Torque adaptive adjustment method of light robot servo joint

Abstract

The present application relates to the technical field of robot joint torque, in particular to a torque adaptive adjustment method of a light robot servo joint. Through step-by-step calculation of actual output torque, expected torque, torque error, load disturbance torque estimation value, compensation torque and reference torque, and in combination with the physical mapping relationship between target motor current, driving voltage, actual motor current and actual output torque, the problems of inability to estimate and compensate load disturbance in real time, difficulty in adaptive adjustment of control parameters with working condition changes, and discontinuity of torque control link between the electrical layer and the torque layer are solved, and the torque control link is constructed to be complete by the compensation torque acting on the reference torque, so that the tracking accuracy of the actual output torque to the reference torque is improved.

Description

Technical Field

[0001] This invention relates to the field of robot joint torque technology, and specifically to a method for adaptive torque adjustment of a lightweight robot servo joint. Background Technology

[0002] With the widespread application of lightweight robots in collaborative manufacturing, service robots, and precision operations, servo joints, as the core execution unit for motion control, directly affect the dynamic performance and operational safety of the entire machine due to their torque control accuracy. Existing technologies often employ control methods based on fixed parameters or indirect adjustment methods based on position and velocity errors to achieve joint drive. However, in actual operation, due to factors such as load changes, frictional nonlinearity, and joint posture changes, the disturbances experienced by the joint exhibit significant time-varying and uncertainties, resulting in a continuous deviation between the desired torque and the actual output torque. Furthermore, existing solutions typically lack effective modeling and compensation mechanisms for load disturbance torque, and the transmission relationship from the reference torque to the drive voltage and then to the actual output torque in the control link is incomplete, making it difficult to form a stable torque closed-loop control. This makes the system prone to problems such as increased torque tracking error and response lag under complex working conditions. Summary of the Invention

[0003] To address the aforementioned shortcomings of existing technologies, this invention provides a torque adaptive adjustment method for lightweight robot servo joints. This method effectively solves the problems in existing technologies where, under the influence of load variations and nonlinear friction, it is difficult to estimate and compensate for load disturbance torque in robot servo joints, and the torque control link is incomplete, resulting in the actual output torque being difficult to accurately track the reference torque.

[0004] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a method for adaptive torque adjustment of a lightweight robot servo joint, comprising at least: Step S1: Obtain joint angle, joint angular velocity, and actual motor current, and construct a state feature vector as input; Step S2: Calculate the actual output torque based on the actual motor current, calculate the desired torque based on the target angular acceleration, joint angular velocity and joint angle, and construct the torque error based on the two. Step S3: Estimate the load disturbance torque based on the torque error, joint angular velocity, and adaptive parameters; Step S4: Generate a compensation torque based on the estimated load disturbance torque, and combine it with the desired torque to obtain a reference torque; Step S5: Update the adaptive parameters based on the torque error and joint angular velocity, and generate the target motor current based on the reference torque; Step S6: Generate a driving voltage based on the deviation between the target motor current and the actual motor current, and adjust the actual motor current through the driving voltage so that the actual output torque tracks the reference torque, thereby achieving torque adaptive adjustment.

[0005] Furthermore, the method for determining the actual output torque is as follows: Based on the linear proportional relationship between the actual motor current and the motor torque constant, the actual motor current is multiplied by the motor torque constant to obtain the actual output torque, where: Considering the noise interference during the current sampling process, a first-order low-pass filter is used to filter the actual motor current.

[0006] Furthermore, the desired torque is obtained by superimposing three parts, including: The product of the equivalent moment of inertia and the target angular acceleration is used to characterize the inertial torque; the product of the damping coefficient and the joint angular velocity is used to characterize the velocity-related drag; and the gravitational torque function calculated based on the joint angle is used to characterize the effect of gravity. The three parts are summed to obtain the desired torque.

[0007] Furthermore, the solution for the gravitational torque function is as follows: Determine the distance from the center of mass to the joint axis; Collect equivalent mass and gravitational acceleration; The gravitational torque function is defined in conjunction with the joint angle.

