Friction compensation method for robot joint, computer readable storage medium and robot system
By collecting robot end-effector displacement data and combining it with inverse kinematics, the direction of static friction compensation torque was determined, which solved the control blind zone of high static friction joints during zero-speed start-up, achieved precise friction compensation, and improved the start-up compliance and operation accuracy of robot joints.
Patent Information
- Application Number
- CN202610254768.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-03
- Publication Date
- 2026-04-07
AI Technical Summary
Existing friction compensation technology has a control blind zone during the zero-speed start-up of high static friction joints, and cannot detect the operator's dragging intention, resulting in dead zones and a feeling of resistance during start-up. It is also prone to sudden jumps or oscillations, affecting the smoothness of human-machine interaction and the accuracy of operation.
By acquiring displacement data from the robot's end effector, the motion trend of the joint to be compensated is calculated using inverse kinematics of the robot, the direction of the static friction compensation torque is determined, and the output static friction compensation torque is controlled to assist joint activation, thus eliminating dependence on local sensors and achieving precise pre-compensation.
It eliminates the dead zone and resistance during startup in zero-force drag mode, and solves the problem that high static friction joints cannot sense the direction of movement intention when stationary due to the failure of local sensors, thus improving the smoothness of startup and the accuracy of operation.
Smart Images
Figure CN121798633A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of robot motion control, and in particular to a friction compensation method for a robot joint, a computer readable storage medium and a robot system. BACKGROUND
[0002] In the application scenarios of medical surgery robots or collaborative robots, an operator often needs to adjust the pose of the robot by manually dragging the end of the robot arm (i.e., zero force dragging or teaching mode). In order to meet the positioning requirements of high load and high rigidity, the joints of such robots (especially linear guide or screw transmission joints) usually adopt a high pre-tightening assembly structure, which results in a significant static friction force, and the amplitude of the static friction force is often much larger than that of the dynamic friction force.
[0003] However, the existing friction compensation technology mostly relies on the speed signal fed back by the local sensor of the joint (such as an encoder installed at the end of the motor or the output end of the joint) to implement control. This control strategy based on local speed feedback has an inherent defect when dealing with the zero-speed start process of a high static friction joint. When the operator pushes the end of the robot arm to try to start the joint in a static state, the joint has not yet produced displacement macroscopically, resulting in zero or only invalid quantization noise in the readings of the local sensor, due to the huge static friction force inside the joint.
[0004] Since there is no effective speed signal at the local joint at this time, the control system cannot perceive the dragging intention of the operator, and cannot determine the direction of application of the static friction compensation torque (i.e., whether to assist in the positive direction or the negative direction). This results in a "control blind area" at the start instant, and the system cannot actively overcome the static friction force. The operator must exert a large enough external force to completely break through the static friction threshold, and after the joint produces a macroscopic speed, the system can intervene in dynamic friction compensation. This phenomenon makes the robot exhibit a significant "dead zone" and a heavy resistance feeling when starting, and is extremely prone to sudden jumps or oscillations at the moment of breaking through the static friction, which seriously affects the compliance and operation accuracy of human-machine interaction. SUMMARY
[0005] An object of the present application is to provide a friction compensation method for a robot joint, a computer readable storage medium and a robot system, at least to solve the above problems.
[0006] To achieve the above object, some embodiments of the present application provide a friction compensation method for a robot joint, comprising:
[0007] obtaining displacement data of the end of the robot;
[0008] determining whether the joint to be compensated is in a static state;
[0009] When the joint to be compensated is in a static state, according to the displacement data, the motion trend of the joint to be compensated is calculated by using the inverse kinematics of the robot;
[0010] According to the direction of the motion trend, the direction of the static friction compensation torque applied to the joint to be compensated is determined;
[0011] The static friction compensation torque is controlled to assist the joint to be compensated to start.
[0012] Some embodiments of the present application provide a computer readable storage medium, comprising a memory and a processor, the memory stores a computer program, and the processor implements the friction compensation method for the robot joint provided by the foregoing embodiments when executing the computer program.
[0013] Some embodiments of the present application provide a robot system, comprising:
[0014] A robot body comprising an end effector and at least one joint to be compensated;
[0015] A position detection device arranged on the end effector or the load side thereof, for collecting displacement data;
[0016] A controller in communication connection with the robot body and the position detection device; the controller is configured to execute the friction compensation method for the robot joint provided by the foregoing embodiments.
