1-9 to rotate relative to the A motor base 3-1-10 around the vertical axis (Z-axis), thus controlling the robotic arm to have a second degree of freedom. The first robotic arm 3-1-9 is connected to the second robotic arm 3-1-8, and a B servo motor 3-1-2 is mounted at the joint, controlling the third degree of freedom of the robotic arm. This means that, with the first robotic arm 3-1-9 as a fixed base, the second robotic arm 3-1-8 is driven to perform pitch motion around the output axis (horizontal axis) of the B servo motor 3-1-2. This joint is crucial in determining the working radius and height of the robotic arm. A C servo motor is mounted on the end of the second robotic arm 3-1-8 via a C motor turntable 3-1-7. This C servo motor is fixed to the end of the second robotic arm 3-1-8 by the end cap, and its output shaft drives the C motor turntable 3-1-7 via a gear set, controlling the fourth degree of freedom (end-effector roll). The C motor turntable 3-1-7 is connected to the third robotic arm 3-1-6, and the C servo motor drives the third robotic arm to rotate around its own axis (along the arm direction), facilitating the adjustment of the roll posture of the end-effector tool (such as a welding torch). The second robotic arm 3-1-8 and the third robotic arm 3-1-6 are connected by a shaft, coupling, and turntable. A D servo motor 3-1-15 is also located between the second robotic arm 3-1-8 and the third robotic arm 3-1-6, such as... Figure 5 As shown, the output shaft of the D servo motor 3-1-15 transmits power to the belt driven gear 3-1-5 at the joint of the third robotic arm via the D servo motor gear 3-1-4 and the synchronous belt 3-1-1. This transmission mechanism is responsible for driving the end-effector to achieve pitch and yaw movements (i.e., the fifth and sixth degrees of freedom). For large-angle attitude adjustments, this system can kinematically coordinate with the B and C servo motors.

[0031] In addition, the joints and connecting ends of the robotic arm are coated with a high-temperature and arc-light resistant protective layer; the electromagnetic chuck and pressure sensor are combined in a closed loop to control the amplitude of vibration during operation. The outer surface of the remote camera is covered with a fixed transparent protective cover.

[0032] In this embodiment, the core drive structure and motion relationship of the robot's six-degree-of-freedom robotic arm module are as follows: First degree of freedom (base rotation): The base stepper motor 3-1-13 is installed inside the robotic arm base 3-1-11. This motor drives the gear 3-1-14 through the turntable bearing on its output shaft, which in turn drives the external turntable bearing 3-1-12 to rotate, thereby causing the entire upper structure of the robotic arm mounted on it to rotate 360° in the horizontal plane.

[0033] Second degree of freedom (shoulder rotation): A motor base 3-1-10 is connected above the turntable bearing 3-1-12, which integrates an A servo motor 3-1-3. The A servo motor 3-1-3 drives the first robotic arm 3-1-9, which is fixed to it, to rotate around the vertical axis (Z-axis) through a built-in gear reduction mechanism.

[0034] The third degree of freedom (arm pitch): A B servo motor 3-1-2 is installed at the joint connecting the first robotic arm 3-1-9 and the second robotic arm 3-1-8. The motor 3-1-2 uses the first robotic arm 3-1-9 as a base, and its output shaft directly drives the second robotic arm 3-1-8 to realize the pitch movement of the second robotic arm 3-1-8 in the vertical plane.

[0035] Fourth degree of freedom (forearm roll): The end of the second robotic arm 3-1-8 is internally encapsulated with a C servo motor (not separately labeled in the figure; its output is represented by a C motor turntable). The C servo motor drives the C motor turntable 3-1-7 to rotate, and the C motor turntable is connected to the proximal end of the third robotic arm 3-1-6, thereby controlling the third robotic arm 3-1-6 to roll around its own longitudinal axis.

[0036] Fifth and Sixth Degrees of Freedom (Wrist Posture Adjustment): To achieve more precise posture control of the end effector, this module is equipped with a remote belt drive mechanism. An independent D servo motor 3-1-15 is mounted at the connection point between the second and third robotic arms. The D servo motor gear 3-1-4 on its output shaft transmits power to the belt-driven gear 3-1-5 at the end joint of the third robotic arm via a synchronous belt 3-1-1. This drive mechanism is responsible for driving the end-effector to achieve pitch and yaw movements (i.e., the fifth and sixth degrees of freedom). For large-angle posture adjustments, this system can kinematically coordinate with the B and C servo motors.

