Bionic wire-driven flexible obstacle avoidance response system and method for robot arm

Abstract

The application provides a kind of bionic wire-driven mechanical arm flexible obstacle avoidance response system and method, multi-modal perception module synchronously activates visual, acoustic, tactile component to collect environmental data;Dynamic impedance field construction module generates virtual mechanical impedance field through impedance field mapping function after data verification, and is converted into three-dimensional space distribution data to extract distribution abstract;Path and tension instruction module plans obstacle avoidance path, generates expected impedance field sequence, calculates tension parameter through wire drive tension pre-adjustment compensation formula and integrates into control instruction;Tension execution module adjusts driving wire tension to drive mechanical arm to move, and when collision, buffer absorbs energy and converts into escape kinetic energy;Closed-loop calibration module compares actual and expected impedance, and generates correction signal feedback to impedance field construction and instruction module if it is over threshold value.The application realizes multi-modal perception fusion, dynamic impedance field adaptation, tension and path coordination and whole-process closed-loop calibration, improves mechanical arm obstacle avoidance safety, adaptability and operation stability.

Description

Technical Field

[0001] This invention belongs to the field of robotic arm control technology, specifically a biomimetic line-driven robotic arm flexible obstacle avoidance response system and method. Background Technology

[0002] With the rapid development of industrial automation, precision manufacturing, and service robots, the demand for robotic arms to operate in complex environments is increasing. Especially in scenarios involving human-robot collaboration, fragile item handling, or dynamic obstacles, higher requirements are placed on the robotic arms' flexible obstacle avoidance capabilities and environmental adaptability. Wire-driven robotic arms, with their advantages of lightweight design, high flexibility, and large workspace, are gradually becoming important execution devices in these scenarios. The core requirement for such robotic arms is to perceive environmental information in real time, accurately plan motion paths, and achieve stable operation and safe obstacle avoidance by dynamically adjusting the tension of the drive line. Currently, robotic arm obstacle avoidance technology mostly relies on a single perception method to obtain environmental data and combine it with a preset path planning algorithm to execute actions. However, in dynamically changing and complex environments, how to integrate multi-dimensional perception information, build accurate environmental risk models, and achieve synergistic linkage between tension adjustment and path optimization has become a key direction for improving the operational performance of wire-driven robotic arms.

[0003] Traditional wire-driven robotic arm obstacle avoidance technology has the following shortcomings:

[0004] First, some technologies rely solely on visual or acoustic sensing to perceive the environment, making it difficult to fully capture the attributes, motion status, and contact information of obstacles. This can lead to misjudgment or omission of obstacles, potentially increasing the risk of collision.

[0005] Second, in the path planning stage, common methods generate paths based on fixed environment models, which have weak real-time adaptability to dynamic obstacles, and it is difficult to ensure both path smoothness and safety at the same time.

[0006] Third, in terms of tension adjustment, most systems use preset parameter control and do not fully combine environmental impedance changes and robotic arm motion state for dynamic compensation. This can easily lead to problems such as excessive tension causing damage to the drive line or insufficient tension causing a decrease in motion accuracy.

[0007] Fourth, existing systems rely heavily on emergency shutdown after a collision, lacking effective energy absorption and active escape mechanisms. This not only affects the continuity of operations but may also cause secondary damage to the robotic arm and the surrounding environment.

[0008] Fifth, existing technologies generally lack a closed-loop calibration mechanism for the entire process, making it difficult to optimize system parameters in real time based on actual operational deviations, resulting in insufficient long-term operational stability.

[0009] In summary, existing obstacle avoidance technologies for wire-driven robotic arms have shortcomings such as limited perception information, lack of dynamic adaptation in path planning, lack of dynamic compensation in tension adjustment, lack of active escape mechanism after collision, and lack of closed-loop calibration throughout the entire process. Summary of the Invention

[0010] To address the aforementioned technical problems, this invention provides a biomimetic line-driven robotic arm flexible obstacle avoidance response system and method, which realizes multimodal perception fusion, dynamic impedance field adaptation, tension and path coordination, and full-process closed-loop calibration, significantly improving the obstacle avoidance safety, adaptability, and operational stability of the robotic arm.

[0011] This invention includes a multimodal perception module that simultaneously activates visual, acoustic, and tactile sensing components to collect environmental data; a dynamic impedance field construction module that generates a virtual mechanical impedance field and extracts a distribution summary; a path and tension command module that plans the path, generates a desired impedance field sequence, and calculates tension parameters; a tension execution module that drives the robotic arm to move and buffers escape during collisions; and a closed-loop calibration module that compares the actual impedance with the expected impedance and corrects it if it exceeds a threshold. This invention comprehensively captures obstacle information through multimodal perception, dynamically constructs an impedance field to achieve precise adaptation, avoids risks and ensures smooth path planning, and dynamically optimizes parameters for tension adjustment. During tension execution, the servo motor and buffer work together to improve adaptability and safety, and the closed-loop calibration forms a dynamically optimized closed loop.

[0012] The present invention achieves the above-mentioned technical objectives through the following technical means.

[0013] A biomimetic line-driven robotic arm flexible obstacle avoidance response system includes a multimodal sensing module, a dynamic impedance field construction module, a path and tension command module, a tension execution module, and a closed-loop calibration module;

[0014] The multimodal perception module includes a vision component, an acoustic component, and a tactile component. The vision component is used to acquire environmental images, the acoustic component is used to acquire ultrasonic scanning data, and the tactile component is used to acquire contact force data. The dynamic impedance field construction module receives data from the multimodal perception module, verifies the data, generates a virtual mechanical impedance field covering the robotic arm's workspace through an impedance field mapping function, converts the virtual mechanical impedance field into spatial distribution data in three-dimensional coordinates, and extracts path impedance values ​​to generate a distribution summary. The path and tension command module receives the distribution summary from the dynamic impedance field construction module, plans an obstacle avoidance path, generates a desired impedance field sequence along the path, calculates the drive line tension pre-adjustment parameters through a line drive tension pre-adjustment compensation formula, and integrates the path, desired impedance field sequence, and tension pre-adjustment parameters into a control command. The tension execution module receives the control command from the path and tension command module, parses it, adjusts the drive line tension through a motor-driven tension controller, drives the robotic arm to move along the path, activates a buffer to absorb energy and convert it into escape kinetic energy during a collision, and collects drive line tension and robotic arm position data in real time. The closed-loop calibration module is used to receive feedback data from the tension execution module, calculate the actual impedance by combining the contact force data from the multimodal sensing module, compare the actual impedance with the expected impedance value to determine the deviation, and if the deviation exceeds the set threshold, generate a correction signal and feed it back to the dynamic impedance field construction module and the path and tension command module respectively.

[0015] In the above scheme, the multimodal perception module includes a vision component comprising an image acquisition unit and an image processing unit. The image acquisition unit is used to acquire environmental images within the workspace, and the image processing unit is used to process the acquired images to extract the shape, outline size, and spatial coordinate information of obstacles. The acoustic component includes an ultrasonic sensor array mounted on the head of the robotic arm, used to scan the space and determine the three-dimensional position of obstacles. The tactile component includes a flexible pressure-sensitive sensing unit deployed along the robotic arm, used to detect the contact pressure between the robotic arm and obstacles. It also includes a data integration unit, used to unify the format of the data output from the vision, acoustic, and tactile components and transmit it to the dynamic impedance field construction module. The tactile component has the highest priority; when tactile detection of contact occurs but there is no visual or acoustic feedback, collision buffering and escape are unconditionally triggered, and the area is marked as high impedance and fed back to the dynamic impedance field construction module.