[0017] Compared with the related art, in the scheme provided by the embodiments of the present application, by collecting the displacement data of the end of the robot and combining the inverse kinematics, the theoretical motion trend of the joint to be compensated can be analyzed when the joint to be compensated has not yet produced macro displacement, so that the correct application direction of the static friction compensation torque is determined, precise pre-compensation with direction is realized, the starting dead zone and the resistance feeling in the zero force drag mode are eliminated, and the technical problem that the high static friction joint cannot perceive the motion intention direction due to the failure of the local sensor in the static state is solved. BRIEF DESCRIPTION OF DRAWINGS
[0018] One or more embodiments are exemplarily illustrated by the pictures in the drawings corresponding thereto, which do not constitute a limitation on the embodiments, and the elements with the same reference numerals in the drawings represent similar elements, unless otherwise specified, and the drawings do not constitute a proportional limit.
[0019] Figure 1 is a flowchart of the friction compensation method provided by the embodiments of the present disclosure;
[0020] Figure 2 is a flowchart of step 1 in the friction compensation method provided by the embodiments of the present disclosure;
[0021] Figure 3FIG. 3 is a flowchart illustrating step 2 of the friction compensation method according to an embodiment of the present disclosure;
[0022] Figure 4 FIG. 4 is a flowchart illustrating step 3 of the friction compensation method according to an embodiment of the present disclosure;
[0023] Figure 5 FIG. 5 is a flowchart illustrating step 33 of the friction compensation method according to an embodiment of the present disclosure. DETAILED DESCRIPTION
[0024] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.
[0025] The terms "first", "second", and the like in the description and claims of the present disclosure and the above drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion.
[0026] In the embodiments of the present disclosure, the terms "upper", "lower", "inner", "middle", "outer", "front", "back", and the like indicate the orientation or positional relationship shown in the drawings. These terms are mainly used to better describe the present disclosure and its embodiments, and are not intended to limit the indicated devices, elements or components to have a specific orientation, or to be constructed and operated in a specific orientation. In addition, in addition to indicating the orientation or positional relationship, the above-mentioned terms can also be used to indicate other meanings, for example, the term "upper" can also be used to indicate a certain attachment relationship or connection relationship in some cases. For those of ordinary skill in the art, the specific meanings of these terms in the embodiments of the present disclosure can be understood according to the specific circumstances.
[0027] In addition, the terms "set", "connected", "fixed" should be broadly understood. For example, "connected" can be fixed connection, detachable connection, or integral structure; can be mechanical connection, or electrical connection; can be direct connection, or indirect connection through an intermediate medium, or internal communication between two devices, elements or components. For those of ordinary skill in the art, the specific meanings of the above terms in the embodiments of the present disclosure can be understood according to the specific circumstances.
[0028] The term "plurality" means two or more, unless otherwise specified.
[0029] In the embodiments of the present disclosure, the character " / " represents an "or" relationship between the preceding and following objects. For example, A / B represents: A or B.
[0030] The term "and / or" is a description of the association relationship between objects, which means that there can be three relationships. For example, A and / or B means: A or B, or, A and B, three relationships.
[0031] It should be noted that, in the case of no conflict, the embodiments in the embodiments of the present disclosure and the features in the embodiments can be combined with each other.
[0032] In combination with Figures 1 to 5 As shown in the drawings, the friction compensation method for robot joint provided by the embodiments of the present disclosure comprises the following steps:
[0033] S1, acquiring displacement data of the robot end;
[0034] S2, judging whether the joint to be compensated is in a static state;
[0035] S3, when the joint to be compensated is in a static state, calculating the motion trend of the joint to be compensated according to the displacement data by using the inverse kinematics solution of the robot;
[0036] S4, determining the direction of the static friction compensation torque applied to the joint to be compensated according to the direction of the motion trend;
[0037] S5, controlling the output of the static friction compensation torque to assist the start of the joint to be compensated.
[0038] By using the friction compensation method provided by the embodiments of the present disclosure, the theoretical motion trend of the joint to be compensated can be analyzed when the joint to be compensated has not yet produced macro displacement by collecting the displacement data of the robot end and combining the inverse kinematics solution, so as to determine the correct application direction of the static friction compensation torque, realize accurate pre-compensation with direction, eliminate the start dead zone and resistance in the zero force drag mode, and solve the technical problem that the high static friction joint cannot sense the motion intention direction in the static state due to the failure of the local sensor.
