An adaptive gripping force control method and system for a mechanical hand of a toothbrush bristle planting machine

Abstract

The present application relates to the technical field of toothbrush automatic manufacturing, and particularly discloses a self-adaptive gripping force control method and system of a mechanical hand of a toothbrush bristle planting machine, which comprises: starting a gripping material quality identification process, identifying a material quality area where a gripping area is located, if it is a transition area, outputting a material quality classification probability graph of the gripping area, executing a gripping contact action and collecting a contact force response signal, online identifying a local contact stiffness, integrating a material quality attribute vector, dynamically adjusting a normal impedance parameter, generating a normal flexible contact flag, if the normal flexible contact flag is received, collecting a three-axis torque signal at a gripping point, identifying a center of mass position vector of a toothbrush handle and calculating an eccentricity parameter, judging whether a tangential force active adjustment preparation mechanism is triggered, marking a control state, if the control state is tangential force to be intervened, calculating an expected tangential force target, introducing a friction force reserve observer to judge whether a normal force collaborative increase process is triggered, and executing tangential force adjustment.

Description

Technical Field

[0001] This invention relates to the field of automated toothbrush manufacturing technology, specifically to an adaptive gripping force control method and system for a toothbrush bristle implantation machine robotic arm. Background Technology

[0002] In the automated production of toothbrushes, the bristle implantation process is a crucial step determining product quality and production efficiency. Toothbrush bristle implantation machines use robotic arms to grasp the toothbrush handle and precisely transfer it to the bristle implantation station, where the implantation head uses high-speed impact to insert the bristles into pre-set holes. As consumers' demands for toothbrush comfort and functionality continue to increase, toothbrush handles are now commonly made using a composite injection molding process combining soft rubber and hard plastic. The soft rubber area enhances the grip, while the hard area ensures structural strength. This trend towards composite materials presents new challenges to the robotic arm's grasping and control.

[0003] In existing technologies, robotic arm grasping control mainly employs a rigid grasping strategy based on position control. The robotic arm moves to the grasping point along a preset trajectory and maintains a locked position, relying on the deformation of the gripper structure to provide contact force. This method can maintain basic stability in scenarios involving a single hard material. However, when the grasping area involves a transition interface between soft rubber and hard material, the stiffness of the soft rubber area is significantly lower than that of the hard area. As a result, the robotic arm is prone to force-controlled oscillations at the moment of contact, and even high-frequency flutter due to sudden changes in stiffness, leading to deviations in the grasping posture or crushing of the soft rubber layer.

[0004] Furthermore, the bristle implantation process itself involves high-frequency impact. When the bristle implantation head falls at high speed to insert the bristles into the holes, it generates periodic pulse forces on the toothbrush handle within a very short time. Since the gripping point of the robotic arm and the center of mass of the toothbrush handle are usually not perfectly aligned, this impact force acting on the center of mass will generate an eccentric rotational torque. When the bristle implantation impact frequency is coupled with the inherent frequency of the robotic arm structure, the eccentric rotational torque may gradually amplify, eventually leading to gripping instability or toothbrush handle posture deflection, directly affecting the bristle implantation accuracy and yield.

[0005] To address the aforementioned problems, this invention proposes an adaptive gripping force control method and system for a toothbrush bristle implantation machine robotic arm. Summary of the Invention

[0006] The purpose of this invention is to provide an adaptive gripping force control method and system for a toothbrush bristle implantation machine robotic arm to solve the aforementioned background problems.

[0007] The objective of this invention can be achieved through the following technical solution: an adaptive gripping force control method for a toothbrush bristle implantation machine robotic arm, comprising:

[0008] Initiate the material identification process, identify the material region where the grasping area is located, and if it is a transitional region, output the material classification probability map of the grasping area.

[0009] Perform the grasping contact action and collect the contact force response signal, identify the local contact stiffness online, integrate the material attribute vector by combining the material region and the material classification probability map, dynamically adjust the normal impedance parameter, and generate a normal flexible contact mark.

[0010] If a normal flexible contact marker is received, the triaxial torque signal at the gripping point is collected, the centroid position vector of the toothbrush handle is identified and the eccentricity parameter is calculated, the eccentric rotational torque characterization is extracted, it is determined whether the tangential force active adjustment preparation mechanism is triggered, and the control state is marked.

[0011] If the control state is tangential force pending intervention, the tangential force active adjustment mechanism is activated, the desired tangential force target is calculated, a friction reserve observer is introduced to determine whether the normal force collaborative increase process is triggered, and tangential force adjustment is executed;

[0012] Furthermore, the material identification process employs a hierarchical identification strategy:

[0013] The minimum Euclidean distance between the preset initial grasping point coordinates and the boundary contours of the extracted soft rubber area and hard area is calculated and defined as the boundary distance parameter. If the boundary distance parameter exceeds the preset boundary neighborhood radius, the grasping area is determined to be located in a single material area; otherwise, the grasping area is determined to be located in a transition area, triggering the hyperspectral imaging mode and outputting the material classification probability map of the grasping area.