[0008] The estimated load disturbance torque consists of three parts: The product of torque error and error weight adaptive parameter is used to characterize the overall disturbance; The product of the joint angular velocity and the velocity weight adaptive parameter is used to characterize velocity-related perturbations; The product of the sign function of the joint angular velocity and the adaptive parameter of the direction weight is used to characterize the direction-related perturbation; The three parts are superimposed to obtain the estimated value of the load disturbance torque.

[0009] Furthermore, by reversing the direction of the estimated load disturbance torque, a compensation torque is generated, and the compensation torque is superimposed with the expected torque to obtain a reference torque.

[0010] Furthermore, the method for updating the adaptive parameters is as follows: The error weight adaptive parameter is updated incrementally based on the torque error; The speed weight adaptive parameter is updated based on the incremental product of torque error and joint angular velocity; The adaptive parameters for directional weights are incrementally updated based on the product of the torque error and the sign function of the joint angular velocity, so that each adaptive parameter is dynamically adjusted according to changes in the operating state, wherein: In the process of updating the adaptive parameters, a learning rate is introduced to control the magnitude of parameter changes.

[0011] The method for adaptive torque adjustment is as follows: The reference torque is proportionally converted according to the motor torque constant to obtain the target motor current; The current error is constructed based on the difference between the target motor current and the actual motor current; The driving voltage is generated based on the current error and the historical cumulative value, thereby regulating the actual motor current change process. After the driving voltage is applied to the motor, the actual motor current changes with time and gradually approaches the target motor current. Based on the motor torque constant, the actual motor current is converted into the actual output torque, thereby realizing the tracking and adjustment of the actual output torque on the reference torque.

[0012] A torque adaptive adjustment system, comprising: The data acquisition module acquires joint angles, joint angular velocities, and actual motor current, and constructs a state feature vector as input. The torque error analysis module calculates the actual output torque based on the actual motor current, calculates the desired torque based on the target angular acceleration, joint angular velocity, and joint angle, and constructs the torque error based on both. The load disturbance analysis module estimates the load disturbance torque based on the torque error, joint angular velocity, and adaptive parameters. The reference torque calculation module generates a compensation torque based on the estimated load disturbance torque and combines it with the desired torque to obtain a reference torque. The torque tracking control module updates adaptive parameters based on the torque error and joint angular velocity, and generates a target motor current based on the reference torque; it generates a drive voltage based on the deviation between the target motor current and the actual motor current, and adjusts the actual motor current through the drive voltage so that the actual output torque tracks the reference torque, thereby achieving adaptive torque adjustment.

[0013] The technical solution provided by this invention has the following advantages compared with the known prior art: By calculating the actual output torque, desired torque, torque error, estimated load disturbance torque, compensation torque, and reference torque step by step, and combining the physical mapping relationship between the target motor current, drive voltage, actual motor current, and actual output torque, this method solves the problems of load disturbance not being estimated and compensated in real time, control parameters not being able to adaptively adjust with changes in operating conditions, and the discontinuity of the torque control link between the electrical layer and the torque layer. Under the influence of load changes and nonlinear friction, adaptive parameter updates can achieve accurate estimation of load disturbance torque. The compensation torque acts on the reference torque to construct a complete torque closed-loop control link, improving the tracking accuracy of the actual output torque to the reference torque. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0015] Figure 1 This is a flowchart of the overall method of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present 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 the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0017] The present invention will be further described below with reference to embodiments.

[0018] Example 1 (see Figure 1 A method for adaptive torque adjustment of a lightweight robot servo joint, comprising at least: Step 1: Real-time acquisition of joint angles using a joint encoder This angle data reflects the current spatial attitude of the joint. Subsequently, the joint angular velocity is obtained either by performing discrete difference calculations on the angle signal or directly through a velocity sensor. This is used to describe the changing trend of the joint's motion state. Simultaneously, the actual motor current of the drive motor is acquired through a current sampling module. This current value directly affects the motor's output torque.

[0019] The above data are acquired synchronously within the same sampling period to avoid data misalignment caused by sampling delay. In engineering implementation, sampling is triggered by a unified clock interrupt, or a synchronous sampling chip is used to complete data acquisition. After acquisition, the acquired joint angle, joint angular velocity, actual motor current, and drive voltage are constructed into a state feature vector according to a predefined order. Subsequent steps are all calculated based on the extracted corresponding components from the state feature vector.