[0016] In the above scheme, the specific steps of converting the virtual mechanical impedance field into spatial distribution data in three-dimensional coordinates in the dynamic impedance field construction module are as follows: taking the center of the robot arm base as the origin of the three-dimensional rectangular coordinate system, setting the X-axis along the horizontal extension direction of the robot arm, the Y-axis along the vertical lifting direction of the robot arm, and the Z-axis along the forward and backward extension direction of the robot arm; based on the divided cubic grid cells, recording the three-dimensional coordinates of the vertices and the three-dimensional coordinates of the center of each grid cell, and associating and binding the impedance value corresponding to each grid cell with its center three-dimensional coordinates; sorting all grid cells with associated impedance values ​​according to the coordinate order of the X-axis, Y-axis, and Z-axis, generating a dataset containing the three-dimensional coordinates of the center of the grid cell, the impedance value, and the grid size, forming the spatial distribution data of the impedance field in three-dimensional coordinates;

[0017] The update frequency of the virtual mechanical impedance field is a preset frequency, using trilinear interpolation. The physical meaning of the impedance value is equivalent contact stiffness, with units of N / mm. Impedance values ​​are assigned to each cubic grid cell through an impedance field mapping function. Specifically, for obstacles with different properties, low impedance values ​​are assigned around fragile obstacles, high impedance values ​​are assigned within the movement path coverage of high-speed moving obstacles, and medium impedance values ​​are assigned around obstacles with unidentified properties, with the detection vibration parameters superimposed.

[0018] The mathematical expression for the impedance field mapping function is:

[0019] ,

[0020] in, Spatial coordinates ,time The impedance value of the grid cell below, Based on the fundamental impedance coefficient, These are the weighting coefficients, and For obstacle attribute functions, Let be the distance decay function. For dynamic adjustment functions, For spatial coordinates, For time, For the speed of the obstacle, The acceleration of the obstacle.

[0021] In the above scheme, when the path and tension command module plans the obstacle avoidance path, it avoids areas where the impedance value exceeds the range of 5N / mm to 500N / mm, controls the radius of curvature of the path to be no less than 5mm, and sets path nodes at 2mm intervals; when generating the desired impedance field sequence, the change amplitude of the target impedance value of adjacent path nodes does not exceed 10N / mm, and the arrival time interval is controlled between 0.05s and 0.2s.

[0022] In the above scheme, the formula for the pre-adjustment compensation of the line drive tension is:

[0023]

[0024] in, Spatial coordinates ,time Pre-adjustment tension of the drive line below, Spatial coordinates ,time The impedance value of the grid cell below, The effective length of the drive line. The motion compensation coefficient, The transition function for the robotic arm motion is expressed as follows:

[0025]

[0026] It is the real-time moving speed of the end effector of the robotic arm. It is the total time of a single motion phase. It is the speed smoothing coefficient. The maximum value of the function is limited to 0.2; The current speed of the robotic arm. This is the error correction factor. The historical error function is calculated as follows:

[0027]

[0028] in, It is the tension of historical theory, calculated as follows:

[0029]

[0030] in, These are the actual tension values ​​from the first three tests under the same impedance. These are the target impedance values ​​corresponding to the first three tests. It is the effective length of the drive line; For spatial coordinates, For time.

[0031] In the above scheme, the tension execution module activates the buffer to absorb 30% to 50% of the collision energy during a collision. At the same time, it sets the escape direction according to the collision direction and controls the drive line to extend and retract at a phase of 0.1π to 0.3π and an amplitude of 5mm to 10mm, converting part of the absorbed collision energy into escape kinetic energy.

[0032] A control method for implementing the biomimetic line-driven robotic arm flexible obstacle avoidance response system includes the following steps: Multimodal perception step: synchronously activating visual, acoustic, and tactile sensing components to acquire environmental images, ultrasonic scanning data, and contact force data; Dynamic impedance field construction step: receiving data from the multimodal perception step, and after data verification, generating a virtual mechanical impedance field covering the robotic arm's workspace through an impedance field mapping function, converting the virtual mechanical impedance field into spatial distribution data in three-dimensional coordinates, and extracting path impedance values ​​to generate a distribution summary; Path and tension command step: receiving the distribution summary from the dynamic impedance field construction step, planning an obstacle avoidance path, generating a desired impedance field sequence along the path, calculating the drive line tension pre-adjustment parameters through a line-driven tension pre-adjustment compensation formula, and integrating the path, desired impedance field sequence, and tension pre-adjustment parameters into a control command; Tension execution step: receiving the control command from the path and tension command step, parsing it, adjusting the drive line tension through a motor-driven tension controller, driving the robotic arm to move along the path, activating a buffer to absorb energy and convert it into escape kinetic energy upon collision, and acquiring drive line tension and robotic arm position data in real time; Closed-loop calibration step: Receive feedback data from the tension execution step, combine it with the contact force data from the multimodal sensing step to calculate the actual impedance, compare the actual impedance with the expected impedance value to determine the deviation, and if the deviation exceeds the set threshold, generate a correction signal and feed it back to the dynamic impedance field construction step and the path and tension command step respectively.

[0033] In the above scheme, the data verification in the dynamic impedance field construction step includes:

[0034] The system checks whether each sensor returns valid frames in a timely manner to determine data integrity, whether the returned data is within a reasonable range to determine coverage, whether the maximum deviation of each sensor's timestamp does not exceed 10ms to determine timing consistency, and the validity of the data format.

[0035] In the above scheme, during the dynamic impedance field construction step, detection vibration parameters are superimposed on unidentified obstacles. These detection vibration parameters are: frequency 5Hz–20Hz, end amplitude 0.1mm–0.5mm, and single-cycle duration 0.5s–1.0s. A sinusoidal disturbance is then superimposed on the drive line based on the target tension during the tension execution step. Where ΔT(t) is the superimposed tension disturbance value, A T f is the disturbance amplitude. vib t represents the vibration frequency and t represents time; the touch component collects changes in contact force to infer the obstacle's properties and adjusts the impedance distribution, then cancels the subsequent vibration.

[0036] In the above scheme, during the closed-loop calibration step, a correction signal is generated when the relative deviation exceeds 5% or the absolute deviation exceeds 2N / mm. After the correction signal is fed back to the dynamic impedance field construction step, the weight coefficients α, β, γ or the distance attenuation coefficient k of the grid cells near the deviation point are locally adjusted in a gradient descent manner according to the direction and magnitude of the deviation between the actual impedance and the expected impedance at the deviation point. Here, α, β, γ are the weight coefficients of the obstacle attribute term, the distance attenuation term, and the dynamic adjustment term in the impedance field mapping function, respectively, and α+β+γ=1.

[0037] Compared with the prior art, the beneficial effects of the present invention are:

[0038] 1. This invention integrates visual, acoustic, and tactile sensor data synchronously through a multimodal perception module to comprehensively capture various information about obstacles in the environment. Combined with a dynamic impedance field construction module, it constructs a virtual mechanical impedance field covering the workspace, achieving precise adaptation to obstacles with different attributes, positions, and motion states. The path planning stage relies on impedance field distribution data to avoid high-risk areas and ensure a smooth path. Tension adjustment uses a scientific pre-adjustment compensation mechanism, combined with the robotic arm's motion state and historical data to dynamically optimize parameters, ensuring precise matching between drive line tension and path requirements. Simultaneously, the coordinated operation of the servo motor and hydraulic buffer in the tension execution module ensures smooth movement of the robotic arm and allows for rapid energy absorption and conversion into escape kinetic energy upon collision, thereby significantly improving the system's adaptability and operational safety in complex environments and effectively reducing losses caused by collisions.

[0039] 2. This invention continuously compares the actual impedance with the expected impedance through a closed-loop calibration module, promptly detecting deviations and generating correction signals, which are then fed back to the impedance field construction and tension command modules, forming a dynamic optimization closed loop throughout the entire process. This multi-module collaborative design not only achieves seamless integration of perception, planning, execution, and calibration, but also adjusts system parameters in real time according to environmental changes and operational feedback, ensuring the robotic arm's motion accuracy and operational flexibility in various scenarios. This invention does not rely on complex external auxiliary equipment; thanks to its multi-dimensional perception and adaptive adjustment capabilities, it significantly improves the timeliness and accuracy of obstacle avoidance response, expanding the application range of robotic arms in precision operations and dynamic environments, while also enhancing the stability and reliability of system operation. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of data transmission between modules of a biomimetic linearly driven robotic arm flexible obstacle avoidance response system.

