Tunnel lining detection vehicle real-time hierarchical obstacle avoidance control method and device

Abstract

The application discloses a tunnel lining detection vehicle real-time hierarchical obstacle avoidance control method and device, and the method comprises the following steps: based on tunnel environment data and parameter data of a tunnel lining detection vehicle, a multi-body dynamics digital twin model of the tunnel lining detection vehicle is constructed, the multi-body dynamics digital twin model comprises a virtual tunnel environment model and a dynamics model of the tunnel lining detection vehicle; environment data and detection vehicle state data collected by detection sensors carried by the tunnel lining detection vehicle in real time are obtained; based on the environment data and the detection vehicle state data, the multi-body dynamics digital twin model is driven to perform synchronous simulation, and simulation detection information is generated; according to obstacle information, a corresponding hierarchical obstacle avoidance strategy is executed in the multi-body dynamics digital twin model simulation, control instructions are generated, the control instructions comprise a control target and a control action; and the control instructions are issued according to the control target. The application can automatically avoid obstacles according to different obstacles, reduce virtual control errors and has good real-time performance.

Description

Technical Field

[0001] This invention relates to the field of intelligent detection and virtual control technology for railway tunnels, and in particular to a real-time graded obstacle avoidance control method and device for a tunnel lining inspection vehicle. Background Technology

[0002] This section is intended to provide background or context for the embodiments of the invention set forth in the claims. The description herein is not an admission that it is prior art simply because it is included in this section.

[0003] A tunnel inspection vehicle is a specialized engineering vehicle designed for rapid, comprehensive, and automated inspection of tunnel structures. It integrates various advanced sensing technologies, data acquisition systems, positioning technologies, and analysis software, enabling it to efficiently acquire detailed status information about the tunnel's appearance and structure. It is an important tool for tunnel operation and maintenance management, safety assessment, and preventative maintenance. In actual operation, it primarily uses robotic arms of different structural types equipped with ground-penetrating radar antennas to approach the tunnel lining surface and detect defects such as voids and looseness within the lining, assessing whether there are any safety hazards in the internal condition of the tunnel lining.

[0004] However, in actual operation, due to the presence of contact wire frames, various cables, bolts, and rivets inside the tunnel, and the potential for variations in the diameter of the lining surface in specific road sections, these structural changes within the tunnel can affect the normal operation of the ground-penetrating radar antenna. Even with obstacle detection sensors, the system will only stop and avoid detected obstacles, resulting in a detection interruption rate exceeding 60%, significantly impacting the efficiency and cost of tunnel inspection operations. Therefore, in order to develop a control algorithm with a low interruption rate, it is of great significance to develop a virtual simulation control system in advance, capable of continuously debugging and optimizing the vehicle control algorithm.

[0005] The robotic arm movements of many existing virtual inspection platforms are based on ideal kinematic models, resulting in large errors between virtual control and physical vehicle control. Furthermore, the system's obstacle avoidance strategy is simplistic and cannot distinguish between different obstacles such as protruding spikes and overhead contact lines. In addition, there is a high response delay, which cannot meet the real-time control requirements of tunnels with variable cross-sections.

[0006] Furthermore, existing technologies have already developed technical solutions to improve the accuracy of individual robotic arm motion trajectory control, but their core lies in solving the execution-level problem of "how to make the robotic arm move more precisely." When facing dense and diverse obstacles within tunnels, existing technologies still lack a system-level decision-making method to direct multiple robotic arms to coordinate obstacle avoidance and maintain the continuity of inspection operations to the greatest extent possible. Therefore, the industry-wide problem of high interruption rates in inspection operations due to obstacle avoidance remains unresolved. Summary of the Invention

[0007] This invention provides a real-time graded obstacle avoidance control method for a tunnel lining inspection vehicle, which can automatically avoid different obstacles, reduce virtual control errors, and has good real-time performance. The method includes:

[0008] Based on tunnel environment data and parameter data of the tunnel lining inspection vehicle, a multibody dynamics digital twin model of the tunnel lining inspection vehicle is constructed. The multibody dynamics digital twin model includes a virtual tunnel environment model and a dynamics model of the tunnel lining inspection vehicle. The virtual tunnel environment includes various obstacle types, and the dynamics model of the tunnel lining inspection vehicle includes the dynamics models of each robotic arm.

[0009] The system obtains real-time environmental data and vehicle status data collected by the detection sensors mounted on the tunnel lining inspection vehicle.

[0010] Based on environmental data and vehicle status data, the multibody dynamics digital twin model is driven to perform synchronous simulation to generate simulation detection information.

[0011] Based on the simulation detection information, the corresponding hierarchical obstacle avoidance strategy for the multi-manipulator system is executed in the multibody dynamics digital twin model simulation, and control commands are generated. The control commands include control targets and control actions. The hierarchical obstacle avoidance strategy includes non-stop obstacle avoidance strategy and stop obstacle avoidance strategy.

[0012] The control commands are issued according to the control objectives.

[0013] This invention provides a real-time graded obstacle avoidance control device for a tunnel lining inspection vehicle, which can automatically avoid different obstacles, reduce virtual control errors, and has good real-time performance. The device includes:

[0014] The model building module is used to construct a multibody dynamics digital twin model of the tunnel lining inspection vehicle based on tunnel environment data and parameter data of the tunnel lining inspection vehicle. The multibody dynamics digital twin model includes a virtual tunnel environment model and a dynamic model of the tunnel lining inspection vehicle. The virtual tunnel environment includes various obstacle types, and the dynamic model of the tunnel lining inspection vehicle includes the dynamic models of each robotic arm.

[0015] The data acquisition module is used to obtain real-time environmental data and vehicle status data collected by the detection sensors mounted on the tunnel lining inspection vehicle;

[0016] The simulation module is used to drive the multibody dynamics digital twin model to perform synchronous simulation based on environmental data and vehicle status data, and generate simulation detection information.

[0017] The control command generation module is used to generate control commands based on the simulation detection information, and to simulate and execute the corresponding hierarchical obstacle avoidance strategy for the multi-manipulator system in the multibody dynamics digital twin model. The control commands include control targets and control actions, and the hierarchical obstacle avoidance strategy includes non-stop obstacle avoidance strategy and stop obstacle avoidance strategy.

[0018] The instruction issuing module is used to issue the control instructions according to the control objectives.

[0019] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described real-time graded obstacle avoidance control method for tunnel lining inspection vehicles.

[0020] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described real-time graded obstacle avoidance control method for tunnel lining inspection vehicles.

[0021] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described real-time graded obstacle avoidance control method for tunnel lining inspection vehicles.

[0022] The technical solution provided by the embodiments of the present invention has achieved the following significant beneficial effects compared with the prior art:

[0023] 1. A technological leap from "precise control" to "intelligent decision-making" has been achieved: For the first time, the solution has shifted from improving the trajectory accuracy of a single robotic arm to building a multi-robotic arm collaborative hierarchical decision-making system based on virtual simulation verification. By matching hierarchical obstacle avoidance strategies with different levels of interruption according to obstacle type and prioritizing non-stop solutions, the fundamental problem of high detection interruption rate has been systematically solved.

[0024] 2. Significantly improved inspection efficiency: By implementing a graded obstacle avoidance strategy (e.g., to deal with protruding nails, contact wires and tunnels with variable cross sections), the inspection interruption rate under such common working conditions can be reduced from more than 60% in traditional methods to less than 15%, achieving an order-of-magnitude improvement in inspection efficiency.

[0025] 3. A safe closed-loop control mechanism was constructed: By simulating and pre-verifying obstacle avoidance actions in a multibody dynamics digital twin model, the feasibility and safety of control commands were ensured, avoiding the risk of collisions between physical equipment due to decision-making errors from the source, and forming a reliable closed loop of "decision-simulation-verification-execution".

[0026] 4. Ensured control accuracy and reliability: Based on a high-precision multibody dynamics digital twin model and real-time acquired environmental data and vehicle status data, the consistency between virtual simulation results and physical execution effects was ensured, providing a reliable basis for the aforementioned intelligent decision-making. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0028] Figure 1 This is a flowchart of the real-time graded obstacle avoidance control method for tunnel lining inspection vehicles in an embodiment of the present invention;

[0029] Figure 2 This is a three-dimensional schematic diagram of the virtual tunnel environment model in an embodiment of the present invention;

[0030] Figure 3 This is a schematic diagram of the dynamic model of the vehicle body in an embodiment of the present invention;

[0031] Figure 4 This is a schematic diagram of the dynamic model of the lifting platform in an embodiment of the present invention;

[0032] Figure 5 This is a schematic diagram of the dynamic model of the first and second robotic arms in an embodiment of the present invention;

[0033] Figure 6 This is a schematic diagram of the dynamic model of the third robotic arm in an embodiment of the present invention;

[0034] Figure 7 This is a schematic diagram of the dynamic models of the fourth and fifth robotic arms in an embodiment of the present invention;

[0035] Figure 8 This is a schematic diagram showing the protruding nail found in an embodiment of the present invention;

[0036] Figure 9 A schematic diagram of the contact wire frame found in the embodiments of the present invention;

[0037] Figure 10 A schematic diagram of the variable cross-section found in the embodiments of the present invention;

[0038] Figure 11 This is a schematic diagram of the real-time graded obstacle avoidance control device for the tunnel lining inspection vehicle in an embodiment of the present invention;

[0039] Figure 12This is a schematic diagram of a computer device in an embodiment of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0041] Existing technologies, in order to improve the control accuracy of the individual motion trajectory of robotic arms, have proposed methods to enable more precise movement of the robotic arms. However, these technologies still have limitations when facing dense and diverse obstacles in tunnels: focusing only on how to make the robotic arms move more precisely does not solve the system-level decision-making problem of "under what circumstances, which or which robotic arms should be commanded, and in what manner, to maximize the continuity of the inspection operation while ensuring safety." Therefore, the industry problem of high interruption rate of inspection operations due to obstacle avoidance still exists. To address this, this invention proposes a real-time graded obstacle avoidance control method for tunnel lining inspection vehicles, such as... Figure 1 This is a flowchart of a real-time graded obstacle avoidance control method for a tunnel lining inspection vehicle in an embodiment of the invention. The method includes:

[0042] Step 101: Based on tunnel environment data and parameter data of the tunnel lining inspection vehicle, construct a multibody dynamics digital twin model of the tunnel lining inspection vehicle. The multibody dynamics digital twin model includes a virtual tunnel environment model and a dynamics model of the tunnel lining inspection vehicle. The virtual tunnel environment includes various obstacle types, and the dynamics model of the tunnel lining inspection vehicle includes the dynamics models of each robotic arm.