[0014] Furthermore, the online identification of local contact stiffness specifically involves:

[0015] Define a contact time window. Within the contact time window, extract the force response signal in the grasping contact direction and mark it as the normal contact force signal. The position change of the end effector in the grasping contact direction is obtained by kinematic forward kinematics calculation through real-time feedback of joint encoder data from the robot control system and marked as the normal displacement signal of the robot. Based on the normal contact force signal and the normal displacement signal of the robot, construct a force-displacement curve. Calculate the slope of the force-displacement curve on the sliding sub-window using the least squares fitting method. Take the median value of all calculated slopes and record it as the local contact stiffness.

[0016] Furthermore, the dynamic adjustment of the normal impedance parameter is specifically as follows:

[0017] When the gripping area is located in a single material area, the normal impedance parameter adopts the preset standard impedance parameter. When the gripping area is located in a transition area, the preset stiffness-impedance parameter mapping curve is queried based on the identified local contact stiffness to obtain the target impedance stiffness and target damping coefficient. The design principle of the stiffness-impedance parameter mapping curve is that the lower the local contact stiffness, the lower the target impedance stiffness and the higher the target damping coefficient.

[0018] Furthermore, the method for obtaining the centroid position vector is as follows:

[0019] After receiving the normal flexible contact mark, the robot arm end effector is controlled to perform a set of preset micro-attitude perturbation actions. During the micro-attitude perturbation actions, the triaxial torque signal and angular acceleration signal at the gripping point are collected, and the least squares identification algorithm is used for fitting to obtain the centroid position vector of the toothbrush handle in the tool coordinate system online.

[0020] Furthermore, the extraction method for the eccentric rotational torque characterization is as follows:

[0021] During the hair implantation cycle, the resultant torque vector and resultant torque magnitude are calculated based on the triaxial torque signal. The sliding time window filtering technique is used to extract the maximum value of the resultant torque magnitude within each sliding time window and record it as the instantaneous torque peak value. After smoothing, the torque saturation index is obtained, and the first derivative of the torque saturation index with respect to time is calculated to obtain the torque saturation change rate index. The torque saturation index and the torque saturation change rate index are integrated and recorded as the eccentric rotation torque characterization.

[0022] Furthermore, the method for determining whether the normal force coordination increase process is triggered is as follows:

[0023] Obtain the torque saturation index included in the eccentric rotation torque characterization, combine it with the eccentric distance in the eccentric parameters, calculate the theoretical tangential force amplitude required at present through the torque balance equation as the expected tangential force target, and process the expected tangential force target and the eccentric distance to obtain the required torque;

[0024] Based on the real-time normal contact force signal and the estimated static friction coefficient of the material at the current gripping point, the maximum static friction torque is calculated and defined as the friction reserve. The estimated static friction coefficient is obtained by querying the preset friction coefficient library based on the material property vector. If the ratio of the required torque to the friction reserve exceeds the preset ratio threshold, the normal force collaborative increase process is triggered.

[0025] Furthermore, the specific process for increasing the synergy of normal forces is as follows:

[0026] Under the constraint of safety boundary, the normal force is gradually increased in a preset step size. The safety boundary constraint includes that the increment of normal force does not exceed the product of the preset maximum allowable compressive stress of the soft rubber layer and the contact area, and the increment of normal force does not exceed the product of the initial normal force and the preset safety factor. The contact area is calculated based on the surface features of the grasping area reconstructed by the 3D vision sensor. The initial normal force is the stable value of the normal contact force signal after the normal impedance is adjusted and stabilized.

[0027] After each increment of normal force, the friction reserve is recalculated until the ratio of the required torque to the friction reserve does not exceed the preset proportional threshold, at which point the increase of normal force is stopped.

[0028] Furthermore, the tangential force adjustment adopts a shockless switching algorithm, which switches the control mode of the two degrees of freedom coupled with the eccentric rotational torque direction in the grasping plane from position control mode to force control mode. The tangential force controller adopts a PID algorithm, takes the desired tangential force target as the control target, and outputs control commands to drive the end effector of the robot to apply tangential force in the grasping plane.

[0029] An adaptive gripping force control method for a robotic arm of a toothbrush bristle implantation machine includes the following modules:

[0030] Material recognition module: Initiates the material recognition process, identifies the material region where the grasped area is located, and outputs the material classification probability map of the grasped area if it is a transitional region.

[0031] Normal control module: Executes grasping contact action and collects contact force response signal, identifies local contact stiffness online, integrates material attribute vector with material region and material classification probability map, dynamically adjusts normal impedance parameter, and generates normal flexible contact mark;

[0032] Eccentricity assessment module: If a normal flexible contact marker is received, the triaxial torque signal at the gripping point is collected, the centroid position vector of the toothbrush handle is identified and the eccentricity parameter is calculated, the eccentric rotation torque characterization is extracted, it is determined whether the tangential force active adjustment preparation mechanism is triggered, and the control status is marked.

