Stiffness control method for flexible robot arm based on dynamic adjustment of feedback control parameters

Abstract

The present application belongs to the technical field of flexible mechanical arm, and provides a stiffness control method of flexible mechanical arm based on dynamic adjustment of feedback control parameters, comprising: establishing a stiffness adjustment model representing the mapping relationship between the feedback controller parameters and the joint stiffness according to the joint dynamics characteristic data of the flexible mechanical arm; obtaining the task characteristic parameters of the flexible mechanical arm, determining the current task scene and the target joint stiffness corresponding to the current task scene based on the task characteristic parameters; according to the stiffness adjustment model, taking the target joint stiffness and the optimal system stability as the optimization objective, determining the optimal parameter combination of the feedback controller corresponding to the target joint stiffness; and dynamically adjusting the gain parameters of the feedback controller according to the optimal parameter combination to control the joint stiffness of the flexible mechanical arm. The scheme improves the adaptability and operation flexibility of the flexible mechanical arm in the task scene, and effectively guarantees the control precision and stability of the joint stiffness.

Description

Technical Field

[0001] This invention relates to the field of flexible robotic arm technology, and in particular to a method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters. Background Technology

[0002] Flexible robotic arms, with their core advantages of lightweight design and high flexibility, are widely used in human-machine collaboration, precision assembly, and unstructured environment operations. The core technological requirement for these robotic arms lies in balancing operational precision with load capacity and interaction safety with flexibility. Specifically, in scenarios involving precision operation and heavy-duty handling, high joint stiffness is required to ensure trajectory tracking accuracy and structural stability, preventing joint deformation caused by load. Conversely, in scenarios involving human-machine contact and collision avoidance, low joint stiffness is needed to absorb impact energy and reduce the risk of damage to personnel or fragile workpieces.

[0003] Existing stiffness control technologies for flexible robotic arms mainly fall into three typical categories. The first category is stiffness control based on mechanical structure adjustments. This type of solution achieves stiffness switching by modifying the mechanical structure, such as changing the transmission ratio, adding variable stiffness mechanisms, or locking components. However, the additional mechanical components significantly increase system complexity, weight, and manufacturing costs. Furthermore, the stiffness adjustment process is discrete and step-like, resulting in slow response and an inability to achieve a continuous and smooth transition between high and low stiffness scenarios. Simultaneously, friction and wear issues in the mechanical structure not only affect adjustment accuracy but also increase energy consumption and maintenance costs, making it difficult to adapt to the demands of dynamic task scenarios for rapid online stiffness adjustment.

[0004] The second category is stiffness control schemes based on smart materials. These schemes utilize the properties of smart materials such as shape memory alloys and electrorheological fluids to change joint stiffness. Although the structure is relatively compact, it is limited by the physical properties of the materials themselves, resulting in a narrow stiffness adjustment range, slow response speed, and obvious nonlinear phenomena such as hysteresis and creep. This leads to low control accuracy and insufficient long-term reliability, making it difficult to meet the stability and adaptability requirements of industrial applications.

[0005] The third category is stiffness control schemes based on control strategies. Some existing control methods attempt to indirectly affect the compliance of the robotic arm's end effector through torque control or impedance control, but they lack systematic stiffness modeling and online adjustment mechanisms, resulting in insufficient stiffness control accuracy and limited adaptability to various scenarios, making it difficult to achieve autonomous, continuous, and precise adjustment of joint stiffness in diverse tasks.

[0006] It is evident that traditional flexible robotic arm stiffness adjustment schemes suffer from technical problems such as low control precision, poor adaptability to various scenarios, and insufficient efficiency and reliability. Summary of the Invention

[0007] This invention provides a method for controlling the stiffness of a flexible robotic arm based on dynamic adjustment of feedback control parameters, which solves the shortcomings of traditional flexible robotic arm stiffness adjustment schemes, such as low control accuracy, poor scene adaptability, and insufficient efficiency and reliability.

[0008] This invention provides a method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters. The drive unit of the flexible robotic arm is connected to a feedback controller. The method includes:

[0009] Based on the joint dynamics data of the flexible robotic arm, a stiffness adjustment model is established to characterize the mapping relationship between the feedback controller parameters and the joint stiffness.

[0010] Obtain the task feature parameters of the flexible robotic arm, and determine the current task scenario and the target joint stiffness corresponding to the current task scenario based on the task feature parameters;

[0011] Based on the stiffness adjustment model, with the optimization objective of satisfying the target joint stiffness and achieving optimal system stability, the optimal parameter combination of the feedback controller corresponding to the target joint stiffness is determined.

[0012] According to the optimal parameter combination, the gain parameter of the feedback controller is dynamically adjusted to control the joint stiffness of the flexible robotic arm.

[0013] The flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters provided by the present invention establishes a stiffness adjustment model characterizing the mapping relationship between feedback controller parameters and joint stiffness based on the joint dynamics characteristic data of the flexible robotic arm, including:

[0014] Based on the joint dynamics characteristic data of the flexible robotic arm, a theoretical transfer function model of each key component is established.

[0015] In the experimental calibration scenario, the measured joint stiffness under different combinations of feedback controller parameters is obtained, and a database of measured parameters and stiffness is constructed.

[0016] Based on the theoretical transfer function model and the data in the parameter and stiffness measurement database, a stiffness adjustment model is established to characterize the mapping relationship between feedback controller parameters and joint stiffness.

[0017] According to the flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters provided by the present invention, the key components include a position controller, a torque controller, a drive unit, a transmission mechanism that takes into account the elastic deformation of the cable, and a joint dynamics component.

[0018] The method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters provided by the present invention obtains the task characteristic parameters of the flexible robotic arm, including:

[0019] The joint positions, output torque, and environmental perception information of the flexible robotic arm are acquired respectively.

[0020] Based on the joint position, output torque, and environmental perception information, task feature parameters are extracted using a data fusion algorithm.

[0021] The flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters provided by the present invention determines the current task scenario based on the task characteristic parameters, including:

[0022] Establish multiple scene feature templates, each containing a corresponding range of feature parameters;

[0023] The task feature parameters are matched with the feature parameter range in each scene feature template to obtain the matching degree between the task feature parameters and each scene feature template.

[0024] The current task scenario is determined based on the scenario feature template with the highest matching degree and a matching degree exceeding the preset matching degree threshold.