[0041] Figure 2 This is a flowchart of a control method for a biomimetic linearly driven robotic arm's flexible obstacle avoidance response system. Detailed Implementation

[0042] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0043] like Figure 1 As shown, a biomimetic line-driven robotic arm flexible obstacle avoidance response system includes a multimodal sensing module, a dynamic impedance field construction module, a path and tension command module, a tension execution module, and a closed-loop calibration module.

[0044] The multimodal perception module includes a vision component, an acoustic component, and a tactile component. The vision component is used to acquire environmental images, the acoustic component is used to acquire ultrasonic scanning data, and the tactile component is used to acquire contact force data. The dynamic impedance field construction module receives data from the multimodal perception module, verifies the data integrity (whether each sensor module returns valid frames in a timely manner), data coverage (whether the data returned by each sensor is within a reasonable range), timing consistency (the maximum deviation of timestamps from each sensor does not exceed 10ms), and format validity. After passing the verification, a virtual mechanical impedance field covering the workspace of the robotic arm is generated through an impedance field mapping function. The virtual mechanical impedance field is converted into spatial distribution data in three-dimensional coordinates, and the path impedance values ​​are extracted to generate a distribution summary. The distribution summary includes key information such as impedance values, gradients, and extrema at each point on the path. The path and tension command module receives the distribution summary from the dynamic impedance field construction module, plans an obstacle avoidance path, generates a desired impedance field sequence along the path, calculates the drive line tension pre-adjustment parameters through the line drive tension pre-adjustment compensation formula, and integrates the path, desired impedance field sequence, and tension pre-adjustment parameters into a control command. The tension execution module receives control commands from the path and tension command module, parses them, and adjusts the drive line tension via a motor-driven tension controller to drive the robotic arm to move along the path. Upon collision, a buffer is activated to absorb energy and convert it into escape kinetic energy. The module also collects drive line tension and robotic arm position data in real time. The closed-loop calibration module receives feedback data from the tension execution module, calculates the actual impedance by combining it with contact force data from the multimodal sensing module, compares the actual impedance with the expected impedance value to determine the deviation, and generates a correction signal if the deviation exceeds a set threshold. This signal is then fed back to the dynamic impedance field construction module and the path and tension command module, respectively.

[0045] The multimodal perception module includes a vision component comprising an image acquisition unit and an image processing unit. The image acquisition unit acquires environmental images within the workspace, while the image processing unit processes the acquired images to extract the shape, outline size, and spatial coordinate information of obstacles. The acoustic component includes an ultrasonic sensor array mounted on the head of the robotic arm, used to scan the space and determine the three-dimensional position of obstacles. The tactile component includes a flexible pressure-sensitive sensing unit deployed along the robotic arm, used to detect the contact pressure between the robotic arm and obstacles. It also includes a data integration unit, used to unify the format of the data output from the vision, acoustic, and tactile components and transmit it to the dynamic impedance field construction module. The tactile component has the highest priority; when tactile detection of contact occurs without visual or acoustic feedback, collision buffering and escape are unconditionally triggered, and the affected area is marked as high impedance and fed back to the dynamic impedance field construction module.

[0046] In the dynamic impedance field construction module, when generating a virtual mechanical impedance field covering the workspace through the impedance field mapping function, the obstacle attributes, obstacle spatial coordinates, and obstacle motion state information transmitted by the multimodal sensing module are first extracted. Then, a three-dimensional rectangular coordinate system is established with the center of the robotic arm base as the origin, and the robotic arm workspace is divided into cubic grid units of a preset size. Based on the impedance field mapping function, a corresponding impedance value is assigned to each cubic grid unit. Specifically, for obstacles with different attributes, the grid unit impedance values ​​are assigned according to the following rules: low impedance values ​​are assigned to grid units within a set range around fragile obstacles; high impedance values ​​are assigned to grid units within the movement path coverage area of ​​high-speed moving obstacles; and medium impedance values ​​are assigned to grid units within a set range around obstacles with unidentified attributes, and a preset frequency detection vibration parameter is superimposed. The detection vibration parameter is: frequency 5Hz~20Hz, end amplitude 0.1mm~0.5mm, single duration 0.5s~1.0s, and a sinusoidal disturbance is superimposed on the drive line by the tension execution module based on the target tension. The vibration is achieved by the micro-extension and retraction of the motor, which causes the robotic arm to make small periodic contact with the unknown obstacle. The tactile component collects the changes in contact force (including amplitude, phase, and waveform distortion). The closed-loop calibration module infers the obstacle properties (rigid / flexible / irregular) based on this and adjusts the impedance distribution to cancel subsequent vibrations. After completing the impedance value distribution of all cubic grid units, a virtual mechanical impedance field covering the entire robotic arm workspace is formed.

[0047] The mathematical expression for the impedance field mapping function is:

[0048] ,

[0049] in, Spatial coordinates ,time The impedance value of the grid cell below, Here, α is the base impedance coefficient, β is the weighting coefficient for the obstacle attribute term, γ is the weighting coefficient for the distance attenuation term, and γ is the weighting coefficient for the dynamic adjustment term. , These are used to balance the contributions of obstacle attributes, distance attenuation, and dynamic adjustment to the impedance value. The allocation of these values ​​needs to be adjusted according to the specific obstacle type. If set improperly, the impedance field may be too sensitive to the obstacle type or ignore the distance factor, resulting in an unreasonable path or collision risk. For obstacle attribute functions, , It is an obstacle type identifier, determined by the multimodal perception module through image recognition and hardness detection. The distance decay function is calculated as follows: , It is the Euclidean distance between the obstacle and the mesh cell, calculated as follows: , These are the three-dimensional coordinates of the center of the current mesh cell. These are the three-dimensional coordinates of the obstacle's center. It is the distance attenuation coefficient, used to control how quickly the impedance value decreases with distance, and determines the severity of the robotic arm's response to impedance. For the dynamically adjusted function, the calculation expression is: , It is the real-time movement speed of the obstacle. It is the real-time acceleration of the obstacle. It is the speed coefficient. It is the acceleration coefficient, which can cause the impedance field to change dynamically with the motion of the obstacle, thus affecting the predictive obstacle avoidance capability for moving obstacles. For spatial coordinates, For time, For the speed of the obstacle, The acceleration of the obstacle.

[0050] The specific steps in the dynamic impedance field construction module to transform the virtual mechanical impedance field into spatial distribution data in three-dimensional coordinates are as follows: Using the center of the robotic arm base as the origin of the three-dimensional rectangular coordinate system, the X-axis is set along the horizontal extension direction of the robotic arm, the Y-axis along the vertical lifting direction of the robotic arm, and the Z-axis along the forward and backward extension direction of the robotic arm. Based on the pre-divided cubic mesh units, the three-dimensional coordinates of the vertices and the center of each mesh unit are recorded, and the impedance value corresponding to each mesh unit is associated and bound to its center three-dimensional coordinates. All mesh units associated with impedance values ​​are sorted according to the coordinate order of the X-axis, Y-axis, and Z-axis to generate a dataset containing the center three-dimensional coordinates of the mesh units, the impedance value, and the mesh size, forming spatial distribution data of the impedance field in three-dimensional coordinates. In this invention, the "impedance value" is defined as the contact force (in N) generated when a unit relative deformation (in mm) occurs between the robotic arm and an obstacle. Its physical meaning is equivalent contact stiffness, and the unit is uniformly N / mm. The larger the impedance value, the more the robotic arm needs to actively avoid or decelerate through the area.