[0033] Tangential control and coordination module: If the control state is tangential force pending intervention, activate the tangential force active adjustment mechanism, calculate the desired tangential force target, introduce a friction reserve observer to determine whether to trigger the normal force coordination increase process, and execute tangential force adjustment.

[0034] The beneficial effects of this invention are as follows:

[0035] 1. This invention initiates the material identification process and outputs a material classification probability map. Combined with online identification of local contact stiffness and dynamic adjustment of normal impedance parameters, it enables the robotic arm to form flexible contact characteristics in the transition area between soft and hard materials. This effectively absorbs the energy impact at the moment of grasping contact, avoids the excitation of high-frequency force-controlled oscillation, solves the problem of unstable grasping caused by material abrupt changes in the prior art, and realizes a smooth transition and flexible contact in the soft-hard interface area.

[0036] 2. This invention identifies the centroid position vector of the toothbrush handle and extracts the eccentric rotational torque characterization, triggering an active tangential force adjustment mechanism. It introduces a friction reserve observer to achieve on-demand coordinated control of normal and tangential forces. Without compromising the flexible contact characteristics, it actively resists the eccentric rotational torque generated by the bristle implantation impact, solving the problem that passive resistance to eccentric disturbances in existing technologies easily leads to gripping instability, and improving the gripping accuracy and system stability under high-speed bristle implantation conditions. Attached Figure Description

[0037] The invention will now be further described with reference to the accompanying drawings.

[0038] Figure 1 This is a flowchart of the steps of an adaptive gripping force control method for a toothbrush bristle implantation machine robot arm according to an embodiment of the present invention;

[0039] Figure 2 This is a module architecture diagram of an adaptive gripping force control system for a toothbrush bristle implantation machine robotic arm, as described in an embodiment of the present invention. Detailed Implementation

[0040] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0041] Example 1

[0042] Please see Figure 1 As shown in the embodiment of the present invention, an adaptive gripping force control method for a toothbrush bristle implantation machine robotic arm aims to solve the problems of high-frequency force-controlled oscillation and difficulty in absorbing contact impact caused by sudden changes in material when gripping in the transition area between soft and hard materials. It also addresses the problems of gripping posture instability and the inability to actively compensate for tangential force caused by the eccentric rotational torque generated by the bristle implantation impact. The method achieves a stable and flexible contact in the soft-hard interface area and a dynamic balance against the eccentric rotational torque, improving gripping stability and bristle implantation accuracy. Specifically, it includes the following steps:

[0043] S1: Start the material recognition process and identify the material region where the grasping area is located. If it is a transitional region, output the material classification probability map of the grasping area.

[0044] Specifically, the material recognition process is initiated before the robotic arm performs the grasping action. The 3D vision sensor integrated into the fixed grasping station performs a rapid three-dimensional contour scan of the toothbrush handle in the acquisition area. The 3D vision sensor adopts the principle of binocular stereo vision. Under the preset exposure parameters and imaging distance, it acquires the depth image and grayscale image of the acquisition area. Based on the grayscale image and depth image, the edge detection algorithm is used to extract the boundary contour between the soft rubber area and the hard area of ​​the toothbrush handle.

[0045] It should be noted that grayscale images record the intensity of reflected light on the surface of the toothbrush handle, reflecting texture, color, and brightness variations, while depth images record the distance information from the toothbrush handle to the 3D vision sensor, reflecting the geometric shape and positional relationship in three-dimensional space. The soft rubber area exhibits different reflective characteristics in the grayscale image compared to the hard area, and may have slight geometric transition features in the depth image.

[0046] The initial gripping point coordinates are preset on the fixed gripping station. The actual contact area of ​​the robot arm gripping the toothbrush handle is defined as the gripping area. The center of action of the resultant force of all contact forces of the robot arm is defined as the gripping point. The initial coordinates of the gripping point are the preset initial gripping point coordinates. In order to balance the recognition accuracy and the production line cycle time, the gripping material recognition process adopts a hierarchical recognition strategy.

[0047] The minimum Euclidean distance between the initial gripping point coordinates and the boundary contour is calculated and defined as the boundary distance parameter. If the boundary distance parameter exceeds the preset boundary neighborhood radius, it is determined that the gripping area of ​​the robot arm is located in a single material area on the toothbrush handle. Based on the recognition results of the 3D vision sensor, it is determined whether the single material area is a soft rubber area or a hard area.

[0048] If the boundary distance parameter does not exceed the boundary neighborhood radius, it is determined that the gripping area of ​​the robotic arm is located in the transition area on the toothbrush handle, triggering the hyperspectral imaging mode for fine material confirmation. The hyperspectral data cube of the gripping area is collected within the preset wavelength range. By comparing it with the pre-built spectral feature library of soft rubber and hard materials, the spectral angle matching algorithm is used to output the material classification probability map of the gripping area pixel by pixel. The value of each pixel point represents the confidence of the corresponding position being judged as soft rubber or hard material in the form of a binary probability distribution, realizing the accurate identification of the distribution of soft and hard materials.