[0025] According to the flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters provided by the present invention, the task feature parameters are matched with the feature parameter range in each scene feature template to obtain the matching degree between the task feature parameters and each scene feature template, including:

[0026] Calculate the parameter deviation between each feature parameter in the task feature parameters and the boundary value of the corresponding feature parameter range in each scene feature template;

[0027] Based on the parameter deviation, the correlation degree between each feature parameter and each scene feature template is calculated;

[0028] The matching degree between the task feature parameters and each scene feature template is calculated by weighted summation of the correlation between all feature parameters and the same scene feature template.

[0029] According to the flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters provided by the present invention, the feedback controller includes at least one of a feedback control module, a feedforward compensation module, and a filter module.

[0030] The feedback control module includes one or more combinations of proportional control submodule, integral control submodule and derivative control submodule;

[0031] The gain parameters include at least one of the following: position loop proportional gain, torque loop proportional gain, integral gain, derivative gain, feedforward gain, and filter parameters.

[0032] The flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters provided by the present invention, according to the stiffness adjustment model, with the optimization objective of satisfying the target joint stiffness and achieving optimal system stability, determines the optimal parameter combination of the feedback controller corresponding to the target joint stiffness, including:

[0033] The control parameter combination corresponding to the target joint stiffness is determined by the stiffness adjustment model.

[0034] With the goal of satisfying the target joint stiffness and achieving optimal system stability, the optimal solution is searched within the feasible region of the feedback controller parameters based on the aforementioned combination of control parameters to obtain the optimal parameter combination.

[0035] The method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters according to the present invention further includes:

[0036] Obtain the joint torque and joint position of the flexible robotic arm;

[0037] Based on the joint torque, the change in joint torque caused by external disturbance is calculated.

[0038] Based on the joint position, the actual change in joint angular displacement is calculated;

[0039] The actual joint stiffness is calculated by dividing the change in joint torque by the change in joint angular displacement.

[0040] Determine the stiffness deviation between the actual joint stiffness and the target joint stiffness;

[0041] If the stiffness deviation is higher than the set deviation threshold, a parameter grading correction strategy is triggered and executed.

[0042] The flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters provided by the present invention triggers and executes a parameter hierarchical correction strategy, including:

[0043] Based on the actual joint stiffness, joint position, and joint torque, multiple deviation evaluation index values ​​are determined, and the stiffness deviation is calculated based on the multiple deviation evaluation index values.

[0044] Determine the number of abnormal joints in the flexible robotic arm whose stiffness deviation exceeds a set deviation threshold;

[0045] If the number of abnormal joints is less than or equal to a set threshold, a first-level correction strategy is triggered and executed. The first-level correction strategy is to adjust at least some of the control parameters in the current optimal parameter combination online.

[0046] If the number of abnormal joints exceeds a set threshold, a secondary correction strategy is triggered and executed. The secondary correction strategy involves adjusting the stiffness adjustment model and regenerating the optimal parameter combination based on the adjusted stiffness adjustment model.

[0047] The flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters provided by this invention establishes a precise mapping model between feedback controller parameters and joint stiffness. It achieves scene recognition and target stiffness matching based on task characteristic parameters, and then completes stiffness control by dynamically adjusting the feedback controller gain parameters. This eliminates the need to modify the mechanical structure of the flexible robotic arm, significantly reducing system complexity, weight, and manufacturing costs, and avoiding wear, hysteresis, and energy consumption problems associated with mechanical adjustments. Simultaneously, the optimal parameter combination determined by the model enables online, continuous, and adaptive adjustment of joint stiffness, quickly responding to the stiffness requirements of different task scenarios and greatly improving the adaptability and operational flexibility of the flexible robotic arm in various task scenarios. Through precise control of feedback parameters, the control accuracy and stability of joint stiffness are effectively guaranteed. Functional improvements can be achieved simply by upgrading the control algorithm, providing efficient and reliable technical support for collaborative robots and other applications requiring high safety and high flexibility. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0049] Figure 1 This is a schematic diagram of the stiffness control system of a flexible robotic arm;

[0050] Figure 2 This is a schematic diagram of the control principle of the stiffness control system for a flexible robotic arm;

[0051] Figure 3 This is a flowchart illustrating the stiffness control method for a flexible robotic arm based on dynamic adjustment of feedback control parameters provided in an embodiment of the present invention.

[0052] Figure 4 This is a schematic diagram illustrating the implementation principle of a flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters.

[0053] Figure 5 This is a schematic diagram illustrating the construction process of the stiffness adjustment model. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0055] The following is combined Figures 1 to 5 This invention describes the detailed scheme of a flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters, as provided in an embodiment of the present invention.

[0056] First, the execution entity upon which the flexible robotic arm stiffness control method based on dynamic adjustment of feedback control parameters provided in this embodiment of the invention relies will be described. Specifically, this method is implemented using a flexible robotic arm stiffness control system. After the system is powered on, a complete initialization self-test process is first executed to ensure that all parts are in normal working condition. For example... Figure 1 As shown, the system specifically includes:

[0057] The central control unit 110, as the core of the system's intelligent decision-making, incorporates a stiffness adjustment algorithm and is responsible for executing logic such as stiffness adjustment model calculation, task scenario recognition, and control parameter decision-making. In practical applications, the central control unit 110 can adopt a multi-core processor architecture, where the main core runs a real-time operating system, specifically responsible for the generation and scheduling of control commands; the auxiliary core runs a general-purpose operating system, mainly handling perception data fusion and decision-making algorithms. During initialization, the central control unit first loads the preset parameter library of the stiffness adjustment model, establishes real-time data channels for communication with each unit, initializes the task scenario recognition engine, and sets system safety monitoring thresholds. These thresholds include the maximum permissible stiffness deviation, the minimum stable stiffness value, and the safe operating range of each joint.

[0058] The control unit 120 receives parameter setting instructions from the central control unit and implements feedback control through a feedback controller. The feedback controller specifically involves a position controller and a torque controller, including dynamically adjustable proportional, integral, and derivative gains, feedforward compensation parameters, and filter parameters. In practical applications, the control unit 120 can employ a reconfigurable field-programmable gate array (FPGA) combined with a microprocessor architecture to achieve high real-time control algorithms. During the initialization phase, the control unit 120 loads default proportional-integral-derivative (PID) parameters, sets a fixed control cycle to ensure the system bandwidth meets requirements, and configures feedforward compensation parameters, including velocity and acceleration feedforward coefficients. The initial values ​​of these parameters are set based on the nominal dynamic parameters of the flexible robotic arm.