[0051] The update frequency of the virtual mechanical impedance field is a preset frequency, using trilinear interpolation; impedance values ​​are assigned to each cubic grid cell through an impedance field mapping function, wherein, for obstacles with different properties, low impedance values ​​are assigned around fragile obstacles, high impedance values ​​are assigned within the movement path coverage of high-speed moving obstacles, and medium impedance values ​​are assigned around obstacles with unidentified properties, and the detected vibration parameters are superimposed.

[0052] When planning obstacle avoidance paths, the path and tension command module avoids areas with impedance values ​​exceeding the range of 5N / mm to 500N / mm, controls the radius of curvature of the path to be no less than 5mm, and sets path nodes at 2mm intervals. When generating the desired impedance field sequence, the change in the target impedance value of adjacent path nodes does not exceed 10N / mm, and the arrival time interval is controlled between 0.05s and 0.2s.

[0053] Specifically, in the path and tension command module, the specific steps for planning the obstacle avoidance path are as follows: taking the current position of the robotic arm as the starting point and the target position as the ending point, and combining the dynamic impedance field spatial distribution data, avoid areas where the impedance value exceeds the range of 5N / mm to 500N / mm, while controlling the radius of curvature between any two points on the path to be no less than 5mm to ensure that the path has no sharp turning points; after the path planning is completed, path nodes are set on the path at 2mm intervals, and the three-dimensional coordinates of each path node and the corresponding path sequence are recorded; when generating the desired impedance field sequence along the path, the three-dimensional coordinates corresponding to each path node are extracted, and the impedance target value of each path node is determined according to the dynamic impedance field spatial distribution data. At the same time, according to the distribution density of the path nodes and the moving speed of the robotic arm, the arrival time of each path node is set, and the interval of the arrival time is controlled between 0.05s and 0.2s; the three-dimensional coordinates, impedance target value and arrival time of each path node are arranged according to the path sequence to form the desired impedance field sequence. The change range of the impedance target value of adjacent path nodes in this sequence does not exceed 10N / mm to ensure a smooth transition of the impedance state.

[0054] In the path and tension command module, the formula for pre-adjustment compensation of the wire drive tension is:

[0055]

[0056] in, Spatial coordinates ,time Pre-adjustment tension of the drive line below, Spatial coordinates ,time The impedance value of the grid cell below, The effective length of the drive line. The motion compensation coefficient, The transition function for the robotic arm motion is expressed as follows:

[0057]

[0058] It is the real-time moving speed of the end effector of the robotic arm. It is the total time of a single motion phase. It is the speed smoothing coefficient. The maximum value of the function is limited to 0.2; The current speed of the robotic arm. This is the error correction factor. The historical error function is calculated as follows:

[0059]

[0060] in, It is the tension of historical theory, calculated as follows:

[0061]

[0062] in, These are the actual tension values ​​from the first three tests under the same impedance. These are the target impedance values ​​corresponding to the first three tests. It is the effective length of the drive line; For spatial coordinates, For time.

[0063] In the tension execution module, the specific steps for adjusting the tension of the drive line via a tension controller driven by a servo motor are as follows: The tension controller first parses the tension pre-adjustment parameters transmitted from the path and tension command module to obtain the target tension value, tension adjustment rate, and tension stability threshold of each drive line. The tension adjustment rate is controlled between 0.1 N / s and 1 N / s, and the tension stability threshold is set to ±1 N. The tension controller outputs a pulse width modulation signal to drive the servo motor. The servo motor drives the drive line to extend and retract via a transmission mechanism, real-time acquisition of the current tension value of the drive line, and feedback to the tension controller, forming a tension adjustment closed loop. When the difference between the current tension value and the target tension value of the drive line is reached... When the value is within the tension stability threshold range, the servo motor speed adjustment is stopped, completing the tension adjustment for this stage. When driving the robotic arm to move along the path, the movement of the robotic arm joints is controlled at a speed of 5mm / s to 15mm / s according to the path node information in the path command. Each joint is driven by 3 to 4 drive lines. The extension and retraction length of each drive line is calculated based on the three-dimensional coordinate difference of the path nodes, and the extension and retraction length error is controlled within ±0.1mm. During the movement of the robotic arm, the tension status of each drive line and the actual position of the robotic arm are continuously monitored to ensure that the robotic arm moves smoothly along the planned path and that the tension of the drive lines is always maintained within the stable range corresponding to the target tension value.

[0064] The tension execution module activates the buffer to absorb 30% to 50% of the collision energy upon collision. At the same time, it sets the escape direction according to the collision direction and controls the drive line to extend and retract at a phase of 0.1π to 0.3π and an amplitude of 5mm to 10mm, converting part of the absorbed collision energy into escape kinetic energy.

[0065] In the tension execution module, the specific steps for activating the micro hydraulic buffer to absorb energy and convert it into escape kinetic energy upon collision are as follows: When a collision occurs, the tactile component detects a collision signal with a contact pressure exceeding 10N and immediately triggers the micro hydraulic buffer to start. The buffer absorbs the collision energy through the damping effect of the internal hydraulic oil, with an energy absorption ratio of 30% to 50%. Simultaneously, the collision direction is determined by the pressure distribution information collected by the tactile component, and the escape direction is set based on the collision direction. A drive line retraction and extension command is generated, controlling the drive line to retract and extend at a phase of 0.1π to 0.3π and an amplitude of 5mm to 10mm, converting part of the absorbed collision energy into the kinetic energy for the robotic arm to move along the escape direction, thereby realizing the robotic arm's backward or lateral escape. The hydraulic buffer has a built-in displacement sensor that collects compression data in real time to calculate the specific value of the absorbed energy, ensuring that the energy conversion process is controllable.

[0066] like Figure 2 As shown, a control method for implementing the biomimetic line-driven robotic arm flexible obstacle avoidance response system includes the following steps: Multimodal perception step: Simultaneously activate visual, acoustic, and tactile sensing components to collect environmental images, ultrasonic scanning data, and contact force data; Dynamic impedance field construction step: Receive the data from the multimodal perception step, and after data verification, generate a virtual mechanical impedance field covering the robotic arm's workspace through an impedance field mapping function. Convert the virtual mechanical impedance field into spatial distribution data in three-dimensional coordinates and extract path impedance values ​​to generate a distribution summary; Path and tension command step: Receive the distribution summary from the dynamic impedance field construction step, plan the obstacle avoidance path, generate a desired impedance field sequence along the path, calculate the drive line tension pre-adjustment parameters through the line-driven tension pre-adjustment compensation formula, and integrate the path, desired impedance field sequence, and tension pre-adjustment parameters into a control command; Tension execution step: Receive the control command from the path and tension command step, parse it, adjust the drive line tension through a motor-driven tension controller, drive the robotic arm to move along the path, activate a buffer to absorb energy and convert it into escape kinetic energy upon collision, and collect drive line tension and robotic arm position data in real time; Closed-loop calibration step: Receive feedback data from the tension execution step, combine it with the contact force data from the multimodal sensing step to calculate the actual impedance, compare the actual impedance with the expected impedance value to determine the deviation, and if the deviation exceeds the set threshold, generate a correction signal and feed it back to the dynamic impedance field construction step and the path and tension command step respectively.

[0067] The data verification in the dynamic impedance field construction step includes:

[0068] The system checks whether each sensor returns valid frames in a timely manner to determine data integrity, whether the returned data is within a reasonable range to determine coverage, whether the maximum deviation of each sensor's timestamp does not exceed 10ms to determine timing consistency, and the validity of the data format.

[0069] In the dynamic impedance field construction step, detection vibration parameters are superimposed on unidentified obstacles. These detection vibration parameters are: frequency 5Hz–20Hz, end amplitude 0.1mm–0.5mm, and single-cycle duration 0.5s–1.0s. A sinusoidal perturbation is then superimposed on the drive line based on the target tension in the tension execution step. Where ΔT(t) is the superimposed tension disturbance value, A T f is the disturbance amplitude. vib t represents the vibration frequency and t represents time; the touch component collects changes in contact force to infer the obstacle's properties and adjusts the impedance distribution, then cancels the subsequent vibration.