[0049] S2: Execute the grasping contact action and collect the contact force response signal, identify the local contact stiffness online, integrate the material attribute vector by combining the material region and the material classification probability map, dynamically adjust the normal impedance parameter, and generate a normal flexible contact mark.

[0050] The robot arm is controlled to perform a grasping contact action based on the initial grasping point coordinates. After the grasping contact action is completed, a contact time window is preset. Within the contact time window, the contact force response signal is collected in real time at a preset sampling frequency by a six-dimensional force sensor installed on the wrist of the robot arm.

[0051] It should be noted that the duration of the contact time window is calibrated offline based on the robot's feed speed and the desired contact stabilization time to ensure that a complete contact force response signal is collected during the contact transient process of the grasping contact action.

[0052] Local contact stiffness is identified online using the collected contact force response signals. Specifically, within the contact time window, the force response signal in the grasping contact direction is extracted and marked as the normal contact force signal. The position change of the end effector in the grasping contact direction is calculated by kinematic forward kinematics using the joint encoder data fed back in real time by the robot control system and marked as the normal displacement signal of the robot. A force-displacement curve is constructed based on the normal contact force signal and the normal displacement signal of the robot. For the contact time window, the slope of the force-displacement curve on the sliding sub-window is calculated using the least squares fitting method, and the median value of all calculated slopes is recorded as the local contact stiffness.

[0053] The material region where the grasping area is located, the material classification probability map of the grasping area, and the local contact stiffness identified online are fused to form the material attribute vector of the grasping area. The normal impedance parameter is dynamically adjusted according to the material attribute vector. The normal impedance parameter is characterized by both impedance stiffness and damping coefficient.

[0054] It should be noted that the normal impedance parameter represents the control parameter of the robot end effector in the grasping contact direction, which describes the dynamic relationship between contact force and displacement.

[0055] Specifically, when the gripping area is located in a single material area, the normal impedance parameter adopts the preset standard impedance parameter to maintain the normal contact's normal rigidity characteristics. When the gripping area is located in a transition area, based on the identified local contact stiffness, the preset stiffness-impedance parameter mapping curve is queried. The stiffness-impedance parameter mapping curve is an offline calibrated nonlinear mapping function. Its input is the local contact stiffness, and its output is the target impedance stiffness and target damping coefficient of the impedance controller. The normal impedance parameter adopts the output target impedance stiffness and target damping coefficient.

[0056] It should be noted that the design principle of the stiffness-impedance parameter mapping curve is that the lower the local contact stiffness, the lower the target impedance stiffness and the higher the target damping coefficient, so that the end effector of the robot arm exhibits stronger flexibility in the grasping contact direction.

[0057] The adjusted normal impedance parameters are injected into the normal impedance controller of the robot in real time to make basic adjustments to the normal impedance parameters, so that the end effector of the robot forms a flexible contact characteristic in the grasping contact direction. During the basic adjustment of the normal impedance parameters, the response signal is continuously monitored and high-frequency components are extracted. If the high-frequency components are extracted and match the preset high-frequency oscillation characteristics, the impedance stiffness is further reduced and the damping coefficient is increased according to the preset gradient until the oscillation is eliminated.

[0058] After the basic adjustment process of the normal impedance parameter is completed, the fluctuation range of the normal contact force signal is calculated within a continuous preset time. If it is lower than the preset force fluctuation threshold and the speed of each joint of the robot is lower than the preset speed threshold, the grasping posture is determined to be initially stable and a normal flexible contact mark is generated.

[0059] It should be noted that the flexible contact characteristics can effectively absorb the energy impact generated at the moment of contact due to the sudden change in soft and hard materials in the grasping area, avoid the excitation of high-frequency force-controlled oscillation, and thus achieve a smooth transition and flexible contact between the soft and hard interface areas.

[0060] S3: If a normal flexible contact marker is received, collect the triaxial torque signal at the gripping point, identify the centroid position vector of the toothbrush handle and calculate the eccentricity parameter, extract the eccentric rotation torque characterization, determine whether the tangential force active adjustment preparation mechanism is triggered, and mark the control state.

[0061] Specifically, after receiving the normal flexible contact mark, a preset identification time window is established. Within the identification time window, the end effector of the robot arm is controlled to perform a set of preset micro-attitude perturbation actions. The micro-attitude perturbation actions include small angular acceleration motions around each coordinate axis. During the micro-attitude perturbation actions, a six-dimensional torque sensor installed on the robot wrist collects and records the three-axis torque signal at the gripping point at a preset high sampling rate. An attitude angle sensor installed on the end effector of the robot arm synchronously collects the angular velocity signal and performs numerical difference calculation to obtain the angular acceleration. The three-axis torque signal represents the torque components around the X-axis, Y-axis and Z-axis in the tool coordinate system. The tool coordinate system is defined as a three-dimensional rectangular coordinate system established with the gripping point as the origin.