[0037] Isolation and vibration reduction control: The floating coupling installed in the connecting shaft system of the second robotic arm 3-1-8 and the third robotic arm 3-1-6 has the following vibration reduction principle: the flexible element inside the coupling can effectively absorb and isolate the low-frequency vibration and impact generated by the movement or obstacle crossing of the four-legged chassis, and prevent it from being transmitted to the end effector, thereby achieving the vibration reduction effect.

[0038] Adaptive closed-loop control of suction force: High-precision pressure sensors are integrated into the telescopic rod of the electromagnetic chuck 2-2-4 and the drive shaft of the wheel mechanism. These sensors monitor the pressure changes and vibration amplitude between the robot body and the working surface in real time and feed the signals back to the control system. Based on this, the system dynamically adjusts the suction force of the electromagnetic chuck and the servo stiffness of the telescopic rod, while fine-tuning the posture of the wheel mechanism during movement, forming an active vibration reduction closed loop. This ensures that the robotic arm maintains extremely high operational stability and repeatability even when suspended from a side wall or working on a ceiling.

[0039] In this embodiment, the visual recognition system includes an intelligent recognition camera fixed to the end effector of a six-degree-of-freedom robotic arm. This camera can upload real-time collected images to the cloud for processing. The visual recognition system is configured to process image data using a YOLOv8 model improved through Mosaic hybrid data augmentation and AFPN feature fusion, and integrates OpenCV edge detection and morphological algorithms to optimize target recognition. The calculation formula is: Final detection result = Ω(YOLOv8_AFPN(Φ_Mosaic(I)),E_Morph(I)) I represents the raw image captured and uploaded by the camera.

[0040] Φ_Mosaic(I) is used during the model training phase to process the training images using Mosaic data augmentation to obtain model weights with stronger generalization ability.

[0041] YOLOv8_AFPN This is the core detection engine of the system. It is a YOLOv8 model improved with an AFPN feature fusion structure, responsible for deep feature extraction and multi-scale object detection of the input image, and outputting preliminary detection boxes and categories.

[0042] E_Morph(I) represents the OpenCV edge detection and morphological optimization module. It performs edge refinement, noise removal, and region optimization directly on the original image or in combination with the model output.

[0043] Ω This represents the fusion and optimization function. It intelligently fuses and post-processes the preliminary results of the deep learning model with the results of traditional image processing optimization, ultimately outputting accurate and robust target recognition results. like Figure 8 As shown, the route for image recognition using a visual recognition system is as follows: 1) Multimodal data acquisition, i.e., real-time acquisition of welding scene images; 2) Data preprocessing, i.e., preprocessing the acquired image data; 3) Feature extraction and fusion, i.e., feature extraction from the preprocessed image data; 4) Intelligent recognition and localization, i.e., accurately recognizing welding targets and complex terrain features, while using OpenCV for edge detection and morphological algorithm optimization to improve recognition robustness; 5) Result output and control.

[0044] In use, the robot of this embodiment is placed near the processing area of ​​the ship section assembly, while maintaining a certain distance to avoid affecting the movement of the remote camera. The remote camera is started by connecting to an external power source through a wire. The capture program is set, and the external rotor motor is connected to the power source through a wire and started. The external rotor motor moves through the motor frame, thereby driving the remote camera to cruise and capture images. The captured image information is transmitted to an external data receiving device (i.e., a computer) via a wireless network.

[0045] In this embodiment, the hierarchical path planning system includes: 1) a global path planner based on the ROS framework, which uses the A* algorithm to optimize the path topology structure according to the pre-stored environment topology map; 2) a local obstacle avoidance module that uses the Dynamic Window Method (DWA) to construct a cost map to respond to dynamic obstacles; and 3) a robotic arm path planning module that generates curvature-continuous B-spline interpolation trajectories in a multi-scale map by improving the RRT*Connect algorithm.