[0070] In the closed-loop calibration step, a correction signal is generated when the relative deviation exceeds 5% or the absolute deviation exceeds 2N / mm. After the correction signal is fed back to the dynamic impedance field construction step, the weight coefficients α, β, γ or the distance attenuation coefficient k of the grid cells near the deviation point are locally adjusted in a gradient descent manner according to the direction and magnitude of the deviation between the actual impedance and the expected impedance at the deviation point. Here, α, β, γ are the weight coefficients of the obstacle attribute term, the distance attenuation term, and the dynamic adjustment term in the impedance field mapping function, respectively, and α+β+γ=1.

[0071] In the closed-loop calibration step, the specific steps for calculating the actual impedance are as follows: combining the real-time tension data of the drive line fed back by the tension execution module, the elastic coefficient of the drive line, and the deformation of the robotic arm joint, the actual impedance value is obtained by the ratio of the contact force to the total deformation; when comparing with the expected impedance value, the expected impedance value under the corresponding three-dimensional coordinates generated by the dynamic impedance field construction module is extracted, and the absolute deviation and relative deviation between the actual impedance value and the expected impedance value are calculated; the relative deviation threshold is set to 5%. When the relative deviation exceeds 5% or the absolute deviation exceeds 2N / mm, a correction signal containing the three-dimensional coordinates of the deviation point, the actual impedance value, the expected impedance value, and the deviation value is generated. The correction signal is fed back to the dynamic impedance field construction module to adjust the impedance distribution rules and to the path and tension command module to correct the tension pre-adjustment parameters.

[0072] This invention collects environmental information through a multimodal sensing module, and a dynamic impedance field construction module assigns impedance values ​​to grid cells through an impedance mapping function, thereby ultimately forming a virtual mechanical impedance field in the robotic arm's workspace. The virtual mechanical impedance field mainly includes a coordinate system with the center of the robotic arm base as the origin, grid cells of preset size divided by the workspace, the coordinates and impedance values ​​corresponding to each cell, and a time dimension. Its update frequency is a preset frequency, using trilinear interpolation. The path and tension command module plans the obstacle avoidance path and calculates the pre-adjustment parameters of the drive line tension. The tension execution module adjusts the drive line tension to drive the robotic arm to move, and activates the buffer to absorb energy and achieve escape upon collision. At the same time, the closed-loop calibration module corrects the impedance field and path commands based on feedback data to ensure that the robotic arm completes obstacle avoidance operations accurately and smoothly.

[0073] Example 1:

[0074] Precision electronic component assembly scene.

[0075] In a precision electronic component assembly workshop, robotic arms need to install microchips into designated positions on circuit boards. The workspace contains high-speed moving automated conveyor belts and may also contain unlabeled tools and other debris. It is necessary to avoid collisions that could damage the chips and to ensure smooth movement that does not affect assembly accuracy.

[0076] The implementation steps are as follows:

[0077] The multimodal perception module is activated. The vision component acquires environmental images of the assembly area using an industrial camera, extracting the shape, outline, size, and spatial coordinates of the microchips. Simultaneously, it identifies the position and trajectory of the high-speed conveyor belt, providing a basis for the appearance and location of obstacles for subsequent precise obstacle avoidance. The acoustic component scans the entire workspace using an ultrasonic sensor array on the robotic arm's head, accurately determining the three-dimensional position of the conveyor belt and the approximate location of debris, supplementing any blind spots in visual perception. The tactile component remains on standby via flexible pressure-sensitive sensing units deployed along the robotic arm, constantly detecting contact pressure and providing real-time feedback for collision detection. The data integration unit processes the data output from the vision and acoustic components to unify the format before transmitting it to the dynamic impedance field construction module, ensuring consistent data format for subsequent processing. It should be noted that the tactile sensor has the highest priority because contact represents a real physical event, followed by vision (which provides a large amount of information but may be subject to occlusion), and finally acoustic (which can supplement blind spots but has lower spatial resolution). If tactile detection reveals obvious contact but there is no visual / acoustic feedback, collision buffering and escape should be triggered unconditionally, the area should be marked as high impedance, and feedback should be sent to the dynamic impedance field construction module for correction. If there is no visual detection but there is acoustic feedback, the acoustic data should be accepted, the obstacle should be marked as "suspected", assigned a medium impedance value, and the detection vibration should be superimposed.

[0078] The dynamic impedance field construction module receives and verifies data, first extracting obstacle attributes, spatial coordinates, and motion state information. A three-dimensional Cartesian coordinate system is established with the center of the robotic arm base as the origin, dividing the workspace into cubic grid cells of preset sizes, allowing for more precise and controllable impedance allocation. Impedance values ​​are assigned to each grid cell according to rules: low impedance values ​​are assigned to grid cells within a set range around the microchip to prevent excessive force from damaging the chip when the robotic arm approaches; high impedance values ​​are assigned to grid cells within the coverage area of ​​the high-speed conveyor belt's movement path to warn the robotic arm to stay away from high-risk areas; and medium impedance values ​​are assigned to grid cells within a set range around unidentified debris, superimposed with preset frequency detection vibration parameters, balancing safety and the detection of unknown obstacles. The impedance field mapping function completes the impedance value allocation for all grid cells, forming a virtual mechanical impedance field covering the entire workspace. The mathematical expression of the impedance field mapping function is: ;in, Spatial coordinates ,time The impedance value of the grid cell below, Based on the fundamental impedance coefficient, The weighting coefficients and For obstacle attribute functions, Let be the distance decay function. For dynamic adjustment functions, For spatial coordinates, For time, For the speed of the obstacle, The impedance field is then converted into spatial distribution data in three-dimensional coordinates and sorted according to the X-axis, Y-axis, and Z-axis coordinates to generate a dataset containing the three-dimensional coordinates of the grid cell center, impedance value, and grid size, making the impedance information more intuitive. The path impedance value is extracted to generate a distribution summary, which facilitates the path and tension command module to quickly obtain key impedance data.

[0079] The path and tension command module receives the distribution summary, taking the current position of the robotic arm as the starting point and the chip mounting position on the circuit board as the ending point. Combined with the dynamic impedance field spatial distribution data, it avoids areas where the impedance value exceeds the range of 5N / mm to 500N / mm, ensuring path safety. Simultaneously, it controls the radius of curvature between any two points on the path to be no less than 5mm, ensuring the path has no sharp inflection points and guaranteeing smooth robotic arm movement. After path planning is completed, path nodes are set at 2mm intervals, and the three-dimensional coordinates and path sequence of each node are recorded, providing clear positional guidance for robotic arm movement. When generating the desired impedance field sequence along the path, the three-dimensional coordinates of each path node are extracted, and the impedance target value is determined based on the spatial distribution data, allowing the robotic arm to have corresponding impedance references at different nodes. Based on the path node distribution density and the robotic arm's moving speed, the arrival time interval of each node is controlled between 0.05s and 0.2s to rationally plan the movement rhythm. The three-dimensional coordinates, impedance target value, and arrival time of each node are arranged according to the path sequence, ensuring that the change in impedance target value between adjacent nodes does not exceed 10N / mm, avoiding impedance abrupt changes that could affect the stability of the robotic arm. The tension pre-adjustment parameters of each drive line are calculated using the drive line tension pre-adjustment compensation formula, providing an accurate basis for drive line tension adjustment. The mathematical expression of the drive line tension pre-adjustment compensation formula is as follows: ;in, Spatial coordinates ,time Pre-adjustment tension of the drive line below, Spatial coordinates ,time The impedance value of the grid cell below, The effective length of the drive line. The motion compensation coefficient, For the transition function of the robotic arm motion, The current speed of the robotic arm. This is the error correction factor. For historical error function, The actual tension under the same impedance for the first three tests. The target impedance for the first three tests. For spatial coordinates, For time; the path, desired impedance field sequence and tension pre-adjustment parameters are integrated into control commands to provide a complete control basis for the tension execution module.