[0062] Based on the principles of rigid body dynamics, the center of mass is identified. When the toothbrush handle moves with a small angular acceleration around any axis, there is a definite mathematical relationship between the inertial torque of the toothbrush handle and the position of the center of mass. The least squares identification algorithm is used to organize the three-axis torque signals recorded during the micro-attitude perturbation into a torque change sequence, and then fit it with the angular acceleration sequence obtained by integrating the real-time recorded angular acceleration. The position vector of the center of mass of the toothbrush handle in the tool coordinate system at the end of the robot arm is calculated online.

[0063] Based on the centroid position vector and the coordinates of the current gripping point in the tool coordinate system, the eccentricity vector of the gripping point relative to the centroid of the toothbrush handle is calculated. The eccentricity vector includes the eccentricity distance and the eccentricity direction angle, collectively referred to as the eccentricity parameter. The identified eccentricity parameter is stored in the local cache. The bristle implantation cycle is defined as the time interval required for the toothbrush bristle implantation machine to complete one bristle implantation action. Before the start of each bristle implantation cycle, a verification is performed. The theoretical triaxial torque signal is obtained by extrapolation and prediction based on the centroid position vector and the current eccentricity parameter. The deviation between the current triaxial torque signal and the theoretical triaxial torque signal is calculated. If the deviation exceeds the preset deviation range, centroid re-identification is triggered, and the centroid position vector is updated to cope with the impact of toothbrush handle model switching or gripping posture changes. Otherwise, the bristle implantation cycle is entered.

[0064] It should be noted that the hair implantation action is not performed when identifying the centroid position vector; that is, neither centroid identification nor centroid re-identification is within the hair implantation cycle.

[0065] During the hair implantation cycle, the resultant torque vector and resultant torque magnitude are calculated based on the triaxial torque signal as comprehensive observation indicators. A sliding time window filtering technique is adopted, defining a sliding time window of fixed duration. The window length of the sliding time window is preset to be an integer multiple of multiple hair implantation cycles to ensure coverage of the complete hair implantation impact process. The end point of each sliding time window is used as the sampling time. At each sampling time, the maximum value of the resultant torque magnitude within the current sliding time window is extracted and defined as the instantaneous torque peak value. The instantaneous torque peak value sequence is obtained by sorting according to the time sequence. The instantaneous torque peak value sequence is further smoothed by a sliding average filter to suppress the periodic noise interference generated by the hair implantation impact. Stable feature values ​​reflecting the cumulative trend of eccentric rotation torque are extracted and recorded as torque saturation index. The first derivative of the torque saturation index with respect to time is calculated in real time to obtain the torque saturation change rate index. The torque saturation index and the torque saturation change rate index are integrated and recorded as the eccentric rotation torque characterization.

[0066] It should be noted that the bristle implantation impact process refers to the dynamic process in which the bristle implantation head, falling at high speed, implants bristles into the toothbrush handle hole, generating periodic pulse force on the toothbrush handle in a very short time. The length of the sliding time window and the filtering order are calibrated offline according to the actual impact frequency of the bristle implantation machine. The typical configuration is to cover the most recent 3 to 5 complete bristle implantation cycles in order to suppress periodic impact noise and retain trend change information. The torque saturation change rate index reflects the cumulative speed and change trend of the eccentric rotation torque.

[0067] The torque saturation index and torque saturation change rate index are compared with preset saturation threshold and change rate threshold, respectively. The saturation threshold is calibrated offline based on the preset maximum allowable anti-deflection capability of the robot end effector under the current normal impedance parameter. The change rate threshold is determined by the preset typical upward slope and safety margin of the bristle impact. When the torque saturation index exceeds the preset saturation threshold, or the torque saturation change rate index exceeds the preset change rate threshold, it is determined that the adjustment based on the normal impedance parameter is no longer able to resist the eccentric rotation torque. Continuing to maintain the current state may lead to gripping instability or toothbrush handle posture deviation, triggering the tangential force active adjustment preparation mechanism.

[0068] It should be noted that eccentric rotational torque refers to the rotational torque around the gripping point generated when the gripping point of the robotic arm does not coincide with the center of mass of the toothbrush handle, caused by the bristle impact force or other external forces acting on the center of mass.