[0020] In addition, before constructing the state feature vector, each signal can be preprocessed, such as low-pass filtering of the actual motor current and noise reduction of the joint angular velocity, thereby improving the stability of subsequent calculations.

[0021] Step 2: Based on the electromagnetic characteristics of the motor, convert the actual motor current into the actual output torque. For servo motors, there is an approximately linear relationship between their output torque and current, which is determined by the motor torque constant. describe. These are inherent parameters of the motor, which can be obtained through factory calibration or experimental measurement.

[0022] First, read the actual motor current obtained in step one, and then calculate the corresponding actual output torque through multiplication. Considering the potential noise interference during current sampling, I(t) can be filtered before calculation, for example, using a first-order low-pass filter:

[0023] in, This represents the filtered current value. Representing the filter coefficients allows us to use the filtered result. Replace the previous It participates in torque calculation, improving the stability of the actual output torque.

[0024] Step 3: Calculate the desired torque required by the joint under the current control objective. Obtain the target angular acceleration from the upper-level control system or trajectory planning module. This parameter reflects the joint's desired dynamic response capability. Subsequently, combined with the current joint angular velocity, a damping term is introduced to describe the torque loss caused by friction or resistance during motion.

[0025] The desired torque is expressed as:

[0026] in, This represents the desired torque, which is the theoretical driving torque required to be applied to the joint at the current time t to achieve the target motion state. It represents the equivalent moment of inertia, which is the equivalent inertia of the joint body mass, load mass, and transmission structure converted onto the joint axis. Indicates joint angular velocity, Represents the damping torque term. This represents the damping coefficient, which describes the velocity-related resistance generated during joint movement due to factors such as friction, lubrication, and air resistance. This represents the target angular acceleration, which is the derivative of the target angular velocity with respect to time.

[0027] By superimposing the inertial term, damping moment term, and gravitational moment function, the complete desired moment is obtained. In engineering implementation, this calculation is typically performed in real time within the control cycle to ensure dynamic updates of the control objective.

[0028] Simultaneously, a gravitational torque function is introduced to compensate for the gravitational effects caused by changes in joint posture. This function is typically expressed as:

[0029] in, Let g represent the gravitational torque function, g be the gravitational acceleration, l be the distance from the center of mass to the joint axis, the lever arm length, and m be the equivalent mass.

[0030] Step 4: Obtain the torque error by calculating the difference between the expected torque and the actual output torque. (That is, the torque error is obtained by subtracting the actual output torque from the expected torque). This error reflects the degree of deviation of the current control system in torque output. In actual implementation, the control system obtains the expected torque from step three and the actual output torque from step two, and performs differential calculation to obtain the torque error. Since sudden interference or sampling anomalies may occur during system operation, to prevent excessive errors from affecting subsequent calculations, the torque error can be limited, for example: , among them In addition to setting a preset upper limit for error, the error can also be smoothed to reduce high-frequency fluctuations. The casing can be set according to the actual situation, which will not be described in detail in this embodiment.

[0031] In this embodiment, since the robot joints may be affected by external load changes (such as grasping objects, posture changes) and internal friction during actual operation, these disturbances are difficult to obtain through direct measurement and need to be estimated through modeling. Therefore: Establish a perturbation model:

[0032] in, This represents the estimated load disturbance torque, which is the estimated equivalent torque of the combined external disturbance and internal unmodeled disturbance experienced by the system at the current time t. This represents the adaptive error weighting parameter, which is dimensionless and dynamically updated over time. This represents the adaptive velocity weighting parameter, which adjusts the contribution of velocity to the disturbance estimation. This represents the adaptive parameter for direction weighting, which adjusts the amplitude of direction-related perturbations. This represents a sign function that describes a disturbance related to the direction of motion but independent of the magnitude of velocity. In this embodiment, the output is 1 when the joint rotates in the forward direction, 0 when the joint is stationary, and -1 when the joint rotates in the reverse direction. Specifically, for ease of understanding, the sign function is expressed as a formula:

[0033] By constructing a load disturbance torque estimation model consisting of a torque error term, a joint angular velocity term, and a sign function term, this model addresses the challenges of directly measuring and uniformly modeling multi-source disturbances such as external load variations, viscous friction, and Coulomb friction during actual robot servo joint operation. The torque error term provides feedback representation of the overall unmodeled disturbances, the joint angular velocity term introduces the ability to characterize continuously varying damped disturbances, and the sign function term introduces the ability to model direction-dependent discontinuous frictional disturbances. Furthermore, by combining an adaptive parameter dynamic adjustment mechanism, the disturbance estimation results can be adjusted in real time under different operating conditions.

[0034] Furthermore, the identified load disturbance torque during robot servo joint operation is not effectively utilized by the control system, thus still directly acting on the joint output and causing persistent torque deviation. That is, even with disturbance estimation but without compensation, the difference between the expected torque and the actual output torque cannot be eliminated, resulting in increased tracking error, lag in dynamic response, and instability with oscillations under sudden load changes. Therefore, constructing a compensation torque... The estimated load disturbance torque is introduced into the control link in the opposite direction to actively cancel out external load disturbances and internal unmodeled disturbances, thus solving the response lag caused by passive correction relying solely on error feedback.

[0035] Step 5: Having obtained the desired torque and compensation torque, the control system lacks a unified final control objective that simultaneously reflects both the target dynamic requirements and the load disturbance compensation results. If the desired torque is still directly used as the control input, the compensation effect formed by the estimated load disturbance torque cannot be effectively integrated into the control link, leading to a disconnect between disturbance compensation and dynamic control. This results in torque deviation and tracking error still occurring when the load changes. Therefore: The calculated desired torque is superimposed with the generated compensation torque to form the final reference torque used for drive control. This process essentially involves introducing corrections for load disturbances based on the original dynamic model output, making the control objective more closely resemble actual operational requirements.

[0036] In one specific embodiment, the control system synchronizes and aligns the two input signals to ensure their time consistency before performing an addition operation. Since the compensation torque is already the inverse expression of the load disturbance torque, this superposition process is equivalent to subtracting the disturbance effect from the original control objective. To ensure system stability, a limiting mechanism can be added in this step, such as limiting the reference torque to the maximum allowable output torque of the system, to prevent the reference torque from becoming too large under abnormal conditions.

[0037] Furthermore, if the adaptive parameters in the load disturbance torque estimation model remain fixed under different operating conditions, they will be unable to adapt to the changes in disturbance characteristics caused by load variations, friction variations, and motion state variations. That is, during the operation of the robot's servo joint, the external load and internal friction characteristics exhibit significant time-varying and nonlinear features. If the adaptive parameters are not dynamically adjusted according to changes in torque error and joint angular velocity, the estimated load disturbance torque will gradually deviate from the actual disturbance, thus affecting the compensation effect. Therefore: Adaptive parameter updates in the load disturbance torque estimation model enable the system to automatically adjust model accuracy based on operating conditions. Specifically, an error-driven gradient update mechanism is employed.

[0038]

[0039]

[0040] in, This represents the updated adaptive error weight parameters. This represents the updated adaptive velocity weight parameters. Indicates the adaptive parameter of direction weight, torque error As an error feedback signal, it reflects the degree of deviation between the current estimate and the actual value. The learning rate is used to control the magnitude of parameter changes. Its value needs to be set according to the system requirements. This embodiment does not limit it. (It should be noted that the torque error and joint angular velocity are normalized with the reference value, and the update term becomes dimensionless so that the same dimensionless learning rate can be used.)