[0059] The drive unit 130, consisting of a servo motor, a reducer, and a cable transmission mechanism, is responsible for converting control signals into actual joint torque outputs. The cable transmission mechanism includes a pulley system and a tension amplification mechanism. Before practical application, the drive unit 130 needs to complete a series of self-test procedures, including drive unit enable status detection, initial cable tension measurement, joint range of motion limit detection, and emergency stop safety circuit testing. During cable tension measurement, the normal tension range should be maintained within a specific interval; excessively high or low tension will affect the accuracy of stiffness adjustment.

[0060] The joint 140 of the flexible robotic arm, as the controlled object, directly affects the stiffness performance of the system due to its dynamic characteristics.

[0061] The sensing unit 150, composed of a multi-source sensor network, includes a joint encoder, a tension sensor (in some scenarios, a motor current sensor may be used instead), a vision sensor, and a tactile sensor. It is responsible for collecting the state and environmental information of the flexible robotic arm, enabling the acquisition of task characteristic parameters applicable to the task scenario. Before normal application, the joint encoder needs to undergo zero-position calibration to establish an absolute position coordinate system; the tension sensor needs zero-drift compensation and sensitivity calibration; the vision sensor needs intrinsic parameter calibration and hand-eye calibration; and the tactile sensor needs pressure sensitivity calibration. During calibration, the system records the baseline data of each sensor, providing an accurate reference benchmark for subsequent real-time data acquisition.

[0062] It is understandable that the task scenario can provide external environmental information and operation instructions, which can serve as the triggering conditions and target basis for stiffness adjustment.

[0063] The core principle of the above-mentioned flexible robotic arm stiffness control system is as follows: Figure 2 As shown, the system employs a dual closed-loop control structure. Specifically, in the outer loop position control channel, a position error is generated by comparing the reference input of position information with the measured position. The position controller 210 can generate the desired torque command based on the position error. In the inner loop torque control channel, the torque controller 220 receives the torque command and compares it with the actual joint torque, then outputs a drive signal containing the motor control voltage to the drive unit. In the servo drive stage, the servo motor and reducer module 230 in the drive unit converts the drive signal into output torque, taking into account the feedback from the drive interference feedback module 240, to improve system stability. In the transmission conversion stage, the transmission mechanism 250 can realize the conversion from motor-end torque to joint-end torque. This process includes modeling of nonlinear factors such as cable transmission ratio and pulley efficiency.

[0064] In the joint dynamics stage, the joint dynamics module 260 of the flexible robotic arm joint, considering dynamic parameters such as inertia, damping, and mechanical stiffness, can describe the core conversion relationship from torque to position, and also includes the input channel for external disturbance torques. In the speed feedback channel, by considering the speed transformation module 270 and the speed differentiation module 280 of the cable and pulley, a complete feedback path from joint speed to motor back electromotive force can be constructed. Figure 2 The theoretical basis of joint stiffness adjustment is clearly demonstrated. By adjusting the parameters of position controller 210 and torque controller 220, the closed-loop stiffness characteristics of the system can be directly changed, realizing online intelligent adjustment of stiffness.

[0065] like Figure 3 and Figure 4 As shown in the embodiment of the present invention, the stiffness control method for a flexible robotic arm based on dynamic adjustment of feedback control parameters is provided. The drive unit of the flexible robotic arm is connected to the feedback controller. The method mainly includes the following steps:

[0066] Step 310: Based on the joint dynamics characteristic data of the flexible robotic arm, establish a stiffness adjustment model that characterizes the mapping relationship between the feedback controller parameters and the joint stiffness.

[0067] Understandably, this step is part of establishing the stiffness adjustment model, that is... Figure 4 In step 410, the stiffness adjustment model can realize the quantitative correlation between the feedback controller parameters and the joint stiffness, and can provide data basis for subsequent stiffness control.

[0068] Specifically, the joint dynamics characteristic data includes, but is not limited to, the joint's inertia, damping, and stiffness parameters, and the feedback controller parameters include, but are not limited to, the gain parameters of the position controller and the gain parameters of the torque controller.

[0069] Step 320: Obtain the task feature parameters of the flexible robotic arm, and based on the task feature parameters, determine the current task scenario and the target joint stiffness corresponding to the current task scenario.

[0070] Specifically, task characteristic parameters can be obtained through data fusion based on information such as joint position, output torque, and environmental perception.

[0071] This step involves acquiring task feature parameters, identifying the task scene, and determining the target joint stiffness. Figure 4 Steps 420, 430 and 440.

[0072] Step 330: Based on the stiffness adjustment model, with the goal of satisfying the target joint stiffness and achieving optimal system stability, determine the optimal parameter combination of the feedback controller corresponding to the target joint stiffness.

[0073] This step is the process of determining the optimal combination of parameters, i.e. Figure 4 Step 450.

[0074] Step 340: Dynamically adjust the gain parameter of the feedback controller according to the optimal parameter combination to control the joint stiffness of the flexible robotic arm.

[0075] This step is the process of performing stiffness adjustment, i.e. Figure 4 Step 460.

[0076] The stiffness control method for flexible robotic arms based on dynamic adjustment of feedback control parameters provided in this embodiment is applicable to flexible robotic arms with various drive methods, including electric, hydraulic, and pneumatic drives. For hydraulic drives, a force and displacement model of the hydraulic actuator needs to be established, and stiffness adjustment is achieved by adjusting the parameters of the hydraulic control valve; for pneumatic drives, the compressibility of gas needs to be considered, and a similar function is achieved by adjusting the air pressure control parameters.

[0077] In one embodiment, based on the joint dynamics data of the flexible robotic arm, a stiffness adjustment model is established to characterize the mapping relationship between the feedback controller parameters and the joint stiffness, specifically including:

[0078] First, based on the joint dynamics characteristic data of the flexible robotic arm, a theoretical transfer function model of each key component is established.

[0079] In this embodiment, as Figure 5 As shown, the construction process of the stiffness adjustment model includes three main stages: theoretical modeling stage 510, experimental calibration stage 520, and model fusion stage 530.