[0043] Step 102: Obtain real-time environmental data and vehicle status data collected by the detection sensors mounted on the tunnel lining inspection vehicle;

[0044] Step 103: Based on environmental data and vehicle status data, drive the multibody dynamics digital twin model to perform synchronous simulation and generate simulation detection information;

[0045] Step 104: Based on the simulation detection information, execute the corresponding hierarchical obstacle avoidance strategy for the multi-manipulator system in the multibody dynamics digital twin model simulation, and generate control commands. The control commands include control targets and control actions. The hierarchical obstacle avoidance strategy includes non-stop obstacle avoidance strategy and stop obstacle avoidance strategy.

[0046] Step 105: Issue the control command according to the control objective.

[0047] In this embodiment of the invention, a multi-body dynamics digital twin model constructed based on tunnel environment data and inspection vehicle parameter data includes a virtual tunnel environment model and an inspection vehicle dynamics model, wherein the virtual tunnel environment covers various obstacle types. Compared with traditional simplified models that can only simulate a single fixed scene, this technology can accurately reproduce the actual obstacle combination of railway tunnels, solving the problem of the disconnect between virtual scenes and real environments, and providing a reliable model foundation for subsequent simulation and obstacle avoidance control. The inspection vehicle dynamics model can fully present the motion characteristics of each component of the inspection vehicle, improve the simulation accuracy of the deformation of the robotic arm end effector, and avoid data distortion or obstacle avoidance deviation caused by model errors. Environmental data and inspection vehicle status data are collected in real time through the detection sensors mounted on the inspection vehicle. Compared with traditional single sensors that can only collect local data (such as only obtaining obstacle positions), this technology collects more comprehensive and accurate data, with a high accuracy rate in obstacle type identification. Based on the real-time collected environmental data and inspection vehicle status data, the multi-body dynamics digital twin model is driven to perform synchronous simulation and generate simulation detection information. Compared to the limitations of traditional offline simulations that cannot match real-time environmental changes, this system achieves a linkage mechanism between real-time data input and dynamic simulation output. Simulation results can reflect the dynamic changes of the tunnel environment and the inspection vehicle in real time, with a small deviation rate between simulation and real data. Based on the obstacle type and status in the simulation detection information, corresponding hierarchical obstacle avoidance strategies are executed. Compared to the traditional one-size-fits-all obstacle avoidance method (such as stopping to avoid obstacles regardless of type), this system can dynamically adjust obstacle avoidance actions according to the obstacle threat level, reducing unnecessary work interruptions and improving detection efficiency. Control commands are categorized according to control objectives and sent to the physical inspection vehicle, ensuring that different components can accurately receive the corresponding control actions, reducing command transmission confusion and delays, and ensuring high accuracy in control command execution, avoiding equipment malfunctions caused by incorrect command transmission. Through model construction, data perception, simulation-driven, obstacle avoidance control, and command execution, the system improves the effectiveness in three core dimensions: accuracy, efficiency, and safety.

[0048] In step 101, based on tunnel environment data and parameter data of the tunnel lining inspection vehicle, a multibody dynamics digital twin model of the tunnel lining inspection vehicle is constructed. The multibody dynamics digital twin model includes a virtual tunnel environment model and a dynamics model of the tunnel lining inspection vehicle. The virtual tunnel environment includes various obstacle types, and the dynamics model of the tunnel lining inspection vehicle includes the dynamics models of each robotic arm.

[0049] In this embodiment of the invention, the tunnel inspection vehicle typically operates inside a tunnel. The tunnel scenario involved in this invention is a railway tunnel, whose interior walls and arches are covered with an inner wall covering layer, referred to as the lining surface.

[0050] In one embodiment, the obstacle type includes at least one of spikes, contact wire frames, and tunnels with variable cross-sections;

[0051] If the obstacle type is a protruding nail and a contact wire frame, the graded obstacle avoidance strategy is a non-stop obstacle avoidance strategy.

[0052] If the obstacle type is a tunnel with variable cross-section, the graded obstacle avoidance strategy is a parking obstacle avoidance strategy.

[0053] Figure 2 This is a three-dimensional schematic diagram of the virtual tunnel environment model in an embodiment of the present invention. The virtual tunnel environment model is modeled at a 1:1 scale according to the actual size of the tunnel environment data, and obstacles are generated based on the tunnel environment data.

[0054] In this embodiment of the invention, the point cloud density of the tunnel three-dimensional point cloud data can be enhanced by using tunnel environment data collected by laser scanning and area array camera, and hidden obstacles can be identified; a dynamic virtual tunnel environment model can be generated based on the hidden obstacles and the current obstacle type.

[0055] In this embodiment of the invention, the hidden obstacles include lining cracks, water leakage, etc., and combined with visible obstacles such as contact wires and variable cross sections, a dynamic environment model is generated, thereby breaking through the limitations of fixed scenes and simulating more than 95% of the actual obstacle combinations in railway tunnels, thus improving the coverage of simulation scenes.

[0056] In one embodiment, the dynamic model of the tunnel lining inspection vehicle also includes a vehicle body dynamic model and a lifting platform dynamic model.

[0057] The tunnel lining inspection vehicle was modeled and textured at a 1:1 scale based on photographs of actual equipment and its parameter data. It mainly includes the power locomotive, transport carriage, robotic arm, and obstacle detection sensors. The dynamic model of the tunnel lining inspection vehicle refers to the wheel-rail contact relationship and the motion connection relationships between bogie components. This modeling process is complex and consumes significant computational resources. This invention primarily focuses on the robotic arm's motion, and since the tunnel inspection vehicle travels at a very low speed, the dynamic model of the vehicle body has been appropriately simplified to highlight the object of focus. Figure 3 This is a schematic diagram of the dynamic model of the vehicle body in an embodiment of the present invention, including the overall shape of the vehicle, the wheel-rail contact relationship between the wheels and the track, etc.

[0058] The lifting platform consists of two scissor lift mechanisms connecting the vehicle body and the robotic arm. The scissor mechanism has many moving parts, and each link needs to create an independent motion unit. The motion units also need to be connected. There are over-constraints between multiple sets of connections, and some constraints need to be appropriately relaxed. Figure 4 This is a schematic diagram of the dynamic model of the lifting platform in an embodiment of the present invention. Figure 4 It includes a first lifting platform and a second lifting platform. The dynamic model of each lifting platform includes 6 connecting rods.

[0059] In this embodiment of the invention, the first and second robotic arms have identical structures, are centrally symmetrically arranged, and contain 18 kinematic pairs. They are the main robotic arms for tunnel inspection vehicles. Figure 5 This is a schematic diagram of the dynamic model of the first and second robotic arms in an embodiment of the present invention. The components included in the dynamic model are a connecting base, a first-stage telescopic arm, a second-stage telescopic arm, a third-stage telescopic arm, a fourth-stage telescopic arm, a lower bending arm, an upper bending arm, a first hydraulic cylinder outer cylinder, a first hydraulic cylinder inner cylinder, a second hydraulic cylinder inner cylinder, a second hydraulic cylinder outer cylinder, a third hydraulic cylinder inner cylinder, a third hydraulic cylinder outer cylinder, a first detection radar base, a first detection radar, a second detection radar base, a second detection radar, and a third detection radar.

[0060] The hydraulic cylinder sliding pair of the third robotic arm has a precision of 0.1mm. Figure 6 This is a schematic diagram of the dynamic model of the third robotic arm in an embodiment of the present invention. The components include a connecting base, a first-stage telescopic arm, a second-stage telescopic arm, a third-stage telescopic arm, a fourth-stage telescopic arm, a fifth-stage telescopic arm, an outer cylinder of a first hydraulic cylinder, an inner cylinder of a first hydraulic cylinder, an inner cylinder of a second hydraulic cylinder, an outer cylinder of a second hydraulic cylinder, and an inner cylinder of a third hydraulic cylinder.

[0061] The fourth and fifth robotic arms have similar structures to the lifting platform, also using scissor mechanisms, but they have more movable parts, more constraints, and a more complex modeling process. Figure 7 This is a schematic diagram of the dynamic model of the fourth and fifth robotic arms in an embodiment of the present invention. The components include a base, an inner cylinder of the first hydraulic cylinder, an outer cylinder of the first hydraulic cylinder, a dust cover, an eighth detection radar, a first scissor lift link, a second scissor lift link, a third scissor lift link, a fourth scissor lift link, a fifth scissor lift link, a sixth scissor lift link, a seventh scissor lift link, an eighth scissor lift link, a ninth scissor lift link, a tenth scissor lift link, a first scissor lift front link, and a first scissor lift rear link. Second scissor lift front link, second scissor lift rear link, third scissor lift front link, third scissor lift rear link, fourth scissor lift front link, fourth scissor lift rear link, fifth scissor lift front link, fifth scissor lift rear link, sixth scissor lift front link, sixth scissor lift rear link, seventh scissor lift front link, seventh scissor lift rear link, eighth scissor lift front link, eighth scissor lift rear link, ninth scissor lift front link, ninth scissor lift rear link, tenth scissor lift front link, tenth scissor lift rear link, scissor lift link pin.