[0041] Step Six: Reference Torque As a torque-level control quantity, it cannot be directly applied to the motor drive system. That is, the actual control quantity executed by the motor driver is motor current, not torque. If there is a lack of an effective conversion relationship from the reference torque to the target motor current, the control system cannot transmit the upper-level torque control target to the execution layer, leading to control link interruption or the inability to achieve the control target. Therefore, by using the reference torque... The target motor current is calculated by comparing the actual motor current with the motor torque constant. However, the target motor current is difficult to track precisely during actual operation due to factors such as motor dynamics, power supply fluctuations, and load changes. In other words, without a closed-loop control mechanism, a continuous deviation will exist between the actual motor current I(t) and the target motor current, resulting in an impact on the actual output torque. Deviation from the reference torque ultimately affects the accuracy and stability of the entire control system. Therefore, the current error is first obtained by subtracting the actual motor current from the target motor current. (Reflecting the current motor current tracking error), this current error is introduced into the PID controller, and a proportional-integral controller is used to adjust the error. The proportional term is used to quickly respond to error changes, and the integral term is used to eliminate steady-state error. Therefore:

[0042] in, This represents the drive voltage, a control voltage signal output by the current controller and applied to the motor drive at the current time t. It directly acts on the motor driver to regulate the actual motor current. This represents the proportional gain coefficient, used to adjust the instantaneous response strength to current errors. This represents the integral gain coefficient, used to adjust the degree of influence of the error integral term on the control output. The time integral term representing the current error, from the initial time 0 to the current time t, is the current error. The cumulative value, This represents the integration variable, a time placeholder used in integration operations, distinguished from the current time t, and ultimately outputs the driving voltage. Specifically, in the driving voltage Under this influence, the motor current changes according to the electrical dynamic equation, causing the actual motor current to gradually converge towards the target motor current. In actual implementation, the driver controls the PWM duty cycle based on the input voltage signal, thereby changing the equivalent voltage applied to the motor windings and regulating the rate of current change. As the actual motor current changes, the motor torque characteristics are adjusted accordingly. The actual output torque of the motor also changes synchronously and gradually approaches the reference torque. The step-by-step transmission from voltage regulation to current regulation and then to torque regulation enables the control system to maintain the stability of torque output even under load disturbances.

[0043] A torque adaptive adjustment system, comprising: The data acquisition module acquires joint angles, joint angular velocities, and actual motor current, and constructs a state feature vector as input. The torque error analysis module calculates the actual output torque based on the actual motor current, calculates the desired torque based on the target angular acceleration, joint angular velocity, and joint angle, and constructs the torque error based on both. The load disturbance analysis module estimates the load disturbance torque based on the torque error, joint angular velocity, and adaptive parameters. The reference torque calculation module generates a compensation torque based on the estimated load disturbance torque and combines it with the desired torque to obtain a reference torque. The torque tracking control module updates adaptive parameters based on the torque error and joint angular velocity, and generates a target motor current based on the reference torque; it generates a drive voltage based on the deviation between the target motor current and the actual motor current, and adjusts the actual motor current through the drive voltage so that the actual output torque tracks the reference torque, thereby achieving adaptive torque adjustment.

[0044] Furthermore, if the aforementioned function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0045] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0046] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0047] Furthermore, in order to provide a concise description of exemplary embodiments, not all features of actual embodiments (i.e., those features that are not relevant to the best mode of carrying out the invention as currently considered, or those features that are not relevant to implementing the invention) may be omitted.

[0048] It should be understood that numerous specific implementation decisions can be made during the development of any practical implementation, such as in any engineering or design project. Such development efforts may be complex and time-consuming, but for those skilled in the art who benefit from this disclosure, the development effort will be a routine work of design, manufacturing, and production without requiring much experimentation.

[0049] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. 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. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for adaptive torque adjustment of a servo joint in a lightweight robot, characterized in that, include: Step S1: Obtain joint angle, joint angular velocity, and actual motor current, and construct a state feature vector as input; Step S2: Calculate the actual output torque based on the actual motor current, calculate the desired torque based on the target angular acceleration, joint angular velocity and joint angle, and construct the torque error based on the two. Step S3: Estimate the load disturbance torque based on the torque error, joint angular velocity, and adaptive parameters; Step S4: Generate a compensation torque based on the estimated load disturbance torque, and combine it with the desired torque to obtain a reference torque; Step S5: Update the adaptive parameters based on the torque error and joint angular velocity, and generate the target motor current based on the reference torque; Step S6: Generate a driving voltage based on the deviation between the target motor current and the actual motor current, and adjust the actual motor current through the driving voltage so that the actual output torque tracks the reference torque, thereby achieving torque adaptive adjustment.