[0039] In some embodiments, the motion trend of the joint to be compensated is not dependent on the joint torque sensor or the joint side displacement sensor in the calculation process, but is only based on the robot end displacement data and the robot kinematics model for inference. Get rid of the dependence on the local sensor of the joint, effectively solve the problem that the local sensor cannot detect the small trend due to the static friction lock or the transmission chain gap under the high static friction working condition, and the signal distortion.
[0040] In some optional embodiments, the calculation of the motion trend is performed in the micro-motion working condition where the robot end displacement is lower than a preset displacement amplitude threshold. Strictly limiting the trend derivation based on the inverse solution of the end to the micro-motion working condition before starting ensures that the algorithm only intervenes in the initial stage where static friction needs to be overcome, avoiding interference with subsequent normal macro-motion control of the robot.
[0041] In some optional embodiments, after outputting the static friction compensation torque, the direction of the static friction compensation torque is prohibited from being updated before the real-time feedback speed of the joint to be compensated exceeds a speed threshold. By locking the direction of the static friction compensation in the starting stage, the frequent reversal of the torque direction between positive and negative caused by sensor noise or unintentional tremor of the operator is prevented, ensuring stable torque output during the joint starting process and avoiding oscillation.
[0042] Optionally, according to the displacement data, the motion trend of the joint to be compensated is calculated using the inverse kinematics solution of the robot, including the following steps:
[0043] S31, constructing a Jacobian matrix of the robot;
[0044] S32, calculating the inverse matrix or pseudo-inverse matrix of the Jacobian matrix;
[0045] S33, converting the displacement data into an end velocity vector;
[0046] S34, left multiplying the inverse matrix or pseudo-inverse matrix by the end velocity vector to obtain a joint velocity vector of the joint to be compensated as the motion trend of the joint to be compensated.
[0047] The motion mapping model of the Cartesian space and the joint space is constructed using the Jacobian matrix and its inverse matrix, and the small displacement data of the robot end is quantitatively converted into the joint velocity vector of the joint to be compensated. This embodiment can effectively decouple the motion coupling relationship of the multi-joint robot, accurately calculate the theoretical motion trend of the joint matching the direction of the end drag when the joint has not produced macro-motion, and prevent compensation direction misjudgment caused by structural coupling.
[0048] Optionally, converting the displacement data into an end velocity vector includes the following steps:
[0049] S331, applying a sliding window algorithm to the time series composed of the collected displacement data;
[0050] S332, performing difference operation or linear fitting operation on the displacement data within the sliding window to extract the displacement rate as the end velocity vector.
[0051] By applying a sliding window algorithm and linear fitting or difference operation to the time series of displacement data, the quantization noise and high-frequency disturbance in the original signal are effectively smoothed. This embodiment can extract stable and reliable displacement rate from discrete displacement sampling points as the end speed vector, improve the signal-to-noise ratio of motion trend detection, and avoid false triggering or fluctuation of friction compensation torque caused by signal jitter.
[0052] Optionally, judging whether the joint to be compensated is in a static state comprises the following steps:
[0053] S21, acquiring a real-time feedback speed of the joint to be compensated;
[0054] S22, judging whether the absolute value of the real-time feedback speed is less than a preset speed threshold;
[0055] S23, if yes, determining that the joint to be compensated is in a static state.
[0056] By setting a speed threshold to construct a zero-speed dead zone, the micro-static state and the macro-motion state of the joint to be compensated are effectively distinguished, and it is ensured that the compensation logic based on the motion trend is activated only when the joint is in a start critical working condition, thereby preventing the control logic from conflicting or the system from oscillating due to false triggering of static friction compensation in the macro-motion stage.
[0057] In some optional embodiments, when it is determined that the joint to be compensated is in a static state, the method further comprises judging whether the real-time feedback speed is less than the speed threshold in a plurality of continuous sampling periods. By using a continuous multi-period state confirmation mechanism, transient noise interference in the speed feedback signal is effectively filtered out, and the accuracy and robustness of the determination of the absolute static state of the joint are improved.