[0062] In one embodiment, based on tunnel environment data and parameter data of the tunnel lining inspection vehicle, a multibody dynamics digital twin model of the tunnel lining inspection vehicle is constructed, including:

[0063] The load-bearing components of the tunnel lining inspection vehicle are modeled as rigid bodies.

[0064] Based on the parameter data of the tunnel lining inspection vehicle, a modal neutral file is generated according to the easily deformable parts of the tunnel lining inspection vehicle, and a flexible body model is constructed based on the modal neutral file;

[0065] Based on the rigid body model and the flexible body model, the vehicle body dynamics model, the lifting platform dynamics model, and the robotic arm dynamics model of each robotic arm are established respectively. Among them, the vehicle body dynamics model is a coupling model of the rigid vehicle body and the flexible wheel axle, the lifting platform dynamics model is a coupling model of the rigid fixed frame and the flexible scissor lift linkage, and the robotic arm dynamics model is a coupling model of the rigid base and the flexible robotic arm.

[0066] In this embodiment of the invention, to meet the high-precision operation requirements of the robotic arm, a rigid-flexible body partitioning modeling method is adopted to model the load-bearing components of the tunnel lining inspection vehicle (vehicle base, lifting platform fixing frame, each robotic arm base, and cylinder outer cylinder) as rigid bodies. Vulnerable components include the four-stage telescopic arm, lower curved arm, and upper curved arm of the first and second robotic arms (which are prone to elastic deformation due to radar loads), the five-stage telescopic arm of the third robotic arm (which needs to meet a sliding pair accuracy of 0.1mm, and the deformation amount needs to be precisely controlled), and the scissor links of the fourth and fifth robotic arms (10 sets of scissor links, which are prone to stress deformation due to over-constraint during movement).

[0067] When establishing the vehicle body dynamics model, the lifting platform dynamics model, and the dynamics models of each robotic arm based on the rigid body model and the flexible body model, respectively, the dynamic relationship between the two is established (e.g., the sliding pair constraint between the flexible telescopic arm of the third robotic arm and the rigid cylinder outer cylinder, with constraint stiffness 1e). 5 The simulation accuracy of the flexible body deformation is ±0.05mm. Through the above steps, the simulation error caused by the neglect of elastic deformation in the dynamic model of the traditional pure tunnel lining inspection vehicle (which is a rigid model) can be solved, and the simulation accuracy of the deformation at the end of the robotic arm can be improved from ±0.2mm to ±0.05mm, which meets the verification requirements of the 0.1mm sliding pair accuracy of the third robotic arm.

[0068] In one embodiment, the rigid-flexible coupling dynamic model of the vehicle body includes wheel-rail contact constraints and damping constraints of the vehicle body shock absorber (damping coefficient 5000 N·s / m); the rigid-flexible coupling dynamic model of the lifting platform includes the rotary joint constraints of the scissor mechanism and the sliding joint constraints of the hydraulic cylinder (maximum thrust 50 kN).

[0069] In one embodiment, the robotic arm dynamics model includes multiple sets of kinematic pair constraints, which include sliding pair constraints and rotational pair constraints.

[0070] In this embodiment of the invention, a deformation compensation coefficient can be added to the kinematic pair constraint of the flexible body component:

[0071] Sliding pair constraint: such as the sliding pair between the second-stage telescopic arm (flexible body) and the first-stage telescopic arm (rigid body) of the first robotic arm, the constraint stiffness is 1e5N / m, and the deformation compensation coefficient is 0.98 (when the flexible body deforms by 0.05mm, the displacement compensation of the sliding pair is 0.049mm).

[0072] Rotary joint constraint: such as the rotary joint between the scissor link (flexible body) and the base (rigid body) of the fourth robotic arm, with a rotation angle range of -60° to 90° and a deformation compensation coefficient of 0.99 (when the scissor link bends and deforms by 0.03mm, the rotation angle compensation is -0.1°).

[0073] In one embodiment, the robotic arm dynamics model includes dynamics models of a first robotic arm, a second robotic arm, a third robotic arm, a fourth robotic arm, and a fifth robotic arm. The first and second robotic arms are telescopic bending arms and are each equipped with multiple ground-penetrating radar antennas. The third robotic arm is a multi-stage hydraulic arm, and the fourth and fifth robotic arms are scissor arms.

[0074] In one embodiment, the kinematic pair constraints of the first and second robotic arms both include the sliding pairs of the inner and outer cylinders of three hydraulic cylinders, the rotary pair between the first-stage telescopic arm and the base, the sliding pairs between the second-stage telescopic arm and the first-stage telescopic arm, the sliding pairs between the third-stage telescopic arm and the second-stage telescopic arm, the sliding pairs between the fourth-stage telescopic arm and the third-stage telescopic arm, the rotary pair between the lower curved arm and the fourth-stage telescopic arm, the rotary pair between the upper curved arm and the lower curved arm, the rotary pair between the outer cylinder of the first hydraulic cylinder and the base, the rotary pair between the inner cylinder of the first hydraulic cylinder and the first-stage telescopic arm, the rotary pair between the outer cylinder of the second hydraulic cylinder and the fourth-stage telescopic arm, the rotary pair between the inner cylinder of the second hydraulic cylinder and the lower curved arm, the rotary pair between the outer cylinder of the third hydraulic cylinder and the fourth-stage telescopic arm, the rotary pair between the inner cylinder of the third hydraulic cylinder and the lower curved arm, the sliding pair between the first detection radar and the base, the sliding pair between the second detection radar and the base, the sliding pair between the third detection radar and the base, the sliding pair between the first detection radar and the upper curved arm, and the sliding pair between the second detection radar and the upper curved arm.

[0075] In one embodiment, the kinematic pair constraints of the third robotic arm include a primary telescopic arm and base rotary joint, a secondary telescopic arm and primary telescopic arm sliding joint, a tertiary telescopic arm and secondary telescopic arm sliding joint, a quaternary telescopic arm and tertiary telescopic arm sliding joint, a quintuplet telescopic arm and quaternary telescopic arm sliding joint, a first cylinder inner and outer barrel sliding joint, a second cylinder inner and outer barrel sliding joint, a third cylinder inner and outer barrel sliding joint, a first cylinder inner barrel and primary telescopic arm rotary joint, a second cylinder inner barrel and primary telescopic arm rotary joint, a third cylinder inner barrel and secondary telescopic arm rotary joint, a first outer barrel and base rotary joint, and a second cylinder outer barrel and base rotary joint.

[0076] In one embodiment, the kinematic pair constraints of the fourth and fifth robotic arms each include a detection radar sliding pair, a first cylinder inner and outer cylinder sliding pair, a tenth middle link and pin rotary pair, a detection radar and pin sliding pair, a dust cover and base rotary pair, a first cylinder outer cylinder and base rotary pair, a first cylinder inner cylinder and dust cover rotary pair, a first front link and dust cover rotary pair, a first rear link and dust cover rotary pair, a rotary pair between each front link and the next middle link of that front link, a rotary pair between each rear link and the next middle link of that rear link, a rotary pair between the first middle link and the dust cover, a rotary pair between each middle link and the next front link of that middle link, a rotary pair between each middle link and the next rear link of that middle link, a rotary pair between the tenth front link and the detection radar, a tenth rear link and the detection radar rotary pair, a rotary pair between each front link and the corresponding middle link, and a rotary pair between each rear link and the corresponding middle link.

[0077] The revolute joints of each front link and the next middle link of that front link include the first front link and the second middle link revolute joint, the second front link and the third middle link revolute joint, the third front link and the fourth middle link revolute joint, the fourth front link and the fifth middle link revolute joint, the fifth front link and the sixth middle link revolute joint, the sixth front link and the seventh middle link revolute joint, the seventh front link and the eighth middle link revolute joint, the eighth front link and the ninth middle link revolute joint, and the ninth front link and the tenth middle link revolute joint.

[0078] The revolute joints of each rear link and the next middle link of that rear link include the first rear link and the second middle link revolute joint, the second rear link and the third middle link revolute joint, the third rear link and the fourth middle link revolute joint, the fourth rear link and the fifth middle link revolute joint, the fifth rear link and the sixth middle link revolute joint, the sixth rear link and the seventh middle link revolute joint, the seventh rear link and the eighth middle link revolute joint, the eighth rear link and the ninth middle link revolute joint, and the ninth rear link and the tenth middle link revolute joint.

[0079] The revolute joints of each middle link and the next front link of that middle link include the first middle link and the second front link revolute joint, the second middle link and the third front link revolute joint, the third middle link and the fourth front link revolute joint, the fourth middle link and the fifth front link revolute joint, the fifth middle link and the sixth front link revolute joint, the sixth middle link and the seventh front link revolute joint, the seventh middle link and the eighth front link revolute joint, the eighth middle link and the ninth front link revolute joint, and the ninth middle link and the tenth front link revolute joint.

[0080] The revolute joints of each middle link and the next rear link of that middle link include the first middle link and the second rear link revolute joint, the second middle link and the third rear link revolute joint, the third middle link and the fourth rear link revolute joint, the fourth middle link and the fifth rear link revolute joint, the fifth middle link and the sixth rear link revolute joint, the sixth middle link and the seventh rear link revolute joint, the seventh middle link and the eighth rear link revolute joint, the eighth middle link and the ninth rear link revolute joint, and the ninth middle link and the tenth rear link revolute joint.

[0081] Each front link and its corresponding middle link revolute joint includes the first front link and first middle link revolute joint, the second front link and second middle link revolute joint, the third front link and third middle link revolute joint, the fourth front link and fourth middle link revolute joint, the fifth front link and fifth middle link revolute joint, the sixth front link and sixth middle link revolute joint, the seventh front link and seventh middle link revolute joint, the eighth front link and eighth middle link revolute joint, the ninth front link and ninth middle link revolute joint, and the tenth rear link and tenth middle link revolute joint.