2. The torque adaptive adjustment method for a lightweight robot servo joint according to claim 1, characterized in that, The method for determining the actual output torque is as follows: Based on the linear proportional relationship between the actual motor current and the motor torque constant, the actual motor current is multiplied by the motor torque constant to obtain the actual output torque, where: Considering the noise interference during the current sampling process, a first-order low-pass filter is used to filter the actual motor current.

3. The torque adaptive adjustment method for a lightweight robot servo joint according to claim 1, characterized in that, The desired torque is obtained by superimposing three parts, including: The product of the equivalent moment of inertia and the target angular acceleration is used to characterize the inertial torque; the product of the damping coefficient and the joint angular velocity is used to characterize the velocity-related drag; and the gravitational torque function calculated based on the joint angle is used to characterize the effect of gravity. The three parts are summed to obtain the desired torque.

4. The torque adaptive adjustment method for a lightweight robot servo joint according to claim 3, characterized in that, The solution to the gravitational torque function is as follows: Determine the distance from the center of mass to the joint axis; Collect equivalent mass and gravitational acceleration; The gravitational torque function is defined in conjunction with the joint angle.

5. The torque adaptive adjustment method for a lightweight robot servo joint according to claim 1, characterized in that, The estimated load disturbance torque consists of three parts: The product of torque error and error weight adaptive parameter is used to characterize the overall disturbance; The product of the joint angular velocity and the velocity weight adaptive parameter is used to characterize velocity-related perturbations; The product of the sign function of the joint angular velocity and the adaptive parameter of the direction weight is used to characterize the direction-related perturbation; The three parts are superimposed to obtain the estimated value of the load disturbance torque.

6. The torque adaptive adjustment method for a lightweight robot servo joint according to claim 1, characterized in that, By reversing the direction of the estimated load disturbance torque, a compensation torque is generated, and the compensation torque is superimposed with the expected torque to obtain a reference torque.

7. The torque adaptive adjustment method for a lightweight robot servo joint according to claim 1, characterized in that, The method for updating the adaptive parameters is as follows: The error weight adaptive parameter is updated incrementally based on the torque error; The speed weight adaptive parameter is updated based on the incremental product of torque error and joint angular velocity; The adaptive parameters for directional weights are incrementally updated based on the product of the torque error and the sign function of the joint angular velocity, so that each adaptive parameter is dynamically adjusted according to changes in the operating state, wherein: In the process of updating the adaptive parameters, a learning rate is introduced to control the magnitude of parameter changes.

8. The torque adaptive adjustment method for a lightweight robot servo joint according to claim 1, characterized in that, The method for adaptive torque adjustment is as follows: The reference torque is proportionally converted according to the motor torque constant to obtain the target motor current; The current error is constructed based on the difference between the target motor current and the actual motor current; The driving voltage is generated based on the current error and the historical cumulative value, thereby regulating the actual motor current change process. After the driving voltage is applied to the motor, the actual motor current changes with time and gradually approaches the target motor current. Based on the motor torque constant, the actual motor current is converted into the actual output torque, thereby realizing the tracking and adjustment of the actual output torque on the reference torque.

9. A torque adaptive adjustment system, applied to the method according to any one of claims 1 to 8, characterized in that, include: The data acquisition module acquires joint angles, joint angular velocities, and actual motor current, and constructs a state feature vector as input. The torque error analysis module calculates the actual output torque based on the actual motor current, calculates the desired torque based on the target angular acceleration, joint angular velocity, and joint angle, and constructs the torque error based on both. The load disturbance analysis module estimates the load disturbance torque based on the torque error, joint angular velocity, and adaptive parameters. The reference torque calculation module generates a compensation torque based on the estimated load disturbance torque and combines it with the desired torque to obtain a reference torque. The torque tracking control module updates adaptive parameters based on the torque error and joint angular velocity, and generates a target motor current based on the reference torque. A driving voltage is generated based on the deviation between the target motor current and the actual motor current, and the actual motor current is adjusted by the driving voltage so that the actual output torque tracks the reference torque, thereby achieving adaptive torque adjustment.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 8.

Images