[0080] In the theoretical modeling stage, the multi-degree-of-freedom dynamic characteristics of the flexible robotic arm can be considered to achieve system modeling. Specifically, through closed-loop control system analysis and dynamic equation derivation, a complete dynamic equation including the inertia matrix, Coriolis force and centrifugal force matrices, gravity term, and joint stiffness matrix is ​​established. By analyzing the dynamic characteristics of each component in the system, transfer function models of the position controller, torque controller, drive unit, transmission mechanism, and joint dynamics are established in the transfer function establishment stage.

[0081] In practical applications, the nonlinear characteristics of the transmission mechanism can significantly affect the accuracy of stiffness adjustment, thus requiring the establishment of a precise compensation mechanism. In particular, for cable transmission mechanisms, a mathematical model considering nonlinear factors needs to be established, as detailed below:

[0082] (1)

[0083] in, This represents the final output torque of the flexible robotic arm joint. This represents the initial torque output by the servo motor in the drive unit. Indicates joint angle The transmission stiffness function can comprehensively reflect the influence of the number of pulley turns, pulley center width, cable mechanical stiffness, and joint angle on transmission efficiency.

[0084] In practical applications, the specific form of the transmission stiffness function can be adjusted according to the design of the transmission mechanism. For example, the above transmission stiffness function can be expressed as:

[0085] (2)

[0086] in, n This indicates the number of times the cable wraps around the pulley. w Indicates the effective radius of the pulley. k 缆绳 This indicates the stiffness coefficient of the cable.

[0087] Meanwhile, in the transmission ratio compensation stage, the impact of joint angle changes on the effective transmission ratio is considered. By establishing a functional relationship between the transmission ratio and the joint angle, the actual transmission ratio at the current angle can be calculated in real time. When the joint is in different positions, the transmission ratio will change regularly, and the compensation algorithm dynamically adjusts the torque transmission relationship based on the angle sensor data.

[0088] In the cable elastic deformation compensation stage, a mathematical model of cable tension and deformation can be established based on the elongation effect of the cable under tension. By monitoring the motor output torque and joint angle, the actual deformation of the cable can be estimated. The compensation algorithm calculates the stiffness correction coefficient based on the deformation to ensure the accuracy of stiffness control.

[0089] In the friction compensation stage, Coulomb friction and viscous friction in the transmission system can be considered. The friction torque at different speeds can be calibrated experimentally to establish a functional relationship between friction torque and velocity. A feedforward friction compensation term can be added to the control command to counteract the influence of friction and improve the linearity of stiffness control.

[0090] During operation, the feedback characteristics of the drive unit need to be compensated. For electric drives, back electromotive force needs to be compensated; for hydraulic drives, pressure fluctuations need to be compensated; and for pneumatic drives, gas compressibility needs to be compensated.

[0091] In the velocity feedforward compensation stage, a compensation signal can be generated in advance based on the planned motion trajectory. By acquiring the expected motion velocity and acceleration, the required feedforward compensation amount can be calculated. This feedforward compensation mechanism can effectively reduce tracking errors and improve the system's response speed.

[0092] In the acceleration compensation stage, to address the influence of inertial forces, a dynamic model of the robotic arm is established to calculate the additional torque required for each joint during acceleration. The compensation algorithm generates inertial compensation torques based on real-time acceleration data, improving the dynamic performance of the system.

[0093] All compensation algorithms undergo rigorous experimental verification to ensure their effectiveness under various operating conditions. Compensation parameters can be tuned online based on the actual state of the flexible robotic arm to adapt to changes in system characteristics. Simultaneously, a compensation limit protection is implemented to prevent system instability caused by overcompensation.

[0094] In practical applications, key components may include position controllers, torque controllers, drive units, transmission mechanisms that take into account the elastic deformation of cables, and joint dynamics components.

[0095] Furthermore, in the derivation of the system stiffness formula, the system stiffness transfer function expression can be derived from the theoretical transfer function models of each key component and the system connection relationships, as follows:

[0096] (3)

[0097] in, G 1 represents the position controller transfer function. G 2 represents the torque controller transfer function. G 3 represents the torque output transfer function of the drive unit. G 4 represents the transfer function of the transmission mechanism. G 5 represents the transfer function for generating the back electromotive force. G 6 represents the velocity transfer function. G 7 represents the velocity differential transfer function. G 8 represents the joint dynamics transfer function of the flexible robotic arm, which can be adjusted... G 1. G The parameter value of 2 can change the stiffness characteristics of the system.

[0098] It is easy to see that the system stiffness transfer function expression clearly reveals the direct influence of the position controller parameters and torque controller parameters on the system stiffness, providing a theoretical basis for subsequent stiffness adjustment. Through partial differential analysis, the monotonicity relationship between controller parameter changes and system stiffness can be verified, ensuring the deterministic direction of stiffness adjustment.

[0099] It should be noted that the above system stiffness transfer function expression is only an example. The actual expression form can be adjusted according to the system configuration, and then the relationship between parameters and stiffness can be analyzed based on the system stiffness transfer function expression.

[0100] Meanwhile, in the experimental calibration scenario, the measured joint stiffness under different combinations of feedback controller parameters was obtained, and a database of measured parameters and stiffness was constructed.

[0101] During the experimental calibration phase, a high-precision calibration experimental platform can be established during the experimental platform construction stage. This platform includes measurement equipment such as a six-dimensional torque sensor, a laser tracker, a dynamic signal analyzer, and a high-precision current sensor. A systematic scan test, i.e., parameter space scanning, is performed in the controller parameter space. By setting different combinations of proportional gain, derivative gain, and integral gain, the stiffness response characteristics of the flexible robotic arm under each parameter combination are recorded.

[0102] In the data acquisition and processing phase, standardized perturbation torques can be applied, and the actual performance stiffness value can be calculated by measuring the angular displacement response of the joints. To ensure data reliability, each parameter combination needs to be repeatedly tested, and the average value is taken as the final result. Simultaneously, frequency domain characteristic data, including the system's amplitude-frequency and phase-frequency characteristics, are recorded to provide supplementary data for model verification.

[0103] Finally, based on the theoretical transfer function model and data from the parameter and stiffness measurement database, a stiffness adjustment model is established to characterize the mapping relationship between the feedback controller parameters and joint stiffness.