[0082] Each rear link and its corresponding middle link revolute joint includes the first rear link and the first middle link revolute joint, the second rear link and the second middle link revolute joint, the third rear link and the third middle link revolute joint, the fourth rear link and the fourth middle link revolute joint, the fifth rear link and the fifth middle link revolute joint, the sixth rear link and the sixth middle link revolute joint, the seventh rear link and the seventh middle link revolute joint, the eighth rear link and the eighth middle link revolute joint, the ninth rear link and the ninth middle link revolute joint, and the tenth rear link and the tenth middle link revolute joint.

[0083] In one embodiment, the method further includes:

[0084] Extract the constraint relationships between the fourth robotic arm, the fifth robotic arm, and the lifting platform;

[0085] Based on the constraint relationships, identify the constrained nodes;

[0086] The constraint stiffness of the constrained nodes has been adjusted.

[0087] To address the over-constraint issue of the fourth and fifth robotic arms and the lifting platform, the constraint stiffness was adjusted. Specific steps included:

[0088] Step 1: Define the parameters and constraints of the scissor lift mechanism

[0089] Obtain the design parameters of the target scissor lift mechanism (fourth robotic arm / fifth robotic arm / lifting platform), including: structural parameters: number of scissor lift links (e.g., the fourth robotic arm contains 10 sets of scissor lift links), dimensions of a single link (length, diameter, wall thickness), and link material (e.g., Q345 steel, elastic modulus 206GPa, Poisson's ratio 0.3).

[0090] Motion parameters: maximum extension stroke of the scissor mechanism, rated speed (e.g., 5 mm / s), and rotation angle range of adjacent link hinge points (e.g., -30°~90°).

[0091] Constraint Types: Define all constraint relationships, including revolute joint constraints at hinge points (such as the revolute joint between the scissor lift and the base, and the revolute joint between adjacent links), sliding joint constraints of hydraulic cylinders (such as the sliding joint between the inner and outer cylinders), and fixed constraints between the dust cover and the base.

[0092] Determine the over-constraint judgment criteria:

[0093] Over-constraint is defined as: when the number of constraints exceeds the minimum number of constraints required for the degree of freedom of the mechanism, and there are two or more constraints that mutually restrict the same degree of freedom of motion (such as two adjacent revolute joints forming conflicting restrictions on the rotation direction of the same link).

[0094] Set a threshold for motion interference: when the scissor lift mechanism is in motion, if the minimum distance between adjacent parts is ≤0.1mm or the overlapping area of ​​the motion trajectory is ≥1mm², it is determined that there is a risk of motion interference.

[0095] Step 2: Set up the parametric modeling environment

[0096] A library of mechanical properties of the materials corresponding to the scissor lift mechanism (such as the density and elastic modulus of Q345 steel).

[0097] Create a new parametric modeling template and define key parameter variables, including:

[0098] Geometric parameter variables: connecting rod length L (e.g., adjustable from 500mm to 800mm), hinge point hole diameter D (e.g., adjustable from 12mm to 15mm);

[0099] Constraint parameter variables: Rotary joint constraint stiffness K_rot (e.g., 1e5N・m / rad ~ 1e6N・m / rad), sliding joint constraint stiffness K_slide (e.g., 1e4N / m ~ 1e5N / m).

[0100] Step 3: Collect measured motion data of the scissor lift mechanism

[0101] Motion tests were conducted on the physical scissor lift mechanism using a high-speed camera (frame rate ≥ 200fps) and a displacement sensor (accuracy ± 0.01mm).

[0102] Control the scissor mechanism to complete 3 full extension and retraction cycles at the rated speed, and record the real-time displacement and rotation angle data of each hinge point;

[0103] Mark the key locations where jamming or abnormal noise occurs during the movement, and initially locate potential over-constraint areas (such as the hinge points of the scissor links in groups 3-4).

[0104] The measured data were compiled into a time-displacement / angle data table, which served as the benchmark data for subsequent simulation verification.

[0105] Step 4: Construct a parametric model of the scissor lift mechanism

[0106] Based on the CAD drawings of the scissor lift mechanism (in STEP or IGES format), and using the parameter variables defined in step 1, complete the geometric model reconstruction:

[0107] Mesh the components such as scissor lift linkage, hydraulic cylinder, and base using tetrahedral elements, with the mesh size controlled between 5mm and 10mm (refined to 2mm in the hinge area).

[0108] Add constraints: Create a revolute joint constraint at the hinge point, a sliding joint constraint at the contact point between the inner and outer cylinders, and a fixed constraint at the fixed component (such as the base).

[0109] Perform static simulations to calculate the stress distribution of the mechanism under initial constraints, mark stress concentration areas (such as locations near constraint nodes where the stress is ≥150MPa), and preliminarily screen high-risk constraints.

[0110] Step 5: Extract the constraint association matrix

[0111] Obtain the node association data for all constraints:

[0112] Establish a constraint-node mapping table to record the node number (e.g., nodes 101-102 corresponding to a revolute joint constraint), constraint type (revolute joint / sliding joint), and constraint direction (X / Y / Z axis) for each constraint.

[0113] Calculate the constraint stiffness matrix K: Based on the constraint type and material parameters, a 6×6 constraint stiffness matrix is ​​generated by a linear solver. The matrix element K_ij represents the stiffness contribution of the i-th constraint to the j-th degree of freedom.

[0114] Identifying constraint dependencies: By calculating the matrix determinant, if the determinant value of the submatrix corresponding to a certain set of constraints approaches 0 (≤1e-6), it is determined that there is a strong dependency relationship between the constraints and they are listed as over-constrained candidate objects.

[0115] Step 6: Quantify motion interference

[0116] Set up motion simulation conditions for the scissor lift mechanism:

[0117] Define the motion path of the scissor mechanism from fully retracted to fully extended according to the rated motion speed in step 1, set the simulation step size to 0.01s, and the total simulation time to be the time required for the mechanism to complete one extension and retraction (e.g., 200s).

[0118] The minimum safe distance between adjacent components is set to 0.2 mm. The motion interference Δ (unit: mm) is calculated in real time for each simulation step, which is the difference between the actual distance between adjacent components and the safe distance (Δ < 0 indicates that interference exists).

[0119] A three-dimensional data table of "simulation step size - constraint stiffness - motion interference" is generated to provide basic data for subsequent optimization.

[0120] Step 7: Build the optimization model

[0121] Define the optimization objective:

[0122] Objective 1 (Minimize constraint stiffness deviation): Ensure that the deviation rate between the adjusted constraint stiffness matrix K' and the initial matrix K is ≤5%, i.e., |(K'_ij-K_ij) / K_ij|×100%≤5%;

[0123] Objective 2 (Minimize motion interference): Ensure that the motion interference Δ ≥ 0 (no interference) at all simulation step sizes, and the average interference (if present) ≤ 0.05 mm;

[0124] Set design variables: Take the relaxation coefficient α of each constraint as the design variable, α∈[0.1,0.8] (α=1 means no relaxation, the smaller α is, the lower the constraint stiffness), and define the variable boundaries: α≥0.3 for hinge point rotary joint (to avoid the mechanism shaking due to excessively low stiffness), α≥0.2 for hydraulic cylinder sliding joint.

[0125] Step 8: Configure the parameters of the sequential quadratic programming algorithm

[0126] Select the Sequential Quadratic Programming (SQP) algorithm and set the algorithm parameters:

[0127] Convergence accuracy: The gradient norm of the objective function ≤ 1e -4 The iteration step size of the variable is ≤0.01;

[0128] Number of iterations: The maximum number of iterations is set to 100. If the target is not improved after 50 iterations, the local optimum exit mechanism will be triggered.

[0129] Weighting: The dual objectives are weighted (objective 1 weight 0.4, objective 2 weight 0.6), transforming it into a single-objective optimization problem: MinF=0.4×(ΔK / K)+0.6×(Δ / Δ_max) (Δ_max is the maximum allowable interference amount of 0.1mm); to achieve an automatic loop of constraint stiffness adjustment-motion simulation-interference amount calculation during the optimization process.

[0130] Step 9: Automatically identify nodes that have exceeded constraints

[0131] Initiate optimization calculations and identify constrained nodes according to the following logic:

[0132] Iterations 1-10: Sort by the determinant value of the constraint stiffness matrix, and prioritize the 10 constraint nodes (strongly dependent constraints) with the smallest determinant value to adjust the relaxation coefficient, with an adjustment step of 0.05 each time;

[0133] Iterations 11-50: Combining motion interference data, for the constraint nodes corresponding to the interference region where Δ < 0, increase the relaxation coefficient adjustment range (step size 0.1), while monitoring the interference changes of other constraint nodes;

[0134] Iterations 51-100: For the remaining constraint nodes that do not meet the objective, a local search strategy is adopted to refine the search within the range of α∈[0.1,0.8] to determine the optimal relaxation coefficient;

[0135] Output a list of constrained nodes, including node number, initial constraint stiffness, optimized relaxation coefficient, and corresponding interference improvement rate (e.g., interference changes from -0.15mm to 0.08mm, improvement rate 153%).

[0136] Step 10: Dynamically adjust constraint stiffness

[0137] The optimized relaxation coefficient α is automatically synchronized to the parameterized model:

[0138] The constraint stiffness of the over-constrained nodes is corrected by the following formula: K'_ij=α×K_ij (K'_ij is the adjusted constraint stiffness, and K_ij is the initial constraint stiffness).