[0069] If not triggered, the control status is marked as tangential force not triggered;

[0070] If triggered, the computing resources of the tangential control and coordination module are activated, the eccentricity parameters obtained from the centroid identification and the current torque saturation index are preloaded, the initial target value of the tangential force controller is preset to the initial theoretical tangential force calculated based on the resultant torque vector of the current grasping point and the contact geometry, and the control state is marked as tangential force to be intervened, so as to ensure that the tangential force can intervene in the control with minimal delay in subsequent steps, and achieve a smooth transition from normal flexible impedance to normal tangential coordinated control;

[0071] S4: If the control state is tangential force waiting to intervene, activate the tangential force active adjustment mechanism, calculate the expected tangential force target, introduce the friction force reserve observer to determine whether the normal force cooperative increase process is triggered, and execute the tangential force adjustment;

[0072] It should be noted that the activation prerequisite for the active adjustment mechanism of tangential force is that the system has completed the material identification process, the basic adjustment of normal impedance has been completed and a normal flexible contact mark has been generated, and the eccentricity parameter of the current gripping point has been successfully obtained through centroid identification. If any of the prerequisites are not met, the intervention of tangential force will be temporarily suspended, the current normal impedance control state will be maintained, and an abnormal status signal will be fed back to the central controller.

[0073] After activating the active adjustment mechanism of tangential force, the torque saturation index included in the eccentric rotation torque characterization is obtained. Combined with the eccentric distance in the eccentric parameters, the theoretical tangential force amplitude required at the moment balance equation is calculated as the expected tangential force target. According to the material attribute vector, the preset friction coefficient library is queried. The friction coefficient library is an offline calibrated mapping table. The input is the material attribute vector, and the output is the corresponding static friction coefficient value range. The median value of the static friction coefficient value range is taken as the estimated static friction coefficient value corresponding to the material of the gripping point.

[0074] A friction reserve observer is introduced to calculate the maximum static friction torque under the current contact state in real time with a preset sampling period. Specifically, the friction reserve observer calculates the maximum available static friction force based on the real-time normal contact force signal and the estimated static friction coefficient corresponding to the material of the current gripping point. The maximum static friction force is multiplied by the eccentricity distance in the eccentricity parameter to obtain the maximum static friction torque, which is defined as the friction reserve. The desired tangential force target is multiplied by the eccentricity distance to calculate the torque corresponding to the current required tangential force, which is defined as the demand torque. If the ratio of the demand torque to the friction reserve does not exceed the preset ratio threshold, tangential force adjustment is performed. Otherwise, it is determined that the static friction reserve provided by the current normal force is insufficient to support the transmission of the required tangential force, and there is a risk of slippage and instability. The normal force collaborative increase process is then initiated.

[0075] The normal force increase process gradually increases the normal force through a normal impedance controller at preset step sizes under safety boundary constraints. The safety boundary constraints are defined by two constraints: first, the increment of normal force does not exceed the product of the preset maximum allowable compressive stress of the soft rubber layer and the contact area; second, the increment of normal force does not exceed the product of the initial normal force and the safety factor.

[0076] Among them, the maximum allowable compressive stress of the soft rubber layer is obtained by querying the preset material allowable stress library based on the soft rubber material type identified by the material classification probability map. The contact area is calculated based on the surface features of the grasping area reconstructed by the 3D vision sensor. The initial normal force is the stable value of the normal contact force signal after the normal impedance is adjusted and stabilized. The safety factor is a preset empirical coefficient greater than 1, which is used to prevent the soft rubber layer from being crushed or the workpiece from being deformed due to excessive increase in normal force.

[0077] After each increment of normal force, wait for the system to stabilize again and recalculate the friction reserve until the ratio of the required torque to the friction reserve does not exceed the preset proportional threshold. Then stop increasing the normal force. If any safety boundary constraint is touched during the increase and the friction reserve still does not meet the requirements, it is determined that the current gripping state cannot be stabilized through normal-tangential coordinated adjustment. The system alarm is triggered and the abnormal state is recorded, prompting the operation and maintenance personnel to check the gripping parameters or the toothbrush handle's incoming posture.

[0078] The tangential force adjustment adopts a shockless switching algorithm, which switches the control mode of the two degrees of freedom coupled with the eccentric rotational torque direction in the grasping plane from the position control mode to the force control mode. The tangential force controller adopts a PID algorithm, starting from the initial target value calculated in the tangential force active adjustment preparatory mechanism, and taking the desired tangential force target as the control target, outputs control commands to drive the end effector of the robot to apply tangential force in the grasping plane.

[0079] It should be noted that the gripping plane refers to a two-dimensional plane perpendicular to the gripping contact direction within the contact area between the robotic end effector and the toothbrush handle. Before the tangential force adjustment is performed, the robotic end effector adopts a position control mode within the gripping plane, and is controlled by the spatial coordinates of the gripping point.

[0080] It should be noted that the purpose of this step is to achieve on-demand coordination between normal flexibility and tangential active control. Under normal operating conditions, the system maintains the established normal impedance flexibility characteristics and absorbs the impact energy of the hair-embedding through flexible contact. When the tangential force active adjustment preparation mechanism is triggered, the tangential force active adjustment is activated. Without disrupting the normal flexibility balance, the friction force reserve observer and the coordinated normal force increment mechanism ensure that the tangential force can be effectively transmitted to resist the eccentric rotational torque. The two control modes are decoupled and coordinated through friction force reserve, which avoids slippage caused by tangential force adjustment and prevents excessive increase of normal force from damaging the flexibility characteristics. Thus, the two control modes can coexist stably, achieving a dynamic balance between absorbing the impact of the soft and hard interface and resisting the eccentric rotational torque.