[0104] In the model fusion phase, theoretical derivations can be combined with experimental calibration results. Specifically, regression analysis can be used to establish a mapping function between feedback controller parameters and joint stiffness. This function includes linear and cross terms, accurately describing the stiffness variation under multi-parameter coupling, and can serve as a stiffness adjustment model. In the model validation and optimization phase, statistical methods can be used to verify the accuracy of the stiffness adjustment model, evaluate the goodness of fit and confidence intervals, and ensure that the stiffness adjustment model has sufficient accuracy.

[0105] Ultimately, a stiffness adjustment model library can be formed, containing stiffness adjustment models for each joint. Each joint's stiffness adjustment model includes specific parameter mapping relationships and boundary conditions, accurately reflecting the stiffness characteristics of the joint at different operating points. In practical applications, the model library can adopt a hierarchical storage structure for easy and rapid querying and real-time updates.

[0106] In one embodiment, obtaining the task characteristic parameters of the flexible robotic arm specifically includes:

[0107] First, the joint positions, output torque, and environmental perception information of the flexible robotic arm are acquired.

[0108] In this embodiment, multi-dimensional data, including joint position, output torque, visual information, and tactile feedback information, can be continuously collected through a sensor network.

[0109] Then, based on the joint position, output torque, and environmental perception information, task feature parameters are extracted through a data fusion algorithm.

[0110] In practical applications, task feature parameters, including load characteristics, motion characteristics, environmental characteristics, and task characteristics, can be extracted from multi-dimensional data.

[0111] The load characteristics are obtained by analyzing the output torque of the torque sensor, and include the load weight and the position of the load's center of mass. Motion characteristics are obtained by processing joint positions, and include maximum velocity and acceleration. Environmental characteristics are obtained through visual and tactile feedback information acquired by vision and tactile sensors, and include environmental stiffness characteristics and contact force magnitude. Task characteristics are determined by analyzing trajectory data and operational commands, and include task accuracy and safety requirements.

[0112] In one embodiment, determining the current task scenario based on task feature parameters specifically includes:

[0113] First, multiple scene feature templates are established, each containing a corresponding range of feature parameters.

[0114] In practical applications, scene feature templates for six typical stiffness scenarios can be pre-established, including scenarios such as high-precision operation, heavy-load handling, human-computer interaction, collision protection, precision assembly, and free movement. Each scene feature template defines the normal range of feature parameters, i.e., the feature parameter range, forming a complete scene recognition knowledge base.

[0115] Then, the task feature parameters are matched with the feature parameter range in each scene feature template to obtain the matching degree between the task feature parameters and each scene feature template.

[0116] In a specific implementation, the task feature parameters are matched with the feature parameter range in each scene feature template to obtain the matching degree between the task feature parameters and each scene feature template, specifically including:

[0117] The first step is to calculate the parameter deviation between each feature parameter in the task feature parameters and the boundary value of the corresponding feature parameter range in each scene feature template.

[0118] In this embodiment, the difference between each feature parameter and the upper and lower limits of the corresponding feature parameter range can be calculated to obtain two difference values, and the smaller difference is taken as the parameter deviation.

[0119] The second step is to calculate the correlation between each feature parameter and each scene feature template based on the parameter deviation.

[0120] Understandably, correlation can characterize the degree of association between a feature parameter and each scene feature template. The higher the correlation, the stronger the association between the feature parameter and the current scene feature template. In practical applications, the correlation can be obtained by adding the parameter deviation to a constant value close to 0 and taking the reciprocal of the sum.

[0121] The third step is to perform a weighted summation of the correlation between all the feature parameters and the same scene feature template to calculate the matching degree between the task feature parameters and each scene feature template.

[0122] In practical applications, the weight values ​​of the feature templates for each scenario in the weighted summation process can be reasonably set according to actual needs.

[0123] Finally, the current task scenario is determined based on the scenario feature template with the highest matching degree that exceeds the preset matching degree threshold.

[0124] In this embodiment, to ensure the reliability of the task scenario determination process, the target scenario feature template is filtered through two limiting conditions. One limiting condition is the highest matching degree, and the other limiting condition is that the matching degree exceeds a preset matching degree threshold, such as higher than 0.7. When both limiting conditions are met simultaneously, a scene feature template that meets the requirements can be filtered out, and the current task scenario can be determined as the task scenario corresponding to the scene feature template that meets the requirements, thereby improving the credibility of the task scenario determination process.

[0125] Furthermore, after determining the current task scenario, the target joint stiffness of each joint under the current task scenario can be determined from a preset stiffness target mapping table. Each task scenario corresponds to a specific stiffness range, including minimum stiffness, maximum stiffness, and recommended stiffness. The recommended stiffness is the optimal value obtained based on a large amount of experimental data, which can ensure the stability of the system while meeting the task requirements.

[0126] Considering the complexity of actual tasks, the process of determining the target joint stiffness also possesses adaptive adjustment capabilities. Specifically, based on real-time monitoring of load conditions, motion states, and environmental changes, the target joint stiffness can be dynamically adjusted from the initially determined value. For example, in heavy-duty handling scenarios, as the load weight increases, the target stiffness value can be appropriately increased to ensure operational stability; in human-computer interaction scenarios, when personnel are detected approaching, the target stiffness value can be decreased to improve safety.

[0127] In some embodiments, the determination of the target joint stiffness can also take into account the synergistic effect of each joint. For end effectors requiring precise positioning, the target joint stiffness of the relevant joints can be increased accordingly; while for the base joints that mainly provide support, the target joint stiffness can be appropriately relaxed. This differentiated stiffness allocation strategy ensures operational performance while optimizing system energy consumption.

[0128] In one embodiment, the feedback controller includes at least one of a feedback control module, a feedforward compensation module, and a filter module.

[0129] The feedback control module includes one or more of the following: proportional control submodule, integral control submodule, and derivative control submodule.

[0130] Gain parameters include at least one of the following: position loop proportional gain, torque loop proportional gain, integral gain, derivative gain, feedforward gain, and filter parameters.

[0131] In this embodiment, the feedback control module in the position controller can be a proportional-integral-derivative (PI-DI) control module, but it is not limited to this and can also be other types of feedback control modules. For the torque controller, the stiffness characteristics of the joint output torque can be directly adjusted by adjusting the force loop gain parameters (such as proportional gain and integral gain).