[0139] For non-over-constrained nodes, the initial constraint stiffness is kept constant to avoid excessive relaxation that could lead to a decrease in mechanism stability. A constraint stiffness adjustment report is generated, recording the parameter comparison before and after adjustment for each constraint node, including constraint type, initial K_ij, adjusted K'_ij, relaxation coefficient α, and corresponding changes in motion interference.

[0140] Step 11: Optimized motion simulation verification

[0141] Based on the adjusted constraint stiffness model, repeat the motion simulation conditions in step 6:

[0142] Complete three full expansion and contraction cycle simulations, and record the motion interference, stress distribution of the mechanism, and deviation of the motion trajectory from the measured data for each simulation step.

[0143] Verification indicators: motion interference Δ≥0, maximum stress ≤120MPa (below the material yield strength of 345MPa), and deviation rate between motion trajectory and measured data ≤3%;

[0144] If all validation metrics are met, proceed to step 13; otherwise, return to step 8, adjust the weight allocation or iteration parameters of the SQP algorithm, and re-optimize.

[0145] Step 12: Physical prototype testing and verification

[0146] Based on the optimized constraint parameters, a physical prototype of the scissor lift mechanism (such as the fourth robotic arm scissor lift mechanism) is fabricated, and motion testing is performed using the testing equipment from step 3.

[0147] Compare the motion performance of the prototype before and after optimization: including the extension cycle time (e.g., 210s before optimization, 200s after optimization, matching the rated speed), the number of times it jams (3 times per cycle before optimization, no jams after optimization), and the occurrence of abnormal noises (no abnormal noises after optimization).

[0148] The actual stress at key constraint nodes was measured using a stress strain gauge (accuracy ±1με) to verify whether it was consistent with the simulation results (deviation ≤10%).

[0149] If the deviation between the physical test and simulation results exceeds 10%, return to step 5, re-extract the constraint correlation matrix, check for modeling errors (such as incorrect mesh size or material parameter input), and optimize again.

[0150] Step 13: Algorithm parameter fixation and iterative update

[0151] The optimized parameters (relaxation coefficient α, algorithm parameters, constraint stiffness adjustment rules) that meet the verification requirements are solidified for rapid adaptation to similar scissor lift mechanisms (such as scissor lift arms of other models of inspection vehicles).

[0152] Establish an iterative mechanism: for every 10 new test data sets of physical scissor lift mechanisms, retrain and optimize the weight parameters of the model (such as the weight allocation of the objective function) to enable the algorithm to adapt to the over-constraint problem of scissor lift mechanisms under different working conditions (such as the change of constraint stiffness in low temperature environment).

[0153] The final output includes an optimized parametric model file and a list of over-constrained nodes, forming a complete technical loop.

[0154] In one embodiment, the method further includes:

[0155] Before obtaining the environmental data and vehicle status data collected in real time by the detection sensors on the tunnel lining inspection vehicle based on the finite element model, a structural reliability analysis is performed on the multibody dynamics digital twin model to obtain the structural reliability analysis results.

[0156] The system obtains real-time environmental data and vehicle status data collected by the sensors mounted on the tunnel lining inspection vehicle, including:

[0157] When the structural reliability analysis results meet the structural reliability requirements, the environmental data and vehicle status data collected in real time by the detection sensors mounted on the tunnel lining inspection vehicle are obtained.

[0158] In this embodiment of the invention, the finite element model is used to supplement and verify the structural safety of the multibody dynamics digital twin model. If the structural reliability analysis result shows that the stress or deformation of the robotic arm exceeds the standard, it is necessary to return to step 101 to adjust the part parameters (such as material and size) of the multibody dynamics digital twin model, and then remodel to ensure that the multibody dynamics digital twin model is structurally feasible. Only when the finite element model verifies that the structural reliability analysis result meets the structural reliability requirements can the dynamic simulation based on the data collected by the detection sensors have practical significance, avoiding the virtual failure of the robotic arm due to structural problems in the simulation. The core logic of remodeling is to solve the structural defects of the original design, which will inevitably lead to differences between the model parameters and the original tunnel and detection vehicle parameter data. There are two specific reasons:

[0159] The original inspection vehicle design may have contained theoretical flaws: the initial multibody dynamics digital twin model was built at a 1:1 scale to the original inspection vehicle's design dimensions. However, the original design may have only met functional requirements (such as the ability of the robotic arm to extend and fold), without fully considering structural strength under extreme conditions (such as the maximum stress when the equipment is deployed). The core value of finite element analysis is to expose such theoretical flaws in advance. If the model parameters are not adjusted (such as increasing the size of key connecting rods or replacing them with higher-strength materials), and the physical equipment is manufactured directly according to the original dimensions, it may lead to equipment damage during actual operation.

[0160] Model adjustments are pre-optimization, and ultimately need to be synchronized to the physical inspection vehicle: One of the core functions of the multibody dynamics digital twin model is virtual verification and optimization, which is equivalent to a digital prototype of the physical inspection vehicle. When model parameters are adjusted due to excessive stress, this adjustment is not arbitrary, but provides a basis for the subsequent design iteration of the physical equipment. For example, if the diameter of the robotic arm link is adjusted from 10mm to 12mm in the model and the stress meets the standard, then the physical inspection vehicle will be manufactured with a diameter of 12mm, ultimately achieving a perfect match between the adjusted model parameters and the optimized physical equipment.

[0161] For example, the force conditions of the first and second robotic arms during operation are analyzed based on the finite element model, and two working conditions are analyzed.

[0162] Operating Condition 1: Running Status

[0163] In addition to its own weight, the robotic arm is subjected to additional loads from the vehicle's power system, the work unit, and uneven road conditions, which are equivalent to dynamic loads (coefficient taken as 1). The overall load on the model is equivalent to twice the weight. Calculations show that the maximum deformation is located at the top of the robotic arm, with a displacement of 0.54 mm. The maximum stress is located near the actuator support at the bottom of the robotic arm, with a stress of 55 MPa.

[0164] Operating Condition 2: Equipment Deployment

[0165] During deployment and retraction, the robotic arm is subjected to the driving force of the hydraulic cylinder and the drag force of the wire rope, respectively, and is supported by the mounting and drive connection points of the lower robotic arm. The load in this operating condition is equivalent to the force and torque at the mounting and drive connection points of the lower robotic arm, and is analyzed using an inertial release calculation method. The maximum deformation is located at the tip of the probe arm, with a displacement of 1.1 mm. The maximum stress is located near the hinge connection of the probe arm, with a stress of 101.1 MPa.

[0166] Analysis Conclusion: Under operating and equipment deployment conditions, the maximum deformation of the robotic arm is 1.1 mm, and the maximum stress is 101.1 MPa. During operation and equipment deployment, the maximum stress borne by the robotic arm is 101.1 MPa, which does not reach the material's yield strength. The maximum deformation of the robotic arm is relatively small. Calculation results show that the stiffness and strength of the probe arm meet the design requirements, meaning the structural reliability analysis results meet the structural reliability requirements. In this case, compared to the traditional kinematic model, the control accuracy is improved.

[0167] In step 102, the environmental data and vehicle status data collected in real time by the detection sensors mounted on the tunnel lining inspection vehicle are obtained;

[0168] In one embodiment, the environmental data is obstacle point cloud data, and the detection vehicle status data includes obstacle location information, distance information between the tunnel lining detection vehicle and the obstacle, and speed information of the tunnel lining detection vehicle.

[0169] In two embodiments of the present invention, multiple detection sensors (e.g., nine) are incorporated, and these sensors can be radars. The radar interface table is shown in Table 1.

[0170] Table 1

[0171]

[0172] In this embodiment of the invention, the inspection vehicle can also be equipped with multi-point cameras, which mainly output status images of various points of interest, so as to provide operators with comprehensive monitoring of the inspection mechanism and inspection status, and provide a basis for manual intervention.

[0173] In step 103, based on environmental data and vehicle status data, the multibody dynamics digital twin model is driven to perform synchronous simulation to generate simulation detection information;

[0174] In this embodiment of the invention, a mature dynamics engine can be used, and there are no restrictions. The smaller the simulation step size, the better the simulation effect. For example, the simulation step size can be less than 20ms. The simulation step size needs to meet the response requirements of the hydraulic system.

[0175] In this embodiment of the invention, the fixed 20ms step size mode is abandoned, and a hydraulic response characteristic-obstacle level correlation model is established. When a level 1 obstacle (protruding nail) is detected, the step size is automatically adjusted to 10ms; when a level 3 obstacle (variable cross-section) is detected, the step size switches to 5ms, and GPU acceleration calculation is invoked simultaneously. Under the premise of meeting the hydraulic system response requirements (≤15ms), the simulation efficiency is improved by 40%, avoiding waste of computing resources.

[0176] In one embodiment, the simulation detection information includes obstacle type, obstacle location simulation information, distance simulation information between the tunnel lining inspection vehicle and the obstacle, and speed simulation information of the tunnel lining inspection vehicle;

[0177] During the simulation, the material information of the obstacles can be analyzed from the obstacle point cloud data, thereby determining the obstacle type. Based on the obstacle location information, the distance between the tunnel lining inspection vehicle and the obstacle, and the speed information of the tunnel lining inspection vehicle, the obstacle location simulation information, the distance between the tunnel lining inspection vehicle and the obstacle, and the speed simulation information of the tunnel lining inspection vehicle are simultaneously simulated.

[0178] In this embodiment of the invention, during the simulation execution of the multibody dynamics digital twin model, all information during the simulation execution process can be displayed through the simulation interface. This includes the output results of the sensor scan, the radar control process, the control process of the robotic arm and vehicle, the speed and position of the detection vehicle, and the position and angle of the lifting platform and the main joints of each robotic arm. Users can also control the simulation process through the simulation interface; for example, the simulation interface includes a start / stop button, and users can input the scanning angle and speed of the sensor. The simulation system formed by the above method includes a hardware control interface that can be directly connected to a game controller to control the movement speed, displacement, angular velocity, and angular displacement of the vehicle body, lifting platform, and robotic arms 1-5 in real time.