[0080] The tension execution module receives and parses control commands. The tension controller acquires the target tension value, tension adjustment rate, and tension stability threshold for each drive line. The tension adjustment rate is controlled between 0.1 N / mm and 1 N / mm to prevent excessively rapid tension adjustment from causing robotic arm vibration. The tension stability threshold is set to ±1 N to ensure tension control accuracy. The tension controller outputs a pulse width modulation signal to drive the servo motor. The servo motor drives the drive lines to extend and retract via a transmission mechanism. The current tension value is collected in real time and fed back, forming a tension adjustment closed loop. When the tension difference is within the stability threshold range, the speed adjustment stops, ensuring that the drive line tension accurately reaches the target value. The robotic arm joint movement is controlled at a moving speed of 5 mm / s to 15 mm / s, balancing efficiency and stability. Each joint is driven by 3 to 4 drive lines in tandem, improving the stability and accuracy of joint movement. The extension and retraction lengths of each drive line are calculated based on the 3D coordinate difference of the path nodes, ensuring that the extension and retraction length error is controlled within ±0.1 mm, guaranteeing that the robotic arm moves along the planned path. During movement, if the haptic component detects a collision signal with a contact pressure exceeding 10N, it immediately activates a miniature hydraulic buffer. Through the damping effect of internal hydraulic oil, it absorbs 30% to 50% of the collision energy, reducing damage to the robotic arm and components. Simultaneously, based on the pressure distribution information collected by the haptic component, it determines the collision direction, sets the escape direction, and controls the drive line to extend and retract at a phase of 0.1π to 0.3π and an amplitude of 5mm to 10mm, converting part of the absorbed collision energy into escape kinetic energy. This allows the robotic arm to retreat or move laterally, quickly escaping the dangerous situation. The displacement sensor built into the hydraulic buffer collects compression data in real time, ensuring the energy conversion process is controllable. It also continuously monitors the drive line tension and robotic arm position data, providing real-time feedback for closed-loop calibration. Simulation tests show that in low-speed static obstacle collisions, approximately 15% of the absorbed energy is converted into escape kinetic energy; in medium-speed scenarios, approximately 25%; and in high-speed sudden collisions, approximately 30% to 35%. Obstacle avoidance success rate: 98% (49 / 50 successful avoidances of dynamic conveyor belt and unknown debris, with no chip damage). Average response time: 0.32s from the appearance of the obstacle to the completion of the first obstacle avoidance action by the robotic arm. Escape kinetic energy conversion rate (medium speed collision): 25.6% (consistent with the simulation value of 25%).

[0081] The closed-loop calibration module receives real-time tension data from the drive line and position data from the tension execution module. Combined with contact force data from the multimodal sensing module, it calculates the actual impedance value by the ratio of contact force to total deformation, accurately reflecting the impedance of the robot arm under actual working conditions. The module extracts the expected impedance value in the corresponding three-dimensional coordinates generated by the dynamic impedance field construction module, calculates the absolute and relative deviations between the actual and expected impedance values, and determines whether the robot arm deviates from the expected working state. When the relative deviation exceeds 5% or the absolute deviation exceeds 2N / mm, a correction signal is generated, containing the three-dimensional coordinates of the deviation point, the actual impedance value, the expected impedance value, and the deviation value. This signal is fed back to the dynamic impedance field construction module to adjust the impedance distribution rules, making the impedance field more closely match the actual situation; and to the path and tension command module to correct the tension pre-adjustment parameters, ensuring more precise drive line tension control and ultimately guaranteeing that the robot arm accurately reaches the target position to complete assembly.

[0082] In summary, in the context of precision electronic component assembly, this biomimetic wire-driven robotic arm's flexible obstacle avoidance response system comprehensively collects environmental and obstacle information through a multimodal perception module, laying the foundation for subsequent processing. The dynamic impedance field construction module rationally allocates impedance based on obstacle properties, forming a precise impedance field by combining it with an impedance field mapping function. The path and tension command module plans a safe and stable path, calculating parameters using a wire-driven tension pre-adjustment compensation formula. The tension execution module precisely controls tension and movement, activating a buffer to ensure safety upon collision. The closed-loop calibration module corrects deviations in real time. The collaborative efforts of all modules ensure the robotic arm accurately completes chip assembly, avoiding component damage and unstable movement.

[0083] Example 2:

[0084] Medical device packaging scenario.

[0085] In the medical device packaging workshop, the robotic arm needs to remove sterile syringes from the material tray and place them into the packaging box. There are intermittently moving material conveyor racks in the work space, and there are some unmarked packaging materials. It is necessary to avoid damaging the syringes and ensure that the picking and placing actions are smooth and meet the requirements of aseptic operation.

[0086] The implementation steps are as follows:

[0087] The multimodal perception module simultaneously activates all sensing components. The vision component uses an industrial camera to capture images of the tray, packaging box, and surrounding environment, extracting the shape, outline size, spatial coordinates of the sterile syringe, as well as the position and movement patterns of the material conveyor, providing precise visual basis for the robotic arm to pick up and place syringes and avoid the conveyor. The acoustic component uses an ultrasonic sensor array to scan the workspace, further accurately determining the three-dimensional position of the material conveyor and the distribution range of packaging materials, compensating for the insufficient perception of vision in complex environments. The tactile component's flexible pressure-sensitive sensing units deployed along the robotic arm are in standby mode, ready to detect contact pressure and promptly identify potential collisions. The data integration unit processes the data collected by the vision and acoustic components into a unified format before transmitting it to the dynamic impedance field construction module, ensuring that the data can be effectively utilized by subsequent modules.

[0088] The dynamic impedance field construction module receives and verifies data, extracts obstacle attributes, spatial coordinates, and motion state information, establishes a three-dimensional Cartesian coordinate system with the center of the robotic arm base as the origin, and divides the workspace into cubic grid cells of preset sizes, making impedance allocation more targeted. According to impedance allocation rules, low impedance values ​​are assigned to grid cells within a set range around the sterile syringe to prevent squeezing damage to the syringe during robotic arm operation; high impedance values ​​are assigned to grid cells within the movement path coverage of the material conveyor to guide the robotic arm to avoid the moving conveyor; and medium impedance values ​​are assigned to grid cells within a set range around the packaging materials, superimposed with preset frequency detection vibration parameters to detect unknown materials while ensuring safety. The impedance field mapping function completes the impedance value allocation for all grid cells, forming a virtual mechanical impedance field. The mathematical expression of the impedance field mapping function is: ;in, Spatial coordinates ,time The impedance value of the grid cell below, Based on the fundamental impedance coefficient, The weighting coefficients and For obstacle attribute functions, Let be the distance decay function. For dynamic adjustment functions, For spatial coordinates, For time, For the speed of the obstacle, The system provides comprehensive impedance references for obstacle acceleration and subsequent path planning. It then transforms these impedances into spatial distribution data in three-dimensional coordinates, sorts them according to the coordinate axes to generate a dataset containing the three-dimensional coordinates of the grid cell centers, impedance values, and grid sizes, making the impedance information clear and concise. The system extracts the path impedance values ​​to generate a distribution summary, which facilitates the path and tension command modules to quickly obtain key impedance information.