[0132] Specifically, gain parameters include proportional gain, integral gain, derivative gain, feedforward gain, and filter parameters. The adjustment principles of various gain parameters are as follows:

[0133] The proportional gain is adjusted primarily based on the magnitude and direction of the stiffness deviation. When the actual stiffness is lower than the target value, the proportional gain can be increased proportionally; when the actual joint stiffness is higher than the target joint stiffness, the proportional gain should be decreased accordingly. A safety boundary is set during the adjustment process to prevent excessive gain from causing system instability. Simultaneously, the current motion state is considered, and the adjustment range is appropriately widened during high-speed motion to avoid oscillations caused by frequent adjustments.

[0134] The adjustment of the derivative gain focuses on improving the system's damping characteristics. This can be achieved by dynamically adjusting the derivative gain by monitoring the vibration level of the flexible robotic arm. When significant vibration is detected, the derivative gain is appropriately increased to suppress oscillations; when the system response is too sluggish, the derivative gain is decreased. The adjustment of the derivative gain is coordinated with the proportional gain to ensure a balance between system stability and speed.

[0135] The adjustment of the integral gain is mainly aimed at eliminating steady-state errors. It can intelligently adjust the intensity of the integral gain by monitoring the accumulation of position tracking errors. When a large steady-state error occurs, the integral gain is increased; when the error is small, the integral gain is decreased to prevent integral saturation. Under specific circumstances, such as detecting continuous small-amplitude oscillations, the system will temporarily disable the integral gain and re-enable it after the system stabilizes.

[0136] The adjustment of the feedforward gain is based on the evaluation of trajectory tracking performance. By analyzing the deviation between the commanded position and the actual position, the feedforward gain setting is optimized. The velocity feedforward gain can be adaptively adjusted according to the motion velocity, while the acceleration feedforward gain can be optimized according to the rate of change of acceleration. Feedforward compensation and feedback control complement each other, jointly improving the tracking accuracy of the system.

[0137] Adjusting the filter parameters is aimed at suppressing high-frequency interference. The high-frequency components of the control signal can be monitored in real time, and the filter cutoff frequency can be adaptively adjusted. In scenarios with high rigidity requirements, the cutoff frequency can be appropriately increased to maintain system bandwidth; in scenarios requiring noise suppression, the cutoff frequency can be decreased to filter out high-frequency interference.

[0138] All gain parameter adjustments follow a smooth transition principle to avoid system shocks caused by abrupt gain parameter settings. Specifically, an incremental adjustment strategy can be adopted, with each adjustment being limited in magnitude to ensure system stability. Simultaneously, establishing a historical record of parameter adjustments provides data support for subsequent optimizations.

[0139] In one embodiment, the optimal parameter combination of the feedback controller corresponding to the target joint stiffness is determined based on the stiffness adjustment model, specifically including:

[0140] First, the combination of control parameters corresponding to the target joint stiffness is determined by using a stiffness adjustment model.

[0141] Then, with the goal of satisfying the target joint stiffness and achieving optimal system stability as the optimization objective, the optimal solution is searched within the feasible region of the feedback controller parameters based on the combination of control parameters to obtain the optimal parameter combination.

[0142] In practical applications, for scenarios with low precision requirements, the combination of control parameters determined based on the stiffness adjustment model can be directly used as the optimal parameter combination for subsequent joint stiffness control. However, for scenarios with high precision requirements, the optimal control parameters can be determined through optimization based on the initially obtained control parameter combination.

[0143] Specifically, the optimization can focus on meeting stiffness targets, ensuring system stability, and optimizing dynamic performance as multi-objective optimization directions, searching for the optimal solution within the feasible region of the controller parameters. First, the objective function is defined, with minimizing the deviation between the actual stiffness and the target stiffness as the primary objective, and minimizing system settling time and overshoot as secondary objectives, assigning different weights to each objective. Then, optimization algorithms such as gradient descent, Newton's method, or the LM algorithm are used to calculate the gradient direction of the objective function based on real-time acquired data, iteratively updating the controller parameter combination. During the iteration process, system stability indicators are monitored in real time; if an unstable trend appears, the iteration is immediately terminated and the system reverts to the previous control parameter combination, ultimately finding the optimal parameter combination that balances stiffness accuracy, response speed, and stability under the current operating conditions.

[0144] In one embodiment, the above-described method for controlling the stiffness of a flexible robotic arm based on dynamic adjustment of feedback control parameters may further include:

[0145] First, obtain the joint torque and joint position of the flexible robotic arm.

[0146] In this embodiment, the operating status of the flexible robotic arm can be continuously monitored online, that is... Figure 4 In step 470, the monitoring process can collect multi-dimensional real-time data, including joint torque, joint position, and joint stiffness. By establishing a real-time data buffer, the operation records for the most recent period can be saved, thus providing a data foundation for performance evaluation.

[0147] Then, based on the joint torque, the change in joint torque caused by external disturbance is calculated.

[0148] At the same time, based on the joint position, the actual change in joint angular displacement is calculated.

[0149] Next, the actual joint stiffness is calculated by dividing the change in joint torque by the change in joint angular displacement.

[0150] In this embodiment, the formula for calculating the actual joint stiffness is:

[0151] (4)

[0152] in, This represents the actual joint stiffness. This represents the change in joint torque caused by external disturbances. This represents the actual change in joint angular displacement.

[0153] Subsequently, the stiffness deviation between the actual joint stiffness and the target joint stiffness is determined.

[0154] In this embodiment, the stiffness deviation can be calculated by subtraction.

[0155] Finally, if the stiffness deviation exceeds the set deviation threshold, a parameter grading correction strategy is triggered and executed.

[0156] In this embodiment, whether correction is needed can be determined based on the stiffness deviation, i.e. Figure 4 In step 480, specifically, if the stiffness deviation exceeds the set deviation threshold, correction is required, and a correction command is generated (step 490). This triggers and executes a parameter hierarchical correction strategy, subsequently requiring either primary or secondary correction depending on the situation. Figure 4 In step 4110; if the stiffness deviation is less than or equal to the set deviation threshold, no correction is needed, and the current parameters are maintained, i.e., step 4100.

[0157] In a specific implementation, triggering and executing the parameter hierarchical correction strategy includes:

[0158] First, based on the actual joint stiffness, joint position, and joint torque, multiple deviation evaluation index values ​​are determined, and the stiffness deviation is calculated based on these multiple deviation evaluation index values.