[0058] In some optional embodiments, the speed threshold is calculated online according to the mean and variance of the real-time feedback speed. The speed threshold is associated with at least one of the joint load, temperature or use time, and is updated as the corresponding working condition changes. In this way, dynamic adaptive adjustment of the speed threshold is realized, and noise baseline fluctuations of the speed sensor caused by environmental temperature drift, joint load change or mechanical wear are automatically offset, thereby ensuring the reliability of the static state determination under different working conditions.
[0059] In some optional embodiments, the speed threshold is adaptively updated according to the speed noise statistical characteristics of the joint to be compensated in the historical running process. The noise statistical characteristics based on the historical running data automatically update the speed threshold, so that the system can set the optimal determination standard according to the actual signal quality of the current hardware, and balance between effectively suppressing false triggering and maintaining start sensitivity.
[0060] Optionally, after it is determined that the joint to be compensated is in a static state, the method further comprises a de-jittering process, comprising the following steps:
[0061] S24, determining whether the joint to be compensated is in a static friction compensation torque output state when the absolute value of the real-time feedback speed is less than a preset speed threshold value;
[0062] S25, if yes, skipping the current motion trend calculation step and maintaining the output of the current static friction compensation torque.
[0063] By identifying the current compensation state, when the joint to be compensated is in the starting assistance stage, the motion trend direction determination fluctuation caused by signal noise or operation microflicker is shielded, the static friction compensation torque direction is prevented from appearing high-frequency repeated switching, and the torque output of the joint starting process is ensured to be continuous and smooth.
[0064] Optionally, after determining the direction of the static friction compensation torque applied to the joint to be compensated, the amplitude of the static friction compensation torque is calculated, and the calculation formula is:
[0065]
[0066] wherein, is the static friction compensation torque; is the maximum static friction force parameter; is the displacement change rate; is the attenuation coefficient; is the shape coefficient.
[0067] The embodiment uses a mathematical model containing an exponential decay term to calculate the amplitude of the static friction compensation torque, simulating the Stribeck characteristic that the friction force of the physical contact surface nonlinearly decays with the increasing motion trend in the pre-slippage stage. In this way, the static friction compensation torque can be smoothly decreased in the joint starting process, avoiding the torque step impact caused by fixed or sudden compensation value, and ensuring the smooth transition of the joint from the static state to the motion state.
[0068] In some embodiments, the static friction compensation torque continuously changes with the displacement change rate, and is continuous and derivable at the displacement change rate of zero. In this way, the continuity and derivability of the static friction compensation torque near zero speed are ensured, the singularity and mutation of the torque output are eliminated, and the soft feel of the operator at the zero force drag starting instant is improved.
[0069] In some embodiments, the amplitude of the static friction compensation torque is limited within a preset safe torque range, and the safe torque range is set according to the rated torque of the robot joint. By setting the safe boundary of the torque output, the overlarge torque caused by calculation abnormality, signal interference or misoperation is prevented, and the mechanical transmission structure and motor actuator are protected from overload damage.
[0070] Optionally, when the absolute value of the real-time feedback speed is greater than the preset speed threshold value, the dynamic friction compensation mode is switched to; in the dynamic friction compensation mode, the torque compensation is performed by using a dynamic friction model.
[0071] By fitting the dynamic friction model, the actual friction characteristics of the joint in the motion stage can be accurately approximated in a data-driven manner, the prediction accuracy of the dynamic friction torque is improved, and the residual damping and following error of the robot joint in the dragging motion process are reduced.
[0072] Optionally, the dynamic friction model is as follows:
[0073]
[0074] wherein, is a dynamic friction compensation torque; is a coulomb friction force; is a viscous friction coefficient; is a joint angular velocity; is a viscous friction term; is a Stribeck friction term; is a maximum static friction torque; is a Stribeck velocity; is an exponential parameter.
[0075] The embodiment establishes a dynamic friction physical model containing coulomb friction, viscous friction and Stribeck effect, uses an exponential decay term to accurately represent the Stribeck phenomenon that the friction force of the joint to be compensated increases nonlinearly with the speed in the low speed area, can effectively correct the prediction deviation of the traditional linear model in the low speed starting and reversing stage, realize accurate torque compensation in the whole process from micro pre-sliding to macro sliding, and eliminate the nonlinear resistance and sudden jump in the joint motion process.