[0179] In step 104, based on the simulation detection information, the corresponding hierarchical obstacle avoidance strategy for the multi-manipulator system is simulated and executed in the multibody dynamics digital twin model to generate control commands. The control commands include control targets and control actions. The hierarchical obstacle avoidance strategy includes non-stop obstacle avoidance strategy and stop obstacle avoidance strategy.

[0180] In one embodiment, the graded obstacle avoidance strategy includes:

[0181] If the obstacle type is a protruding nail, the graded obstacle avoidance strategy is to only retrieve the ground radar antenna containing the obstacle's location information;

[0182] If the obstacle type is a contact wire frame, the graded obstacle avoidance strategy is to fold the third robotic arm.

[0183] If the obstacle type is a tunnel with a variable cross-section, the graded obstacle avoidance strategy is to stop the vehicle and collaboratively retract the telescopic parts of the first, second, and fourth robotic arms.

[0184] In one embodiment, based on the simulation detection information, the corresponding hierarchical obstacle avoidance strategy for the multi-manipulator system is executed in the multibody dynamics digital twin model simulation, generating control commands, including:

[0185] Based on the simulation information of the distance between the tunnel lining inspection vehicle and the obstacle, the speed simulation information of the tunnel lining inspection vehicle, and the interruption time requirements of different graded obstacle avoidance strategies, the control speed is determined.

[0186] When the graded obstacle avoidance strategy is to only retract the ground radar antenna that provides obstacle location information, the control target is determined to be the inner and outer cylinder sliding pair of the second cylinder of the tunnel lining inspection vehicle. The control action is determined to retract the ground radar antenna according to the control speed and extend the ground radar antenna after the tunnel lining inspection vehicle has passed.

[0187] When the graded obstacle avoidance strategy is to fold the third robotic arm, the control targets are determined to be the sliding joints of the second-level telescopic arm and the first-level telescopic arm, the sliding joints of the third-level telescopic arm and the second-level telescopic arm, the sliding joints of the fourth-level telescopic arm and the third-level telescopic arm, and the sliding joints of the fifth-level telescopic arm and the fourth-level telescopic arm. The control action is to retract the third robotic arm according to the control speed and extend the ground radar antenna after waiting for the tunnel lining inspection vehicle to pass.

[0188] When the graded obstacle avoidance strategy involves stopping and collaboratively retracting the telescopic parts of the first, second, and fourth robotic arms, four control commands are determined. The first control command targets the tunnel lining inspection vehicle, and its action is to stop. The second control command targets the rotating joints of the upward and downward bending arms, the rotating joints of the downward bending arm and the fourth-stage telescopic arm, the sliding joints of the second-stage and first-stage telescopic arms, the sliding joints of the third-stage and second-stage telescopic arms, and the sliding joints of the fourth-stage and third-stage telescopic arms, and its action is to retract the folding robotic arms at the controlled speed and wait for completion. The third control command targets the tunnel lining inspection vehicle, and its action is to start and wait for the vehicle to completely pass. The fourth control command has the same target as the second control command, and its action is to restore the robotic arms.

[0189] In a specific embodiment, when the obstacle type is a contact wire frame, the interruption time of the corresponding graded obstacle avoidance strategy is <1s; when the obstacle type is a contact wire frame, the interruption time of the corresponding graded obstacle avoidance strategy is <3s; when the obstacle type is a tunnel with a variable cross section, the interruption time of the corresponding graded obstacle avoidance strategy is <5s. Based on the above interruption time requirements, as well as the simulation information of the distance between the tunnel lining inspection vehicle and the obstacle and the simulation information of the speed of the tunnel lining inspection vehicle, the control speed can be determined.

[0190] In step 105, the control command is issued according to the control objective.

[0191] In this embodiment of the invention, control commands are sent to the physical inspection vehicle via the UDP communication protocol or a hardware control interface.

[0192] In one embodiment, the method further includes:

[0193] The priority of each graded obstacle avoidance strategy is marked so that when the tunnel lining inspection vehicle receives control instructions that include multiple graded obstacle avoidance strategies, it executes control actions according to the priority.

[0194] The priority is calculated based on the obstacle distance and obstacle type, weighted accordingly.

[0195] In this embodiment of the invention, existing priority rules can be optimized to construct a dynamic evaluation model of obstacle threat level and task weight: the threat level is calculated based on the obstacle distance (≤5m high threat) and obstacle type (variable cross-section > space frame > protruding nail), and the task weight is dynamically allocated according to the radar detection area (arch > side wall > base plate). Temporary priority overloading is supported (e.g., priority can be increased for specific robotic arms in emergency detection tasks).

[0196] For example, when the arch contact grid (level 2) and the side wall protrusion nail (level 1) appear at the same time, the grid obstacle avoidance (higher task weight) is executed first, and the protrusion nail obstacle avoidance is executed with a delay of 0.5 seconds.

[0197] In addition, for level 3 obstacles, a multi-arm kinematic decoupling algorithm can be used: the joint motion space of the first, second and fourth robotic arms is calculated by using the Jacobian matrix, and a collision-free retraction path is automatically planned to ensure that the retraction time difference of the telescopic arm is ≤0.2s and the rotation angle error of the scissor link is ≤±1°.

[0198] Before generating control commands, a 100ms pre-simulation step is added: based on the current environmental data, the obstacle avoidance action is simulated, and the minimum safe distance between the end of the robotic arm and the obstacle is calculated (≥50mm is acceptable). If it is unacceptable, the control speed is readjusted (adjustment step size 0.1mm / s), the obstacle avoidance action error rate is calculated, and the risk of physical damage caused by virtual failure is completely resolved.

[0199] The following is a specific embodiment to illustrate the specific application of the method proposed in this invention.

[0200] After the simulation starts, click the start button on the simulation interface. The robotic arm of the inspection vehicle will automatically extend to the working state. Set the speed and position information of the inspection vehicle and start the vehicle.

[0201] The inspection vehicle travels in the virtual tunnel at a predetermined speed, the robotic arm is normally extended, and the ground-penetrating radar antenna scans the lining surface. During this time, the position and angle of the robotic arm, lifting platform and ground-penetrating radar antenna can be adjusted through the handle and simulation interface.

[0202] If the protruding nail is detected at a distance of 0.5m in front of the inspection vehicle, such as Figure 8 The diagram shows the discovery of the protruding nail in the embodiment of the present invention. The obstacle avoidance strategy with a priority of level 1 is activated, the control command is determined and issued: the ground-penetrating radar antenna retracts into the safety chamber within 1 second, the robotic arm continues to maintain the extended state, and the ground-penetrating radar antenna automatically resets after passing the protruding nail, reducing the scanning interruption time.

[0203] If the inspection vehicle identifies the overhead contact line frame at a distance of 5 meters in front of it, such as Figure 9 In this embodiment of the invention, a schematic diagram of the contact wire frame is found. An obstacle avoidance strategy with a priority of level 2 is activated, a control command is determined, and the following is issued: the third robotic arm quickly retracts, the other robotic arms work normally, and the third robotic arm automatically resets after the detection vehicle passes the obstacle.

[0204] If a variable cross-section is detected 5m in front of the detection vehicle, such as Figure 10 In this embodiment of the invention, a schematic diagram of a variable cross section is found. An obstacle avoidance strategy with a priority of level 3 is initiated, a control command is determined, and issued as follows: the tunnel inspection vehicle stops, the first, third, and fourth robotic arms quickly retract, and then after the inspection vehicle starts and passes the obstacle, the three robotic arms automatically reset and continue inspection and driving.

[0205] Depend on Figure 8 , Figure 9 , Figure 10 As can be seen, in condition 1, after detecting a protruding nail, the ground-penetrating radar antenna retracts, the detection vehicle does not stop, and the remaining ground-penetrating radar antennas operate normally; in condition 2, after detecting a contact network frame, the third robotic arm retracts, the detection vehicle does not stop, and the remaining ground-penetrating radar antennas operate normally; in condition 3, after detecting a variable cross-section, the detection vehicle automatically avoids the obstacle and drives over it. In all three conditions, the detection vehicle automatically controls its passage over obstacles, and the unaffected ground-penetrating radar antennas operate normally, significantly reducing the detection interruption rate. Therefore, the control method proposed in this embodiment of the invention is effective.

[0206] This invention also proposes a real-time graded obstacle avoidance control device for a tunnel lining inspection vehicle. Figure 11This is a schematic diagram of the real-time graded obstacle avoidance control device for a tunnel lining inspection vehicle in an embodiment of the present invention, including:

[0207] The model building module 1101 is used to build a multi-body dynamics digital twin model of the tunnel lining inspection vehicle based on tunnel environment data and parameter data of the tunnel lining inspection vehicle. The multi-body dynamics digital twin model includes a virtual tunnel environment model and a dynamic model of the tunnel lining inspection vehicle. The virtual tunnel environment includes various obstacle types, and the dynamic model of the tunnel lining inspection vehicle includes the dynamic model of each robotic arm.

[0208] Data acquisition module 1102 is used to obtain environmental data and vehicle status data collected in real time by the detection sensors mounted on the tunnel lining inspection vehicle;

[0209] Simulation module 1103 is used to drive the multibody dynamics digital twin model to perform synchronous simulation based on environmental data and vehicle status data, and generate simulation detection information;

[0210] The control command generation module 1104 is used to generate control commands based on the simulation detection information, and to simulate and execute the corresponding hierarchical obstacle avoidance strategy for the multi-manipulator system in the multibody dynamics digital twin model. The control commands include control targets and control actions. The hierarchical obstacle avoidance strategy includes non-stop obstacle avoidance strategy and stop obstacle avoidance strategy.