[0089] The path and tension instruction module receives the distribution summary and plans obstacle avoidance paths in two segments: the current position of the robotic arm as the starting point, the syringe pickup position in the tray as the first stage endpoint, and the placement position in the packaging box as the second stage endpoint. This makes the picking and placing actions more organized. During the planning process, areas with impedance values ​​exceeding 5N / mm to 500N / mm are strictly avoided to ensure path safety. The radius of curvature between any two points on the path is controlled to be no less than 5mm, with no sharp turning points, ensuring smooth robotic arm movement and meeting the stability requirements for aseptic operation. Path nodes are set at 2mm intervals on both paths, and the three-dimensional coordinates and path sequence of each node are recorded to provide clear guidance for each stage of robotic arm movement. When generating the desired impedance field sequence, the impedance target value corresponding to the three-dimensional coordinates of each node is extracted to ensure that the robotic arm has a suitable impedance standard at different movement stages. The arrival time interval is reasonably set between 0.05s and 0.2s to optimize the movement rhythm. The three-dimensional coordinates, impedance target value, and arrival time of each node are arranged in the path sequence to ensure that the change in impedance target value between adjacent nodes does not exceed 10N / mm, avoiding sudden impedance changes that could affect the stability of the robotic arm. The tension pre-adjustment parameters of each drive line are calculated using the wire drive tension pre-adjustment compensation formula. The mathematical expression of the wire drive tension pre-adjustment compensation formula is: ;in, Spatial coordinates ,time Pre-adjustment tension of the drive line below, Spatial coordinates ,time The impedance value of the grid cell below, The effective length of the drive line. The motion compensation coefficient, For the transition function of the robotic arm motion, The current speed of the robotic arm. This is the error correction factor. For historical error function, The actual tension under the same impedance for the first three tests. The target impedance for the first three tests. For spatial coordinates, It provides accurate data for time and drive line tension control; it integrates two paths, the corresponding desired impedance field sequence and tension pre-adjustment parameters to form a complete control command, providing comprehensive operational basis for the tension execution module.

[0090] The tension execution module receives and parses control commands. The tension controller determines the target tension value, tension adjustment rate (0.1N / mm to 1N / mm), and tension stability threshold (±1N) for each drive line. The tension adjustment rate setting ensures smooth tension adjustment and avoids impact; the stability threshold setting ensures tension control accuracy. A pulse width modulation signal is output to drive the servo motor, which in turn drives the drive lines to extend and retract, providing real-time feedback of tension data to form a closed-loop adjustment until the tension reaches a stable state, ensuring the drive line tension meets requirements. The robotic arm moves along the first path at a speed of 5mm / s to 15mm / s, balancing material handling efficiency and stability. Each joint is driven by 3 to 4 drive lines in coordination, enhancing the stability and accuracy of joint movement. Precise control of the drive line extension and retraction length, with an error not exceeding ±0.1mm, ensures the robotic arm accurately reaches the material tray position to pick up the syringe. It then moves along the second path towards the packaging box, continuously monitoring the tension status and position data during the movement to promptly understand the robotic arm's working status. In the event of a collision, once the tactile component detects a contact pressure exceeding 10N, it immediately activates a miniature hydraulic buffer to absorb 30% to 50% of the collision energy, mitigating damage to the syringe and robotic arm. Based on the pressure distribution, it determines the collision direction and sets the escape direction, controlling the drive line to extend and retract at a phase of 0.1π to 0.3π and an amplitude of 5mm to 10mm, converting some of the collision energy into escape kinetic energy for safe escape. The displacement sensor collects buffer compression data in real time to monitor the energy conversion process, ensuring that the escape process is controllable.

[0091] The closed-loop calibration module receives real-time tension and position data from the tension execution module. Combined with contact force data from the tactile components, it calculates the actual impedance value using the ratio of contact force to total deformation, accurately determining the robotic arm's actual impedance. This is compared with the expected impedance value in the corresponding three-dimensional coordinates generated by the dynamic impedance field construction module, calculating absolute and relative deviations to determine if the robotic arm is operating normally. When the relative deviation exceeds 5% or the absolute deviation exceeds 2 N / mm, a correction signal is generated, containing the three-dimensional coordinates of the deviation point, the actual impedance value, the expected impedance value, and the deviation value. This signal is fed back to the dynamic impedance field construction module and the path and tension command module, adjusting the impedance distribution rules and tension pre-adjustment parameters to better match the robotic arm's operating state. This ensures the robotic arm smoothly and accurately places the syringe into the packaging box, completing the packaging operation and meeting aseptic operation requirements.

[0092] Simulation tests in a medical device packaging scenario yielded the following escape kinetic energy conversion rates for different collision types: At low speeds (robotic arm end-effector speed ≤ 50 mm / s) colliding with sterile syringes or static excipients, the proportion of absorbed energy converted into escape kinetic energy was approximately 12%–18%; at medium speeds (50–100 mm / s) lateral contact with intermittently moving material conveyors resulted in a conversion rate of approximately 22%–28%; and at high speeds (≥ 100 mm / s) sudden collisions with moving conveyors, the conversion rate reached 30%–35%. This conversion effect maintains a consistent trend with the precision assembly scenario in Example 1. Obstacle avoidance success rate: 96% (48 / 50 successful avoidances of intermittently moving conveyors and excipients without syringe breakage). Average response time: 0.38 s (response was slightly slower than in Example 1 due to the more complex obstacle movement patterns in the scenario). Escape kinetic energy conversion rate (medium-speed collision): 24.3% (consistent with the simulated value of 22%–28%).

[0093] In summary, in the medical device packaging scenario, the system's multimodal sensing module accurately acquires information about syringes, delivery racks, and excipients. The dynamic impedance field construction module allocates impedance according to obstacle characteristics and generates an impedance field through an impedance field mapping function. The path and tension command module plans the path in stages and calculates parameters using a wire-driven tension pre-adjustment compensation formula. The tension execution module smoothly controls the robotic arm to pick up and place syringes, with buffers and escape mechanisms ensuring safety in case of collisions. The closed-loop calibration module promptly corrects deviations, and the coordination of all components ensures that the robotic arm meets aseptic operation requirements, smoothly and accurately completing syringe packaging and avoiding device damage.

[0094] It should be understood that although this specification is described according to various embodiments, not every embodiment contains only one independent technical solution. This way of describing the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

[0095] The detailed descriptions listed above are merely specific illustrations of feasible embodiments of the present invention and are not intended to limit the scope of protection of the present invention. All equivalent embodiments or modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention.

Claims

1. A biomimetic line-driven robotic arm flexible obstacle avoidance response system, characterized in that, It includes a multimodal sensing module, a dynamic impedance field construction module, a path and tension command module, a tension execution module, and a closed-loop calibration module; The multimodal perception module includes a vision component, an acoustic component, and a tactile component. The vision component is used to acquire environmental images, the acoustic component is used to acquire ultrasonic scanning data, and the tactile component is used to acquire contact force data. The dynamic impedance field construction module receives data from the multimodal perception module, verifies the data, generates a virtual mechanical impedance field covering the robotic arm's workspace through an impedance field mapping function, converts the virtual mechanical impedance field into spatial distribution data in three-dimensional coordinates, and extracts path impedance values ​​to generate a distribution summary. The path and tension command module receives the distribution summary from the dynamic impedance field construction module, plans an obstacle avoidance path, generates a desired impedance field sequence along the path, calculates the drive line tension pre-adjustment parameters through a line drive tension pre-adjustment compensation formula, and integrates the path, desired impedance field sequence, and tension pre-adjustment parameters into a control command. The tension execution module receives the control command from the path and tension command module, parses it, adjusts the drive line tension through a motor-driven tension controller, drives the robotic arm to move along the path, activates a buffer to absorb energy and convert it into escape kinetic energy during a collision, and collects drive line tension and robotic arm position data in real time. The closed-loop calibration module is used to receive feedback data from the tension execution module, calculate the actual impedance by combining the contact force data from the multimodal sensing module, compare the actual impedance with the expected impedance value to determine the deviation, and if the deviation exceeds the set threshold, generate a correction signal and feed it back to the dynamic impedance field construction module and the path and tension command module respectively.