[0076] In some optional embodiments, the switching between the static friction compensation mode and the dynamic friction compensation mode uses different speed thresholds to form a hysteresis interval. The switching of the static friction and dynamic friction control modes is managed by using the speed hysteresis interval, the frequent mode jumping caused by signal fluctuations at the critical speed point is eliminated, and the stability of the control system is enhanced.
[0077] In some optional embodiments, when the static friction compensation mode and the dynamic friction compensation mode are switched, the compensation torque is smoothly transitioned to avoid sudden changes in the compensation torque. By eliminating the torque step that may be generated at the moment of switching the static and dynamic friction models, the smooth transition of the friction compensation torque in the whole motion process is ensured, and the joint operation is avoided from appearing jerky or impact.
[0078] In some optional embodiments, multiple sets of dynamic friction model parameters are stored for different load or temperature conditions, and the corresponding parameter set is selected for compensation according to the current operating condition. A multi-parameter lookup table mechanism addresses the nonlinear effects of temperature and load on friction characteristics, ensuring that the system can always call upon matching physical model parameters and maintain high-precision compensation effects under complex conditions such as low-temperature startup or variable load.
[0079] Optionally, the parameter update steps for the dynamic friction model include:
[0080] Collect actual output torque and velocity data of the joint under different uniform motion states;
[0081] Construct a loss function, which is used to characterize the sum of squared errors between the actual output torque data and the torque calculated by the dynamic friction model;
[0082] Calculate the gradient of the loss function with respect to the parameters of the kinetic friction model, and iteratively update the parameters of the kinetic friction model in the opposite direction of the gradient.
[0083] By collecting actual working condition data to construct an error loss function and using gradient information to iteratively correct model parameters, it is possible to effectively eliminate model mismatch caused by mechanical manufacturing tolerances, component wear, or changes in lubrication conditions. This ensures that the dynamic friction model always accurately matches the real physical friction characteristics of the joint during the long-term operation of the robot, thereby improving the robustness of friction compensation accuracy.
[0084] In some embodiments, the updating of the dynamic friction model parameters is paused when the robot performs non-uniform motion or acceleration / deceleration. By eliminating unsteady-state data containing inertial force coupling, the online identification of friction model parameters is ensured to be based solely on high-quality uniform motion data, preventing inertial terms during acceleration / deceleration from interfering with the accuracy of friction parameter estimation.
[0085] Optionally, acquiring the displacement data of the robot's end effector includes the following steps:
[0086] S11. Use a high-resolution encoder array located on the load side of the robot's end effector to acquire position signals;
[0087] S12. Perform differential processing and low-pass filtering on the position signal to obtain the displacement data of the robot end effector.
[0088] By utilizing a high-resolution encoder array located on the load side to directly acquire end-effector position signals, the blind spot problem of local sensors in the joint to be compensated being unable to detect minute displacements due to static friction lock-up or transmission backlash is effectively avoided. Combined with differential and low-pass filtering processing, high signal-to-noise ratio displacement data can be extracted from the noisy raw signal, improving the system's sensitivity to the operator's subtle dragging intentions and signal stability.
[0089] In some alternative embodiments, during the real-time control of static friction compensation, friction compensation is achieved solely based on robot end-effector displacement data and joint velocity feedback, without relying on joint torque sensors. This eliminates the constraints of expensive torque sensor hardware during the real-time control phase, enabling high-performance friction compensation using only conventional position and velocity sensors, thus reducing mass production costs.
[0090] In some alternative embodiments, the joint to be compensated includes at least two joints; based on the same end-effector displacement data, the motion trends of at least two joints are simultaneously calculated using inverse kinematics of the robot; and static friction compensation torques in the corresponding directions are output for each of the at least two joints. This embodiment utilizes the decoupling characteristics of inverse kinematics to accurately decompose the single operational intent of the end-effector into multi-joint coordinated motion trend commands, thereby achieving coordinated activation of all joints in a complex robotic arm configuration.
[0091] In some optional embodiments, joints with amplitudes greater than a preset trend threshold are selected as the target joints for actual friction compensation based on the magnitude of each joint component in the joint velocity vector. This embodiment filters out weak signal interference from non-dominant motion joints and focuses on key joints on the main kinematic chain for precise compensation, improving the computational efficiency and energy utilization of the control system.