[0046] Specifically, at the motion control level of the robot platform, a heterogeneous computing platform based on the ROS framework is used. The global path planner integrating the A* algorithm generates the optimal path framework based on the pre-stored environmental topology map. The DWA local obstacle avoidance module responds to obstacles in real time through a dynamic cost map, and collaboratively drives the Mecanum wheels of the quadruped chassis to achieve omnidirectional movement. It also crosses obstacles through the wheel leg height adjustment mechanism. With the switching of the adsorption force of the liftable electromagnetic chuck and the guide rail electromagnetic feet, it can achieve stable operation in multiple postures on vertical walls, ceilings, etc.

[0047] In terms of motion planning for six-axis robotic arms, such as Figure 9As shown, a multi-scale environment map is first constructed. The robot's initial pose, target point pose, and obstacle information are input. In the coarse-scale space, a pre-defined tree strategy is used to quickly explore the initial path. After mapping to the fine-scale map, an improved RRT algorithm is used for refined path search. During the planning process, the kinematic constraints of the robot's joints are continuously monitored to determine if joint rotational problems occur. If so, i.e., when a potential rotational problem risk is detected, virtual spherical obstacles are automatically generated in Cartesian space to reconstruct the path constraints. (The core of this "virtual spherical obstacle" generation rule is an active safety strategy that transforms kinematic risks into geometric constraints.) When a joint approaches its limit or the robot approaches a singularity, the path is reconstructed. During configuration, the system dynamically generates a spherical obstacle with an adaptive radius that increases with the risk level, centered on the current risk point. This obstacle is then inserted into the environment map, forcing the path planning algorithm to actively avoid this "forbidden zone" that could lead to motion failure. Ultimately, a smooth and feasible trajectory is replanned while ensuring kinematic safety. Path safety is ensured through iterative optimization. If not, a cyclic pruning strategy is implemented to prune redundant nodes on conflict-free paths to improve efficiency. Finally, path curvature continuity is achieved through cubic B-spline curve interpolation (NURBS curve interpolation), generating an executable trajectory that meets the dynamic requirements and outputting the final path.

[0048] When the robot in this embodiment is working, the electromagnetic chuck 2-2-4 provides adsorption force according to the operation requirements. The 6-DOF robotic arm drives the turntable bearing 3-1-12 through the stepper motor 3-1-13 to achieve the first degree of freedom rotation. The A servo motor 3-1-3, the B servo motor 3-1-2 and the belt drive device work together to control the movement of each joint, driving the end welding gun, spray gun or gripper and other tools to work precisely along the planned trajectory. The whole process is monitored in a closed loop by the vision system, which adjusts the welding parameters or the position of the robotic arm in real time to ensure high precision and high adaptability of ship section assembly operations.

[0049] The method for performing ship section assembly operations using a robot in this embodiment is as follows: 1) Real-time acquisition of welding scene images through a visual recognition system to identify targets and terrain features; 2) Based on the optimal path framework generated by the global path planner, the Mecanum wheels are driven to achieve omnidirectional movement, and the wheel leg height is adjusted and the electromagnetic adsorption force is switched to cross obstacles. 3) The robotic arm trajectory is corrected in real time through force feedback data, and full circumferential welding / repair operations are completed within DNV classification society standard pipe fittings.

[0050] During the welding operation, the vision recognition system monitors the weld morphology in real time and dynamically adjusts the welding parameters and the position of the robotic arm to control straightness error.

[0051] The above are merely preferred embodiments of the present invention and do not constitute any limitation on the present invention. Any equivalent substitutions or modifications made by those skilled in the art to the technical solutions and content disclosed in the present invention without departing from the scope of the present invention shall be deemed to have remained within the protection scope of the present invention.