2. The biomimetic line-driven robotic arm flexible obstacle avoidance response system according to claim 1, characterized in that, The multimodal perception module includes a vision component comprising an image acquisition unit and an image processing unit. The image acquisition unit acquires environmental images within the workspace, while the image processing unit processes the acquired images to extract the shape, outline size, and spatial coordinate information of obstacles. The acoustic component includes an ultrasonic sensor array mounted on the head of the robotic arm, used to scan the space and determine the three-dimensional position of obstacles. The tactile component includes a flexible pressure-sensitive sensing unit deployed along the robotic arm, used to detect the contact pressure between the robotic arm and obstacles. It also includes a data integration unit, used to unify the format of the data output from the vision, acoustic, and tactile components and transmit it to the dynamic impedance field construction module. The tactile component has the highest priority; when tactile detection of contact occurs without visual or acoustic feedback, collision buffering and escape are unconditionally triggered, and the affected area is marked as high impedance and fed back to the dynamic impedance field construction module.

3. The biomimetic line-driven robotic arm flexible obstacle avoidance response system according to claim 1, characterized in that, The specific steps in the dynamic impedance field construction module to transform the virtual mechanical impedance field into spatial distribution data in three-dimensional coordinates are as follows: taking the center of the robotic arm base as the origin of the three-dimensional rectangular coordinate system, setting the X-axis along the horizontal extension direction of the robotic arm, the Y-axis along the vertical lifting direction of the robotic arm, and the Z-axis along the forward and backward extension direction of the robotic arm; based on the divided cubic grid cells, recording the three-dimensional coordinates of the vertices and the three-dimensional coordinates of the center of each grid cell, and associating and binding the impedance value corresponding to each grid cell with its center three-dimensional coordinates; sorting all grid cells with associated impedance values ​​according to the coordinate order of the X-axis, Y-axis, and Z-axis, generating a dataset containing the three-dimensional coordinates of the center of the grid cell, the impedance value, and the grid size, forming the spatial distribution data of the impedance field in three-dimensional coordinates; The update frequency of the virtual mechanical impedance field is a preset frequency, using trilinear interpolation; impedance values ​​are assigned to each cubic grid cell through an impedance field mapping function, wherein, for obstacles with different properties, low impedance values ​​are assigned around fragile obstacles, high impedance values ​​are assigned within the movement path coverage of high-speed moving obstacles, and medium impedance values ​​are assigned around obstacles with unidentified properties, and the detected vibration parameters are superimposed. The mathematical expression for the impedance field mapping function is: ， in, Spatial coordinates ,time The impedance value of the grid cell below, Based on the fundamental impedance coefficient, These are the weighting coefficients, and For obstacle attribute functions, Let be the distance decay function. For dynamic adjustment functions, For spatial coordinates, For time, For the speed of the obstacle, The acceleration of the obstacle.

4. The biomimetic line-driven robotic arm flexible obstacle avoidance response system according to claim 1, characterized in that, When planning obstacle avoidance paths, the path and tension command module avoids areas with impedance values ​​exceeding the range of 5N / mm to 500N / mm, controls the radius of curvature of the path to be no less than 5mm, and sets path nodes at 2mm intervals. When generating the desired impedance field sequence, the change in the target impedance value of adjacent path nodes does not exceed 10N / mm, and the arrival time interval is controlled between 0.05s and 0.2s.

5. The biomimetic line-driven robotic arm flexible obstacle avoidance response system according to claim 1, characterized in that, The formula for pre-adjustment compensation of the linear drive tension is: in, Spatial coordinates ,time Pre-adjustment tension of the drive line below, Spatial coordinates ,time The impedance value of the grid cell below, The effective length of the drive line. The motion compensation coefficient, The transition function for the robotic arm motion is expressed as follows: It is the real-time moving speed of the end effector of the robotic arm. It is the total time of a single motion phase. It is the speed smoothing coefficient. The maximum value of the function is limited to 0.2; The current speed of the robotic arm. This is the error correction factor. The historical error function is calculated as follows: in, It is the tension of historical theory, calculated as follows: in, These are the actual tension values ​​from the first three tests under the same impedance. These are the target impedance values ​​corresponding to the first three tests. It is the effective length of the drive line; For spatial coordinates, For time.

6. The biomimetic line-driven robotic arm flexible obstacle avoidance response system according to claim 1, characterized in that, The tension execution module activates the buffer to absorb 30% to 50% of the collision energy upon collision. At the same time, it sets the escape direction according to the collision direction and controls the drive line to extend and retract at a phase of 0.1π to 0.3π and an amplitude of 5mm to 10mm, converting part of the absorbed collision energy into escape kinetic energy.

7. A control method for implementing the biomimetic line-driven robotic arm flexible obstacle avoidance response system as described in claims 1-6, characterized in that, Includes the following steps: Multimodal perception step: Simultaneously activate visual, acoustic, and tactile sensing components to acquire environmental images, ultrasonic scanning data, and contact force data; Dynamic impedance field construction step: Receive data from the multimodal perception step, verify the data, generate a virtual mechanical impedance field covering the robotic arm's workspace through an impedance field mapping function, convert the virtual mechanical impedance field into spatial distribution data in three-dimensional coordinates, and extract path impedance values ​​to generate a distribution summary; Path and tension command step: Receive the distribution summary from the dynamic impedance field construction step, plan an obstacle avoidance path, generate a desired impedance field sequence along the path, calculate the drive line tension pre-adjustment parameters through the line drive tension pre-adjustment compensation formula, and integrate the path, desired impedance field sequence, and tension pre-adjustment parameters into a control command; Tension execution step: Receive the control command from the path and tension command step, parse it, adjust the drive line tension through a motor-driven tension controller, drive the robotic arm to move along the path, activate the buffer to absorb energy and convert it into escape kinetic energy upon collision, and collect drive line tension and robotic arm position data in real time; Closed-loop calibration step: Receive feedback data from the tension execution step, combine it with the contact force data from the multimodal sensing step to calculate the actual impedance, compare the actual impedance with the expected impedance value to determine the deviation, and if the deviation exceeds the set threshold, generate a correction signal and feed it back to the dynamic impedance field construction step and the path and tension command step respectively.

8. The control method for the biomimetic line-driven robotic arm flexible obstacle avoidance response system according to claim 7, characterized in that, The data verification in the dynamic impedance field construction step includes: The system checks whether each sensor returns valid frames in a timely manner to determine data integrity, whether the returned data is within a reasonable range to determine coverage, whether the maximum deviation of each sensor's timestamp does not exceed 10ms to determine timing consistency, and the validity of the data format.

9. The control method for the biomimetic line-driven robotic arm flexible obstacle avoidance response system according to claim 7, characterized in that, In the dynamic impedance field construction step, detection vibration parameters are superimposed on unidentified obstacles. These detection vibration parameters are: frequency 5Hz–20Hz, end amplitude 0.1mm–0.5mm, and single-cycle duration 0.5s–1.0s. A sinusoidal perturbation is then superimposed on the drive line based on the target tension in the tension execution step. Where ΔT(t) is the superimposed tension disturbance value, A T f is the disturbance amplitude. vib Let t be the vibration frequency and t be the time. By collecting changes in contact force through tactile components, the properties of obstacles are inferred and the impedance distribution is adjusted to cancel subsequent vibrations.

10. The control method for the biomimetic line-driven robotic arm flexible obstacle avoidance response system according to claim 7, characterized in that, In the closed-loop calibration step, a correction signal is generated when the relative deviation exceeds 5% or the absolute deviation exceeds 2N / mm. After the correction signal is fed back to the dynamic impedance field construction step, the weight coefficients α, β, γ or the distance attenuation coefficient k of the grid cells near the deviation point are locally adjusted in a gradient descent manner according to the direction and magnitude of the deviation between the actual impedance and the expected impedance at the deviation point. Here, α, β, γ are the weight coefficients of the obstacle attribute term, the distance attenuation term, and the dynamic adjustment term in the impedance field mapping function, respectively, and α+β+γ=1.

Images