[0092] This disclosure also provides a robot joint friction compensation device, comprising: a data acquisition module for acquiring displacement data of the robot end effector; a trend calculation module for calculating the motion trend of the joint to be compensated based on the displacement data and using inverse kinematics of the robot when it is determined that the joint to be compensated is in a stationary state; a direction determination module for determining the direction of the static friction compensation torque applied to the joint to be compensated according to the direction of the motion trend; and a torque control module for controlling the output of the static friction compensation torque of the joint to be compensated to assist the joint to be compensated in starting.
[0093] This disclosure also provides a computer-readable storage medium, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the friction compensation method as described in the above embodiments.
[0094] This disclosure also provides a robot system, including: a robot body, comprising an end effector and at least one joint to be compensated; a position detection device disposed on the end effector or its load side for collecting displacement data; and a controller communicatively connected to the robot body and the position detection device; the controller is configured to execute the friction compensation method for robot joints as described in the above embodiments. Specifically, the controller is configured to acquire displacement data collected by the position detection device; when the joint to be compensated is in a stationary state, based on the displacement data, deduce the motion trend of the joint to be compensated through inverse kinematics of the robot; and control the joint to be compensated to output static friction compensation torque according to the direction of the motion trend.
[0095] For example, this embodiment applies to a minimally invasive surgical robot. The robot's joints are not equipped with local torque sensors, but a high-resolution absolute encoder array is installed on the end effector. During control, the system first smooths the time-series position signal acquired by the end effector encoder using a sliding window algorithm and performs differential operations to extract the minute displacement change rate of the end effector. When the local feedback velocity of the joint is lower than a preset zero-velocity dead zone threshold (e.g., 0.001 m / s) for several consecutive sampling periods, the controller determines that the joint is stationary. At this time, the system constructs a real-time Jacobian matrix using the robot's kinematic model, calculates the pseudo-inverse of this matrix, and maps the end effector displacement change rate to the joint space, thereby resolving the theoretical motion trend of the joint (i.e., virtual velocity). Once a significant motion trend is detected, the controller immediately determines the direction of the static friction compensation torque based on the sign of this trend. To simulate the pre-slip characteristics of the physical contact surface, the amplitude of the compensation torque is not a step signal, but rather determined according to the formula... Calculations are performed to ensure that the compensating force decreases smoothly exponentially as the motion trend intensifies. Furthermore, to prevent frequent changes in torque direction due to hand tremors at the starting critical point, the system incorporates anti-shake logic. Once the static friction compensation torque begins to output, even if the direction of any subsequent detected minute trend momentarily reverses, the system will forcibly lock the current torque direction until the joint enters a macroscopic motion state, as long as the joint has not exceeded its dead zone speed.
[0096] For example, the system presets a speed threshold range with hysteresis characteristics. When the real-time feedback speed of the joint exceeds the dynamic friction intervention threshold, the control mode seamlessly switches from static friction pre-compensation to dynamic friction compensation. In dynamic friction mode, the controller employs a composite physical model including Coulomb friction terms, viscous friction terms, and Stribek terms. The output torque precisely counteracts the nonlinear resistance within the joint. Key parameters of this model (such as Coulomb friction) and viscosity coefficient The mechanism is not static but possesses online learning capabilities. Specifically, when the controller detects that a joint is in a uniform motion state, it automatically collects the current actual output torque and velocity data to construct a sum-of-squares loss function representing the model's prediction error. The system calculates the gradient of the loss function with respect to each model parameter and performs small iterative updates in the opposite direction of the gradient. This mechanism enables the robot to automatically adapt to frictional characteristic drift caused by changes in lubricant viscosity or mechanical wear over time, ensuring a low-resistance dragging feel throughout its entire lifespan.
[0097] For example, none of the joints of the robotic arm in this embodiment are equipped with expensive joint torque sensors; smooth dragging relies entirely on end-effector position feedback and joint velocity feedback. When the operator pushes the end of the robotic arm, the resulting displacement vector often involves the coupled motion of multiple joints simultaneously. The controller uses a unified inverse kinematics operation to decompose a Cartesian displacement vector at the end into motion trend components for each of the six joints. Based on the magnitude of each component, the system identifies several target joints (e.g., shoulder and elbow joints) that are dominated by the current operational intent, and ignores the weak signals from other unrelated joints with extremely small amplitudes (e.g., wrist joints) to suppress noise. Subsequently, the controller generates static friction compensation torques for these target joints that match their motion trend directions. This centralized computing architecture based on full-chain inverse kinematics not only solves the problem that local sensors on a single joint cannot detect external dragging forces when static friction is locked, but also achieves coordinated activation of multiple joints through software algorithms, enabling the operator to obtain a consistent and smooth force feedback experience when dragging in any direction.