[0081] The technical solution of this invention is as follows: Initiate a material identification process for grasping, identify the material region where the grasping area is located. If it is a transition region, output a material classification probability map of the grasping area. Execute a grasping contact action and collect contact force response signals. Identify local contact stiffness online. Integrate the material attribute vector by combining the material region and the material classification probability map. Dynamically adjust the normal impedance parameter and generate a normal flexible contact marker. If a normal flexible contact marker is received, collect the triaxial torque signal at the grasping point. Identify the centroid position vector of the toothbrush handle and calculate the eccentricity parameter. Extract the eccentric rotational torque characterization. Determine whether the tangential force active adjustment preparatory mechanism is triggered and mark the control state. If the control state is tangential force pending intervention, activate the tangential force active adjustment mechanism, calculate the desired tangential force target, introduce a friction reserve observer to determine whether the normal force collaborative increase process is triggered, and execute tangential force adjustment.

[0082] Example 2

[0083] Please see Figure 2 As shown in the embodiment of the present invention, an adaptive gripping force control system for a toothbrush bristle implantation machine robotic arm specifically includes the following modules:

[0084] Material recognition module: Initiates the material recognition process, identifies the material region where the grasped area is located, and outputs the material classification probability map of the grasped area if it is a transitional region.

[0085] Normal control module: Executes grasping contact action and collects contact force response signal, identifies local contact stiffness online, integrates material attribute vector with material region and material classification probability map, dynamically adjusts normal impedance parameter, and generates normal flexible contact mark;

[0086] Eccentricity assessment module: If a normal flexible contact marker is received, the triaxial torque signal at the gripping point is collected, the centroid position vector of the toothbrush handle is identified and the eccentricity parameter is calculated, the eccentric rotation torque characterization is extracted, it is determined whether the tangential force active adjustment preparation mechanism is triggered, and the control status is marked.

[0087] Tangential control and coordination module: If the control state is tangential force pending intervention, activate the tangential force active adjustment mechanism, calculate the desired tangential force target, introduce a friction reserve observer to determine whether to trigger the normal force coordination increase process, and execute tangential force adjustment.

[0088] The embodiments of the present invention have been described in detail above, but the content described is only a preferred embodiment of the present invention and should not be considered as limiting the scope of the present invention. All equivalent changes and improvements made in accordance with the scope of the present invention should still fall within the scope of the present invention.

Claims

1. An adaptive gripping force control method for a robotic arm of a toothbrush bristle implantation machine, characterized in that: Includes the following steps: Initiate the material identification process, identify the material region where the grasping area is located, and if it is a transitional region, output the material classification probability map of the grasping area. Perform the grasping contact action and collect the contact force response signal, identify the local contact stiffness online, integrate the material attribute vector by combining the material region and the material classification probability map, dynamically adjust the normal impedance parameter, and generate a normal flexible contact mark. If a normal flexible contact marker is received, the triaxial torque signal at the gripping point is collected, the centroid position vector of the toothbrush handle is identified and the eccentricity parameter is calculated, the eccentric rotational torque characterization is extracted, it is determined whether the tangential force active adjustment preparation mechanism is triggered, and the control state is marked. If the control state is tangential force pending intervention, the tangential force active adjustment mechanism is activated, the desired tangential force target is calculated, a friction reserve observer is introduced to determine whether the normal force collaborative increase process is triggered, and tangential force adjustment is executed.

2. The adaptive gripping force control method for a toothbrush bristle implantation machine robotic arm according to claim 1, characterized in that: The material identification process employs a hierarchical identification strategy: The minimum Euclidean distance between the preset initial gripping point coordinates and the boundary contours of the extracted soft rubber area and hard area is calculated and defined as the boundary distance parameter. If the boundary distance parameter exceeds the preset boundary neighborhood radius, the gripping area is determined to be located in a single material area; otherwise, the gripping area is determined to be located in a transition area, triggering the hyperspectral imaging mode and outputting the material classification probability map of the gripping area.

3. The adaptive gripping force control method for a toothbrush bristle implantation machine robotic arm according to claim 2, characterized in that: Online identification of local contact stiffness specifically involves: Define a contact time window. Within the contact time window, extract the force response signal in the grasping contact direction and label it as the normal contact force signal. The position change of the end effector in the grasping contact direction is obtained by kinematic forward kinematics calculation using the joint encoder data fed back in real time by the robot control system and labeled as the robot normal displacement signal. Construct a force-displacement curve based on the normal contact force signal and the robot normal displacement signal. Calculate the slope of the force-displacement curve on the sliding sub-window using the least squares fitting method. Take the median value of all calculated slopes and record it as the local contact stiffness.