[0159] In this embodiment, the deviation evaluation index values ​​specifically include stiffness consistency index values, stability index values, and tracking performance index values. The stiffness consistency index value reflects the degree of consistency between the actual joint stiffness and the target joint stiffness, and can be obtained by calculating the statistical characteristics of the stiffness deviation. The stability index value assesses the system's operational smoothness by analyzing the signal fluctuations of joint position and joint torque. The tracking performance index value evaluates the accuracy of trajectory tracking by comparing the deviation between the target value and the actual value of the joint position.

[0160] In practical applications, the absolute value of the difference between each deviation evaluation index value and the corresponding preset index reference value can be calculated. The absolute values ​​of the index differences corresponding to each deviation evaluation index value are weighted and summed. The weighted summation of index deviations is then multiplied by the preset deviation correction coefficient to calculate the stiffness deviation.

[0161] In this embodiment, the formula for calculating the stiffness deviation is:

[0162] (5)

[0163] in, f For stiffness deviation, For deviation correction factor, For the first i The weight value of each deviation from the evaluation index value w i For the first i The absolute value of the difference between the actual value and the reference value of each deviation from the evaluation index.n This represents the number of deviations from the evaluation index value.

[0164] Understandably, the calculation of stiffness deviation integrates multiple deviation evaluation index values, each of which is assigned a corresponding weight. Through weighted summation and other operations, the stiffness deviation can be obtained, objectively reflecting the gap between the current system performance and the expected level. Furthermore, the calculation of stiffness deviation can consider the time factor, with recent data receiving higher weight, ensuring the timeliness of the evaluation results.

[0165] Next, the number of abnormal joints in the flexible robotic arm whose stiffness deviation exceeds the set deviation threshold is determined.

[0166] On the one hand, if the number of abnormal joints is less than or equal to the set number threshold, a first-level correction strategy is triggered and executed. The first-level correction strategy is to adjust at least some of the control parameters in the current optimal parameter combination online.

[0167] This step is the first-level correction stage, which specifically adjusts at least some control parameters, namely... Figure 4 Step 4120.

[0168] On the other hand, if the number of abnormal joints exceeds the set threshold, a secondary correction strategy is triggered and executed. The secondary correction strategy is to adjust the stiffness adjustment model and regenerate the optimal parameter combination based on the adjusted stiffness adjustment model.

[0169] This step is a secondary correction process, specifically updating the stiffness adjustment model, i.e. Figure 4 Step 4130.

[0170] In this embodiment, the number of abnormal joints can be counted based on the stiffness deviation, and different correction strategies can be adopted according to the number of abnormal joints. When the stiffness deviation is small or the number of abnormal joints is small, a first-level correction measurement is performed, that is, at least some control parameters are adjusted online. Specifically, the relevant control parameters can be adjusted in a targeted manner by analyzing the specific performance of each performance index. For example, when the actual joint stiffness is detected to be low, the proportional gain can be appropriately increased; when the signal fluctuations of joint position and joint torque are detected to be severe, the derivative gain can be increased.

[0171] When the stiffness deviation is large and persistently abnormal, or when there are many abnormal joints, a second-level correction measurement is performed, i.e., the stiffness adjustment model is updated. In this case, experimental data under the current working conditions can be collected to recalibrate the stiffness adjustment model. The update of the stiffness adjustment model can adopt an incremental learning approach, incorporating new experience while maintaining the original knowledge, and gradually improving the adaptability of the stiffness adjustment model.

[0172] In practical applications, the generation of correction commands can take into account the real-time operating status of the system. During critical operation phases, the system will suspend non-urgent correction operations to ensure task continuity. A safety monitoring mechanism can be set up during the correction process to monitor the correction effect in real time; if performance degradation is detected, the system should immediately revert to the previous state. The triggering conditions for correction commands can be set based on the statistical characteristics of stiffness deviation.

[0173] In some embodiments, a performance trend analysis function can also be set up, which can identify changes in system characteristics through long-term monitoring data, predict system performance degradation by analyzing the changing trends of indicators such as stiffness deviation and vibration level, and then arrange maintenance measures in advance to avoid the occurrence of sudden failures.

[0174] In practical applications, the knowledge accumulation stage can save all correction records and corresponding correction effect evaluation data. This historical data can provide valuable references for subsequent model optimization, forming a continuous accumulation of experience and knowledge. As the running time increases, the system's self-adjustment ability can be continuously improved.

[0175] In the application phase, combining the aforementioned control methods and relying on the stiffness control system of the flexible robotic arm, the various functional units in the system can be organically integrated to form a complete intelligent stiffness adjustment closed loop. The system adopts a hierarchical control architecture, with high-level decision-making and low-level control working together to ensure the accuracy and real-time performance of stiffness adjustment. The main control loop runs at a fixed cycle, with each control cycle completing the entire process from sensor data acquisition to drive command output. The control cycle can be optimized according to system requirements, ensuring both control accuracy and computational efficiency. Critical tasks have high execution priority to ensure timely system response.

[0176] In the data flow management phase, it is necessary to ensure smooth information transmission between functional units. Specifically, a unified data format and timestamps can be used to guarantee data consistency and timeliness. Multiple backups should be implemented for critical data to prevent system anomalies caused by data loss. Simultaneously, regarding anomaly handling mechanisms, the operational status of each component must be monitored in real time to promptly detect and address anomalies. When sensor malfunctions, actuator abnormalities, or communication interruptions are detected, the system will automatically switch to a safe mode to prevent accidents. Fault information can be recorded in detail to provide a basis for subsequent analysis and processing.

[0177] In the performance verification phase, multiple test items can be set to comprehensively evaluate various performance indicators of the system. Specifically, in the stiffness accuracy test verification phase, the control accuracy of the system can be tested under different target joint stiffnesses, and the evaluation is carried out by statistically analyzing the distribution characteristics of the actual joint stiffness. In the response speed test phase, the time required from receiving stiffness control commands to reaching a steady state can be measured to evaluate the dynamic performance of the system.

[0178] In the repeatability testing phase, the consistency of the system can be verified by repeatedly performing the same stiffness adjustment process. In practical applications, tests can be conducted under different load conditions and motion states to verify the system's adaptability. Subsequent long-term operation tests can be performed to examine the system's stability and reliability during continuous operation.