[0211] The instruction issuing module 1105 is used to issue the control instructions according to the control target.

[0212] In one embodiment, the model building module 1101 is used for:

[0213] The load-bearing components of the tunnel lining inspection vehicle are modeled as rigid bodies.

[0214] Based on the parameter data of the tunnel lining inspection vehicle, a modal neutral file is generated according to the easily deformable parts of the tunnel lining inspection vehicle, and a flexible body model is constructed based on the modal neutral file;

[0215] Based on the rigid body model and the flexible body model, the vehicle body dynamics model, the lifting platform dynamics model, and the robotic arm dynamics model of each robotic arm are established respectively. Among them, the vehicle body dynamics model is a coupling model of the rigid vehicle body and the flexible wheel axle, the lifting platform dynamics model is a coupling model of the rigid fixed frame and the flexible scissor lift linkage, and the robotic arm dynamics model is a coupling model of the rigid base and the flexible robotic arm.

[0216] In one embodiment, the obstacle type includes at least one of spikes, contact wire frames, and tunnels with variable cross-sections;

[0217] If the obstacle type is a protruding nail and a contact wire frame, the graded obstacle avoidance strategy is a non-stop obstacle avoidance strategy.

[0218] If the obstacle type is a tunnel with variable cross-section, the graded obstacle avoidance strategy is a parking obstacle avoidance strategy.

[0219] In one embodiment, the dynamic model of the tunnel lining inspection vehicle includes a vehicle body dynamic model and a lifting platform dynamic model.

[0220] In one embodiment, the robotic arm dynamics model includes multiple sets of kinematic pair constraints, which include sliding pair constraints and rotational pair constraints.

[0221] In one embodiment, the robotic arm dynamics model includes dynamics models of a first robotic arm, a second robotic arm, a third robotic arm, a fourth robotic arm, and a fifth robotic arm. The first and second robotic arms are telescopic bending arms and are each equipped with multiple ground-penetrating radar antennas. The third robotic arm is a multi-stage hydraulic arm, and the fourth and fifth robotic arms are scissor arms.

[0222] In one embodiment, the kinematic pair constraints of the first and second robotic arms both include the sliding pairs of the inner and outer cylinders of three hydraulic cylinders, the rotary pair between the first-stage telescopic arm and the base, the sliding pairs between the second-stage telescopic arm and the first-stage telescopic arm, the sliding pairs between the third-stage telescopic arm and the second-stage telescopic arm, the sliding pairs between the fourth-stage telescopic arm and the third-stage telescopic arm, the rotary pair between the lower curved arm and the fourth-stage telescopic arm, the rotary pair between the upper curved arm and the lower curved arm, the rotary pair between the outer cylinder of the first hydraulic cylinder and the base, the rotary pair between the inner cylinder of the first hydraulic cylinder and the first-stage telescopic arm, the rotary pair between the outer cylinder of the second hydraulic cylinder and the fourth-stage telescopic arm, the rotary pair between the inner cylinder of the second hydraulic cylinder and the lower curved arm, the rotary pair between the outer cylinder of the third hydraulic cylinder and the fourth-stage telescopic arm, the rotary pair between the inner cylinder of the third hydraulic cylinder and the lower curved arm, the sliding pair between the first detection radar and the base, the sliding pair between the second detection radar and the base, the sliding pair between the third detection radar and the base, the sliding pair between the first detection radar and the upper curved arm, and the sliding pair between the second detection radar and the upper curved arm.

[0223] In one embodiment, the kinematic pair constraints of the third robotic arm include a primary telescopic arm and base rotary joint, a secondary telescopic arm and primary telescopic arm sliding joint, a tertiary telescopic arm and secondary telescopic arm sliding joint, a quaternary telescopic arm and tertiary telescopic arm sliding joint, a quintuplet telescopic arm and quaternary telescopic arm sliding joint, a first cylinder inner and outer barrel sliding joint, a second cylinder inner and outer barrel sliding joint, a third cylinder inner and outer barrel sliding joint, a first cylinder inner barrel and primary telescopic arm rotary joint, a second cylinder inner barrel and primary telescopic arm rotary joint, a third cylinder inner barrel and secondary telescopic arm rotary joint, a first outer barrel and base rotary joint, and a second cylinder outer barrel and base rotary joint.

[0224] In one embodiment, the kinematic pair constraints of the fourth and fifth robotic arms each include a detection radar sliding pair, a first cylinder inner and outer cylinder sliding pair, a tenth middle link and pin rotary pair, a detection radar and pin sliding pair, a dust cover and base rotary pair, a first cylinder outer cylinder and base rotary pair, a first cylinder inner cylinder and dust cover rotary pair, a first front link and dust cover rotary pair, a first rear link and dust cover rotary pair, a rotary pair between each front link and the next middle link of that front link, a rotary pair between each rear link and the next middle link of that rear link, a rotary pair between the first middle link and the dust cover, a rotary pair between each middle link and the next front link of that middle link, a rotary pair between each middle link and the next rear link of that middle link, a rotary pair between the tenth front link and the detection radar, a tenth rear link and the detection radar rotary pair, a rotary pair between each front link and the corresponding middle link, and a rotary pair between each rear link and the corresponding middle link.

[0225] In one embodiment, the apparatus further includes a finite element analysis module for:

[0226] Before obtaining the environmental data and vehicle status data collected in real time by the detection sensors on the tunnel lining inspection vehicle based on the finite element model, a structural reliability analysis is performed on the multibody dynamics digital twin model to obtain the structural reliability analysis results.

[0227] The data acquisition module is used for:

[0228] When the structural reliability analysis results meet the structural reliability requirements, the environmental data and vehicle status data collected in real time by the detection sensors mounted on the tunnel lining inspection vehicle are obtained.

[0229] In one embodiment, the environmental data is obstacle point cloud data, and the detection vehicle status data includes obstacle location information, distance information between the tunnel lining detection vehicle and the obstacle, and speed information of the tunnel lining detection vehicle;

[0230] The simulation detection information includes obstacle type, obstacle location simulation information, distance simulation information between the tunnel lining inspection vehicle and the obstacle, and speed simulation information of the tunnel lining inspection vehicle.

[0231] In one embodiment, the graded obstacle avoidance strategy includes:

[0232] If the obstacle type is a protruding nail, the graded obstacle avoidance strategy is to only retrieve the ground radar antenna containing the obstacle's location information;

[0233] If the obstacle type is a contact wire frame, the graded obstacle avoidance strategy is to fold the third robotic arm.

[0234] If the obstacle type is a tunnel with a variable cross-section, the graded obstacle avoidance strategy is to stop the vehicle and collaboratively retract the telescopic parts of the first, second, and fourth robotic arms.

[0235] In one embodiment, the control command generation module is used to:

[0236] Based on the simulation information of the distance between the tunnel lining inspection vehicle and the obstacle, the speed simulation information of the tunnel lining inspection vehicle, and the interruption time requirements of different graded obstacle avoidance strategies, the control speed is determined.

[0237] When the graded obstacle avoidance strategy is to only retract the ground radar antenna that provides obstacle location information, the control target is determined to be the inner and outer cylinder sliding pair of the second cylinder of the tunnel lining inspection vehicle. The control action is determined to retract the ground radar antenna according to the control speed and extend the ground radar antenna after the tunnel lining inspection vehicle has passed.

[0238] When the graded obstacle avoidance strategy is to fold the third robotic arm, the control targets are determined to be the sliding joints of the second-level telescopic arm and the first-level telescopic arm, the sliding joints of the third-level telescopic arm and the second-level telescopic arm, the sliding joints of the fourth-level telescopic arm and the third-level telescopic arm, and the sliding joints of the fifth-level telescopic arm and the fourth-level telescopic arm. The control action is to retract the third robotic arm according to the control speed and extend the ground radar antenna after waiting for the tunnel lining inspection vehicle to pass.

[0239] When the graded obstacle avoidance strategy involves stopping and collaboratively retracting the telescopic parts of the first, second, and fourth robotic arms, four control commands are determined. The first control command targets the tunnel lining inspection vehicle, and its action is to stop. The second control command targets the rotating joints of the upward and downward bending arms, the rotating joints of the downward bending arm and the fourth-stage telescopic arm, the sliding joints of the second-stage and first-stage telescopic arms, the sliding joints of the third-stage and second-stage telescopic arms, and the sliding joints of the fourth-stage and third-stage telescopic arms, and its action is to retract the folding robotic arms at the controlled speed and wait for completion. The third control command targets the tunnel lining inspection vehicle, and its action is to start and wait for the vehicle to completely pass. The fourth control command has the same target as the second control command, and its action is to restore the robotic arms.

[0240] In summary, the method and apparatus proposed in this invention utilize a multi-body dynamics digital twin model constructed based on tunnel environment data and inspection vehicle parameter data. This model includes a virtual tunnel environment model and an inspection vehicle dynamics model, where the virtual tunnel environment covers various obstacle types. Compared to traditional simplified models that can only simulate a single fixed scene, this technology can accurately reproduce the actual obstacle combinations in railway tunnels, solving the problem of the disconnect between the virtual scene and the real environment, and providing a reliable model foundation for subsequent simulation and obstacle avoidance control. The inspection vehicle dynamics model can fully present the motion characteristics of each component of the inspection vehicle, improving the simulation accuracy of the deformation at the end of the robotic arm and avoiding data distortion or obstacle avoidance deviation caused by model errors. Environmental data and inspection vehicle status data are collected in real time through the detection sensors mounted on the inspection vehicle. Compared to traditional single sensors that can only collect local data (such as only obtaining obstacle positions), this technology collects more comprehensive and accurate data, with a high accuracy rate in obstacle type identification. Based on the real-time collected environmental data and inspection vehicle status data, the multi-body dynamics digital twin model is driven to perform synchronous simulation, generating simulation detection information. Compared to the limitations of traditional offline simulations that cannot match real-time environmental changes, this system achieves a linkage mechanism between real-time data input and dynamic simulation output. Simulation results can reflect the dynamic changes of the tunnel environment and the inspection vehicle in real time, with a small deviation rate between simulation and real data. Based on the obstacle type and status in the simulation detection information, corresponding hierarchical obstacle avoidance strategies are executed. Compared to the traditional one-size-fits-all obstacle avoidance method (such as stopping to avoid obstacles regardless of type), this system can dynamically adjust obstacle avoidance actions according to the obstacle threat level, reducing unnecessary work interruptions and improving detection efficiency. Control commands are categorized according to control objectives and sent to the physical inspection vehicle, ensuring that different components can accurately receive the corresponding control actions, reducing command transmission confusion and delays, and ensuring high accuracy in control command execution, avoiding equipment malfunctions caused by incorrect command transmission. Through model construction, data perception, simulation-driven, obstacle avoidance control, and command execution, the system improves the effectiveness in three core dimensions: accuracy, efficiency, and safety.