[0098] The foregoing description and accompanying drawings fully illustrate embodiments of the present disclosure to enable those skilled in the art to practice them. Other embodiments may include structural and other changes. The embodiments represent only possible variations. Individual components and functions are optional unless explicitly required, and the order of operation may vary. Parts and features of some embodiments may be included or substituted for parts and features of other embodiments. Embodiments of the present disclosure are not limited to the structures described above and shown in the accompanying drawings, and various modifications and changes may be made without departing from its scope. The scope of the present disclosure is limited only by the appended claims, and the foregoing embodiments should be considered exemplary and non-limiting.
Claims
1. A friction compensation method for robot joints, characterized in that, include: Acquire displacement data of the robot's end effector; Determine whether the joint to be compensated is in a stationary state. If so, calculate the motion trend of the joint to be compensated using the inverse kinematics of the robot based on the displacement data. Determine the direction of the static friction compensation torque applied to the joint to be compensated based on the direction of the motion trend; Control the output static friction compensation torque to assist the joint to be compensated in starting.
2. The friction compensation method according to claim 1, characterized in that, Based on the displacement data, the motion trend of the joint to be compensated is calculated using inverse kinematics of the robot, including: Construct the Jacobian matrix for the robot; Calculate the inverse or pseudo-inverse of the Jacobian matrix; The displacement data is converted into an end-effector velocity vector, and the end-effector velocity vector is multiplied by the inverse matrix or pseudo-inverse matrix to obtain the joint velocity vector of the joint to be compensated, which serves as the motion trend of the joint to be compensated.
3. The friction compensation method according to claim 1, characterized in that, Converting displacement data into an end-velocity vector includes: A sliding window algorithm is applied to the time series of the collected displacement data. Perform differential or linear fitting operations on the displacement data within a sliding window to extract the rate of change of displacement as the end velocity vector.
4. The friction compensation method according to claim 1, characterized in that, Determining whether the joint to be compensated is in a static state includes: Obtain the real-time feedback velocity of the joint to be compensated; Determine whether the absolute value of the real-time feedback speed is less than a preset speed threshold; If so, the joint to be compensated is determined to be in a static state.
5. The friction compensation method according to claim 1, characterized in that, After determining the direction of the static friction compensation torque applied to the joint to be compensated, the calculation of the amplitude of the static friction compensation torque is also included. The calculation formula is as follows: in, This is the static friction compensation torque; This is the parameter for maximum static friction. The rate of change of displacement; The attenuation coefficient; This is the shape factor.
6. The friction compensation method according to claim 4, characterized in that, When the absolute value of the real-time feedback speed is greater than the preset speed threshold, switch to dynamic friction compensation mode; In the dynamic friction compensation mode, the dynamic friction model is used for torque compensation.
7. The friction compensation method according to claim 6, characterized in that, The calculation formula for the kinetic friction model is: in, This is the torque for compensating for dynamic friction. Coulomb friction; It is the coefficient of viscous friction; Joint angular velocity; This is the viscous friction term; For the Stribeck friction term; This represents the maximum static friction torque. For Strybeck speed; For exponential parameters.
8. The friction compensation method according to claim 1, characterized in that, Obtain the displacement data of the robot's end effector, including: Position signals are acquired using an encoder array located on the load side of the robot's end effector; The position signal is processed by differential processing and low-pass filtering to obtain the displacement data of the robot end effector.
9. A computer-readable storage medium, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the friction compensation method for robot joints as described in any one of claims 1 to 8.
10. A robot system, characterized in that, include: The robot body includes an end effector and at least one joint to be compensated; A position detection device is installed on the end effector or its load side to collect displacement data; The controller is communicatively connected to the robot body and the position detection device; the controller is configured to perform the friction compensation method for robot joints as described in any one of claims 1 to 8.