4. The adaptive gripping force control method for a toothbrush bristle implantation machine robotic arm according to claim 3, characterized in that: The dynamic adjustment of the normal impedance parameter is specifically as follows: When the grasping area is located in a single material area, the normal impedance parameter adopts the preset standard impedance parameter. When the grasping area is located in a transition area, the target impedance stiffness and target damping coefficient are obtained by querying the preset stiffness-impedance parameter mapping curve based on the identified local contact stiffness. The design principle of the stiffness-impedance parameter mapping curve is that the lower the local contact stiffness, the lower the target impedance stiffness and the higher the target damping coefficient.

5. The adaptive gripping force control method for a toothbrush bristle implantation machine robotic arm according to claim 1, characterized in that: The method for obtaining the centroid position vector is as follows: Upon receiving the normal flexible contact marker, the robot arm end effector is controlled to perform a set of preset micro-attitude perturbation actions. During the micro-attitude perturbation actions, the triaxial torque signal and angular acceleration signal at the gripping point are collected. The least squares identification algorithm is used for fitting, and the centroid position vector of the toothbrush handle in the tool coordinate system is calculated online.

6. The adaptive gripping force control method for a toothbrush bristle implantation machine robotic arm according to claim 5, characterized in that: The extraction method for the eccentric rotational torque characterization is as follows: During the hair implantation cycle, the resultant torque vector and resultant torque magnitude are calculated based on the triaxial torque signal. Using a sliding time window filtering technique, the maximum value of the resultant torque magnitude within each sliding time window is extracted and recorded as the instantaneous torque peak value. After smoothing, the torque saturation index is obtained, and the first derivative of the torque saturation index with respect to time is calculated to obtain the torque saturation change rate index. The torque saturation index and the torque saturation change rate index are integrated and recorded as the eccentric rotation torque characterization.

7. The adaptive gripping force control method for a toothbrush bristle implantation machine robotic arm according to claim 3, characterized in that: The method for determining whether the normal force coordination process is triggered is as follows: Obtain the torque saturation index included in the eccentric rotation torque characterization, combine it with the eccentric distance in the eccentric parameters, calculate the theoretical tangential force amplitude required at present through the torque balance equation as the expected tangential force target, and process the expected tangential force target and the eccentric distance to obtain the required torque; Based on the real-time normal contact force signal and the estimated static friction coefficient of the material at the current gripping point, the maximum static friction torque is calculated and defined as the friction reserve. The estimated static friction coefficient is obtained by querying the preset friction coefficient library based on the material property vector. If the ratio of the required torque to the friction reserve exceeds the preset ratio threshold, the normal force co-increase process is triggered.

8. The adaptive gripping force control method for a toothbrush bristle implantation machine robotic arm according to claim 7, characterized in that: The specific process for increasing the normal force coordination is as follows: Under the constraint of safety boundary, the normal force is gradually increased in a preset step size. The safety boundary constraint includes that the increment of normal force does not exceed the product of the preset maximum allowable compressive stress of the soft rubber layer and the contact area, and the increment of normal force does not exceed the product of the initial normal force and the preset safety factor. The contact area is calculated based on the surface features of the grasping area reconstructed by the 3D vision sensor. The initial normal force is the stable value of the normal contact force signal after the normal impedance is adjusted and stabilized. After each increment of normal force, the friction reserve is recalculated until the ratio of the required torque to the friction reserve does not exceed a preset proportional threshold, at which point the increase of normal force is stopped.

9. The adaptive gripping force control method for a toothbrush bristle implantation machine robotic arm according to claim 7, characterized in that: The tangential force adjustment adopts a shockless switching algorithm, which switches the control mode of the two degrees of freedom coupled with the eccentric rotational torque direction in the grasping plane from position control mode to force control mode. The tangential force controller adopts a PID algorithm, takes the desired tangential force target as the control target, and outputs control commands to drive the end effector of the robot to apply tangential force in the grasping plane.

10. An adaptive gripping force control system for a toothbrush bristle implantation machine robotic arm, the system being used to implement the control method as described in any one of claims 1-9, characterized in that, include: Material recognition module: Initiates the material recognition process, identifies the material region where the grasped area is located, and outputs the material classification probability map of the grasped area if it is a transitional region. Normal control module: Executes grasping contact action and collects contact force response signal, identifies local contact stiffness online, integrates material attribute vector with material region and material classification probability map, dynamically adjusts normal impedance parameter, and generates normal flexible contact mark; Eccentricity assessment module: If a normal flexible contact marker is received, the triaxial torque signal at the gripping point is collected, the centroid position vector of the toothbrush handle is identified and the eccentricity parameter is calculated, the eccentric rotation torque characterization is extracted, it is determined whether the tangential force active adjustment preparation mechanism is triggered, and the control status is marked. Tangential control and coordination module: If the control state is tangential force pending intervention, activate the tangential force active adjustment mechanism, calculate the desired tangential force target, introduce a friction reserve observer to determine whether to trigger the normal force coordination increase process, and execute tangential force adjustment.

Images