[0241] This invention also provides a computer device. Figure 12 This is a schematic diagram of a computer device in an embodiment of the present invention. The computer device 1200 includes a memory 1210, a processor 1220, and a computer program 1230 stored in the memory 1210 and executable on the processor 1220. When the processor 1220 executes the computer program 1230, it implements the above-mentioned real-time graded obstacle avoidance control method for tunnel lining inspection vehicle.

[0242] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described real-time graded obstacle avoidance control method for tunnel lining inspection vehicles.

[0243] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described real-time graded obstacle avoidance control method for tunnel lining inspection vehicles.

[0244] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0245] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0246] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0247] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0248] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A real-time graded obstacle avoidance control method for a tunnel lining inspection vehicle, characterized in that, include: Based on tunnel environment data and parameter data of the tunnel lining inspection vehicle, a multibody dynamics digital twin model of the tunnel lining inspection vehicle is constructed. The multibody dynamics digital twin model includes a virtual tunnel environment model and a dynamics model of the tunnel lining inspection vehicle. The virtual tunnel environment includes various obstacle types, and the dynamics model of the tunnel lining inspection vehicle includes the dynamics models of each robotic arm. The system obtains real-time environmental data and vehicle status data collected by the detection sensors mounted on the tunnel lining inspection vehicle. Based on environmental data and vehicle status data, the multibody dynamics digital twin model is driven to perform synchronous simulation to generate simulation detection information. Based on the simulation detection information, the corresponding hierarchical obstacle avoidance strategy for the multi-manipulator system is executed in the multibody dynamics digital twin model simulation, and control commands are generated. The control commands include control targets and control actions. The hierarchical obstacle avoidance strategy includes non-stop obstacle avoidance strategy and stop obstacle avoidance strategy. The control commands are issued according to the control objectives.

2. The method as described in claim 1, characterized in that, Obstacle types include at least one of the following: protruding spikes, overhead contact lines, and tunnels with variable cross-sections; If the obstacle type is a protruding nail and a contact wire frame, the graded obstacle avoidance strategy is a non-stop obstacle avoidance strategy. If the obstacle type is a tunnel with variable cross-section, the graded obstacle avoidance strategy is a parking obstacle avoidance strategy.

3. The method as described in claim 1, characterized in that, The dynamic model of the tunnel lining inspection vehicle also includes the vehicle body dynamic model and the lifting platform dynamic model.

4. The method as described in claim 3, characterized in that, Based on tunnel environmental data and parameter data of the tunnel lining inspection vehicle, a multibody dynamics digital twin model of the tunnel lining inspection vehicle is constructed, including: The load-bearing components of the tunnel lining inspection vehicle are modeled as rigid bodies. Based on the parameter data of the tunnel lining inspection vehicle, a modal neutral file is generated according to the easily deformable parts of the tunnel lining inspection vehicle, and a flexible body model is constructed based on the modal neutral file; Based on the rigid body model and the flexible body model, the vehicle body dynamics model, the lifting platform dynamics model, and the robotic arm dynamics model of each robotic arm are established respectively. Among them, the vehicle body dynamics model is a coupling model of the rigid vehicle body and the flexible wheel axle, the lifting platform dynamics model is a coupling model of the rigid fixed frame and the flexible scissor lift linkage, and the robotic arm dynamics model is a coupling model of the rigid base and the flexible robotic arm.

5. The method as described in claim 3, characterized in that, The robotic arm dynamics model includes multiple sets of kinematic pair constraints, which include sliding pair constraints and rotary pair constraints.

6. The method as described in claim 5, characterized in that, The dynamic model of the robotic arm includes the dynamic models of the first robotic arm, the second robotic arm, the third robotic arm, the fourth robotic arm, and the fifth robotic arm. The first and second robotic arms are telescopic bending arms and are equipped with multiple ground-penetrating radar antennas. The third robotic arm is a multi-stage hydraulic arm, and the fourth and fifth robotic arms are scissor arms.

7. The method as described in claim 1, characterized in that, Also includes: Before obtaining the environmental data and vehicle status data collected in real time by the detection sensors on the tunnel lining inspection vehicle based on the finite element model, a structural reliability analysis is performed on the multibody dynamics digital twin model to obtain the structural reliability analysis results. The system obtains real-time environmental data and vehicle status data collected by the sensors mounted on the tunnel lining inspection vehicle, including: When the structural reliability analysis results meet the structural reliability requirements, the environmental data and vehicle status data collected in real time by the detection sensors mounted on the tunnel lining inspection vehicle are obtained.

8. The method as described in claim 1, characterized in that, The environmental data is obstacle point cloud data, and the detection vehicle status data includes obstacle location information, distance information between the tunnel lining detection vehicle and the obstacle, and speed information of the tunnel lining detection vehicle. The simulation detection information includes obstacle type, obstacle location simulation information, distance simulation information between the tunnel lining inspection vehicle and the obstacle, and speed simulation information of the tunnel lining inspection vehicle.

9. The method as described in claim 2, characterized in that, Tiered obstacle avoidance strategies include: If the obstacle type is a protruding nail, the graded obstacle avoidance strategy is to only retrieve the ground radar antenna containing the obstacle's location information; If the obstacle type is a contact wire frame, the graded obstacle avoidance strategy is to fold the third robotic arm. If the obstacle type is a tunnel with a variable cross-section, the graded obstacle avoidance strategy is to stop the vehicle and collaboratively retract the telescopic parts of the first, second, and fourth robotic arms.

10. The method as described in claim 9, characterized in that, Based on the simulation detection information, the corresponding hierarchical obstacle avoidance strategy for the multi-manipulator system is executed in the multibody dynamics digital twin model simulation, generating control commands, including: Based on the simulation information of the distance between the tunnel lining inspection vehicle and the obstacle, the speed simulation information of the tunnel lining inspection vehicle, and the interruption time requirements of different graded obstacle avoidance strategies, the control speed is determined. When the graded obstacle avoidance strategy is to only retract the ground radar antenna that provides obstacle location information, the control target is determined to be the inner and outer cylinder sliding pair of the second cylinder of the tunnel lining inspection vehicle. The control action is determined to retract the ground radar antenna according to the control speed and extend the ground radar antenna after the tunnel lining inspection vehicle has passed. When the graded obstacle avoidance strategy is to fold the third robotic arm, the control targets are determined to be the sliding joints of the second-level telescopic arm and the first-level telescopic arm, the sliding joints of the third-level telescopic arm and the second-level telescopic arm, the sliding joints of the fourth-level telescopic arm and the third-level telescopic arm, and the sliding joints of the fifth-level telescopic arm and the fourth-level telescopic arm. The control action is to retract the third robotic arm according to the control speed and extend the ground radar antenna after waiting for the tunnel lining inspection vehicle to pass. When the graded obstacle avoidance strategy involves stopping and collaboratively retracting the telescopic parts of the first, second, and fourth robotic arms, four control commands are determined. The first control command targets the tunnel lining inspection vehicle, and its action is to stop. The second control command targets the rotating joints of the upward and downward bending arms, the rotating joints of the downward bending arm and the fourth-stage telescopic arm, the sliding joints of the second-stage and first-stage telescopic arms, the sliding joints of the third-stage and second-stage telescopic arms, and the sliding joints of the fourth-stage and third-stage telescopic arms, and its action is to retract the folding robotic arms at the controlled speed and wait for completion. The third control command targets the tunnel lining inspection vehicle, and its action is to start and wait for the vehicle to completely pass. The fourth control command has the same target as the second control command, and its action is to restore the robotic arms.

11. A real-time graded obstacle avoidance control device for a tunnel lining inspection vehicle, characterized in that, include: The model building module is used to construct a multibody dynamics digital twin model of the tunnel lining inspection vehicle based on tunnel environment data and parameter data of the tunnel lining inspection vehicle. The multibody dynamics digital twin model includes a virtual tunnel environment model and a dynamic model of the tunnel lining inspection vehicle. The virtual tunnel environment includes various obstacle types, and the dynamic model of the tunnel lining inspection vehicle includes the dynamic models of each robotic arm. The data acquisition module is used to obtain real-time environmental data and vehicle status data collected by the detection sensors mounted on the tunnel lining inspection vehicle; The simulation module is used to drive the multibody dynamics digital twin model to perform synchronous simulation based on environmental data and vehicle status data, and generate simulation detection information. The control command generation module is used to generate control commands based on the simulation detection information, and to simulate and execute the corresponding hierarchical obstacle avoidance strategy for the multi-manipulator system in the multibody dynamics digital twin model. The control commands include control targets and control actions, and the hierarchical obstacle avoidance strategy includes non-stop obstacle avoidance strategy and stop obstacle avoidance strategy. The instruction issuing module is used to issue the control instructions according to the control objectives.

12. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 10.

14. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 10.

Images