[0179] During the safety verification phase, it is necessary to ensure the system's safety protection capabilities under abnormal operating conditions. Specific tests include the effectiveness of functions such as emergency shutdown, limit protection, and overload protection. Simultaneously, the system's operational stability under harsh conditions such as power fluctuations and signal interference must be verified.

[0180] The compatibility testing phase verifies the system's adaptability to different models of flexible robotic arms. By replacing the flexible robotic arm itself, the system's versatility is tested, ensuring the scalability of the technical solution. Simultaneously, the communication interface with external devices is tested to verify the system's integration capabilities.

[0181] Based on the test results above, final parameter optimization can be performed on the entire system. Control parameters, compensation coefficients, and safety thresholds are adjusted to bring the system to its optimal operating state. The optimization process can consider the needs of different application scenarios, providing customizable performance configurations.

[0182] Ultimately, the system possesses complete intelligent stiffness adjustment capabilities, enabling it to adapt to diverse application needs. Software updates can continuously improve system performance, and the modular design facilitates maintenance and upgrades while also offering excellent scalability.

[0183] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters, characterized in that, The drive unit of the flexible robotic arm is connected to a feedback controller, and the method includes: Based on the joint dynamics data of a flexible robotic arm, a stiffness adjustment model is established to characterize the mapping relationship between feedback controller parameters and joint stiffness. This includes: establishing theoretical transfer function models for each key component based on the joint dynamics data of the flexible robotic arm; obtaining measured joint stiffness under different combinations of feedback controller parameters in an experimental calibration scenario, and constructing a database of measured parameters and stiffness; and establishing a stiffness adjustment model characterizing the mapping relationship between feedback controller parameters and joint stiffness based on the theoretical transfer function model and the data in the measured database of parameters and stiffness. Specifically, a regression analysis method is used to establish a mapping function between feedback controller parameters and joint stiffness, which includes linear and cross terms, serving as the stiffness adjustment model. Obtain the task feature parameters of the flexible robotic arm, and determine the current task scenario and the target joint stiffness corresponding to the current task scenario based on the task feature parameters; Based on the stiffness adjustment model, with the optimization objective of satisfying the target joint stiffness and achieving optimal system stability, the optimal parameter combination of the feedback controller corresponding to the target joint stiffness is determined. According to the optimal parameter combination, the gain parameter of the feedback controller is dynamically adjusted to control the joint stiffness of the flexible robotic arm.

2. The method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters according to claim 1, characterized in that, The key components include a position controller, a torque controller, a drive unit, a transmission mechanism that takes into account the elastic deformation of the cable, and a joint dynamics component.

3. The method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters according to claim 1, characterized in that, Obtain the task characteristic parameters of the flexible robotic arm, including: The joint positions, output torque, and environmental perception information of the flexible robotic arm are acquired respectively. Based on the joint position, output torque, and environmental perception information, task feature parameters are extracted using a data fusion algorithm.

4. The method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters according to claim 1, characterized in that, Based on the aforementioned task feature parameters, the current task scenario is determined, including: Establish multiple scene feature templates, each containing a corresponding range of feature parameters; The task feature parameters are matched with the feature parameter range in each scene feature template to obtain the matching degree between the task feature parameters and each scene feature template. The current task scenario is determined based on the scenario feature template with the highest matching degree and a matching degree exceeding the preset matching degree threshold.

5. The method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters according to claim 4, characterized in that, The task feature parameters are matched with the feature parameter range in each scene feature template to obtain the matching degree between the task feature parameters and each scene feature template, including: Calculate the parameter deviation between each feature parameter in the task feature parameters and the boundary value of the corresponding feature parameter range in each scene feature template; Based on the parameter deviation, the correlation degree between each feature parameter and each scene feature template is calculated; The matching degree between the task feature parameters and each scene feature template is calculated by weighted summation of the correlation between all feature parameters and the same scene feature template.

6. The method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters according to claim 1, characterized in that, The feedback controller includes at least one of a feedback control module, a feedforward compensation module, and a filter module. The feedback control module includes one or more combinations of proportional control submodule, integral control submodule and derivative control submodule; The gain parameters include at least one of the following: position loop proportional gain, torque loop proportional gain, integral gain, derivative gain, feedforward gain, and filter parameters.

7. The method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters according to claim 1, characterized in that, Based on the stiffness adjustment model, with the optimization objective of satisfying the target joint stiffness and achieving optimal system stability, the optimal parameter combination of the feedback controller corresponding to the target joint stiffness is determined, including: The control parameter combination corresponding to the target joint stiffness is determined by the stiffness adjustment model. With the goal of satisfying the target joint stiffness and achieving optimal system stability, the optimal solution is searched within the feasible region of the feedback controller parameters based on the aforementioned combination of control parameters to obtain the optimal parameter combination.

8. The method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters according to claim 1, characterized in that, The method further includes: Obtain the joint torque and joint position of the flexible robotic arm; Based on the joint torque, the change in joint torque caused by external disturbance is calculated. Based on the joint position, the actual change in joint angular displacement is calculated; The actual joint stiffness is calculated by dividing the change in joint torque by the change in joint angular displacement. Determine the stiffness deviation between the actual joint stiffness and the target joint stiffness; If the stiffness deviation is higher than the set deviation threshold, a parameter grading correction strategy is triggered and executed.

9. The method for stiffness control of a flexible robotic arm based on dynamic adjustment of feedback control parameters according to claim 8, characterized in that, Trigger and execute the parameter hierarchical correction strategy, including: Based on the actual joint stiffness, joint position, and joint torque, multiple deviation evaluation index values ​​are determined, and the stiffness deviation is calculated based on the multiple deviation evaluation index values. Determine the number of abnormal joints in the flexible robotic arm whose stiffness deviation exceeds a set deviation threshold; If the number of abnormal joints is less than or equal to a set threshold, a first-level correction strategy is triggered and executed. The first-level correction strategy is to adjust at least some of the control parameters in the current optimal parameter combination online. If the number of abnormal joints exceeds a set threshold, a secondary correction strategy is triggered and executed. The secondary correction strategy involves adjusting the stiffness adjustment model and regenerating the optimal parameter combination based on the adjusted stiffness adjustment model.

Images