Claims

1. A multifunctional all-terrain intelligent welding robot for ships, characterized in that, include: The quadruped chassis module includes a chassis and four independently controllable wheel-leg mechanisms. The four wheel-leg mechanisms are evenly distributed and fixed under the chassis. Each wheel-leg mechanism is controlled by a first servo motor and a second servo motor to rotate in the xy plane and yz plane, respectively. Mecanum wheels are installed at the ends of the wheel-legs. A liftable electromagnetic chuck is provided in the center of the chassis via a bracket. Electromagnetic feet with guide rails are symmetrically arranged on both sides of the bracket. The electromagnetic chuck and electromagnetic feet work together to enable the robot to suspend from the side wall and perform ceiling operations. Six-degree-of-freedom robotic arm module: Fixed above the chassis, with a welding gun, spray gun or gripper tool detachably installed at the end. Its movement is driven by a stepper motor to achieve the first degree of freedom of rotation of the turntable bearing, and the movement of each joint is controlled by a servo motor group and belt drive mechanism. Visual recognition system: including an intelligent recognition camera fixed to the end of the robotic arm, configured to process image data using a YOLOv8 model improved by Mosaic hybrid data augmentation and AFPN feature fusion, and integrating OpenCV edge detection and morphological algorithms to optimize target recognition; A hierarchical path planning system includes: A global path planner based on the ROS framework uses the A* algorithm to optimize the path topology structure based on a pre-stored environment topology map. The local obstacle avoidance module uses the Dynamic Window Method (DWA) to construct a cost map in response to dynamic obstacles. The robotic arm path planning method generates curvature-continuous B-spline interpolation trajectories in multi-scale maps by improving the RRT*Connect algorithm.

2. The multifunctional all-terrain intelligent welding robot for ships according to claim 1, characterized in that, The wheel-leg mechanism includes a linkage structure, consisting of a leg housing, a wheel assembly, a linkage shaft, and a bracket, and achieves obstacle-crossing function by independently adjusting the height of the wheel legs.

3. The multifunctional all-terrain intelligent welding robot for ships according to claim 2, characterized in that, In the wheel-leg mechanism, two connecting rod shafts are fixed to the bracket by two universal joints, and the two connecting rod shafts and the two universal joints form a parallelogram connecting rod structure.

4. The multifunctional all-terrain intelligent welding robot for ships according to claim 1, characterized in that, The six-degree-of-freedom robotic arm module includes a first robotic arm, a second robotic arm, and a third robotic arm: The six-degree-of-freedom robotic arm module is fixed to the robotic arm base via a turntable bearing. The robotic arm base is fixed to the chassis and has a base stepper motor installed inside. The base stepper motor is connected to the turntable bearing through a drive gear, which drives the turntable bearing to rotate, giving the robotic arm its first degree of freedom. The turntable bearing is mounted on the base of the robotic arm and connected to the first robotic arm via the base of motor A. The base of motor A contains a servo motor A for controlling the first robotic arm. The first robotic arm and the second robotic arm are connected and a B servo motor is installed at the connecting joint; The end of the second robotic arm is equipped with a C motor turntable, on which a C servo motor is mounted; The third robotic arm is connected to a C servo motor, which is connected to a belt drive via a belt-driven gear.

5. The multifunctional all-terrain intelligent welding robot for ships according to claim 1, characterized in that, The robotic arm module uses a floating coupling to isolate the chassis from vibration; the joint connection ends of the robotic arm are coated with a high-temperature and arc-resistant protective layer; the electromagnetic chuck and pressure sensor are combined in a closed loop to control the amplitude of vibration during operation.

6. The multifunctional all-terrain intelligent welding robot for ships according to claim 1, characterized in that, The method for path planning of the robotic arm is as follows: through multi-scale map construction and pose input, an initial path is first generated in the coarse-scale map using a pre-set tree strategy, and then mapped to the fine-scale map and fine-planned using an improved RRT algorithm; during the planning process, the risk of joint rotation is automatically detected, and if necessary, constraints are reconstructed and iteratively optimized by adding virtual spherical obstacles; for safe paths, redundant nodes are pruned, and finally, curvature continuity is achieved through curve interpolation, and an executable trajectory is output.

7. A method for assembling ship sections, using the multifunctional all-terrain intelligent welding robot described in any one of claims 1-6, characterized in that the steps include... include: 1) Real-time acquisition of welding scene images through a visual recognition system to identify targets and terrain features; 2) Based on the optimal path framework generated by the global path planner, the Mecanum wheels are driven to achieve omnidirectional movement, and the wheel leg height is adjusted and the electromagnetic adsorption force is switched to cross obstacles. 3) The robotic arm trajectory is corrected in real time through force feedback data, and full circumferential welding / repair operations are completed within DNV classification society standard pipe fittings.

8. The method for assembling ship sections according to claim 7, characterized in that, During the welding operation, the vision recognition system monitors the weld morphology in real time and dynamically adjusts the welding parameters and the position of the robotic arm.

Images