A method, device and storage medium for setting control parameters of a robot

Abstract

This application relates to a method, apparatus, device, and storage medium for tuning control parameters of a robot, applied to a control unit in a robot control system. The system further includes: a robot and an asynchronous interactive tuning unit, with the robot situated in a preset free space. The method includes: controlling the robot to perform a preset motion within the preset free space; during the robot's execution of the preset motion, detecting the robot through the asynchronous interactive tuning unit to obtain a dynamic reference input; tuning the robot's position-following performance to obtain a first set of tuning data; determining the corresponding spatial proportional gain parameter value for the robot; and determining a second set of tuning data based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input. Using this application can improve the tuning effect of robot control parameters.

Description

Technical Field

[0001] This application relates to the field of parameter tuning technology, and in particular to a method, apparatus, device and storage medium for tuning control parameters of a robot. Background Technology

[0002] Forward Dynamics Compliance Control (FDCC) is a double-loop nested nonlinear system (the inner loop is a virtual robot dynamics simulation, and the outer loop is a real contact force interaction). It is the core solution for robot compliant force-position hybrid control and is widely used in scenarios such as human-robot collaboration and precision assembly.

[0003] Currently, the traditional PID tuning method is commonly used to tune the control parameters of FDCC. However, the PID tuning method is based on the assumption of single-loop linear feedback and attempts to simultaneously satisfy high-precision position following in free space (inner loop dominance) and compliant contact stability in constrained space (outer loop dominance) with a set of fixed gain parameters. These two are inherently contradictory and cannot be balanced, resulting in poor tuning effect of control parameters.

[0004] Therefore, improving the tuning effect of robot control parameters has become an urgent problem to be solved. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, apparatus, device, and storage medium for tuning robot control parameters to address the aforementioned technical problems, thereby improving the tuning effect of robot control parameters.

[0006] In a first aspect, this application provides a method for tuning control parameters of a robot, applied to a control unit in a robot control system. The system further includes: a robot and an asynchronous interactive tuning unit, wherein the robot is situated in a preset free space; the method includes: Control the robot to perform preset movements within a preset free space; During the robot's execution of the preset motion, the robot is detected by the asynchronous interactive tuning unit to obtain the dynamic reference input. The robot's position-following performance was tuned to obtain the first set of tuning data; Determine the spatial proportional gain parameter value for the robot; The second set of tuning data is determined based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input.

[0007] Secondly, this application provides a robot control parameter tuning device, applied to a control unit in a robot control system. The system further includes: a robot, an asynchronous interactive tuning unit, and the robot is located in a preset free space. The device includes: The control module is used to control the robot to perform preset movements in a preset free space; The tuning module is used to detect the robot through an asynchronous interactive tuning unit during the robot's execution of preset movements, obtain the dynamic reference input, tune the robot's position following performance to obtain the first set of tuning data, determine the corresponding spatial proportional gain parameter value of the robot, and determine the second set of tuning data based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input.

[0008] Thirdly, this application provides a computer device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the method described above.

[0009] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described method.

[0010] Fifthly, this application provides a computer program product comprising a computer program that, when executed by a processor, implements the steps of the method described above.

[0011] The aforementioned robot control parameter tuning method, device, equipment, and storage medium control the robot to execute preset movements, collect data using an asynchronous interactive tuning unit, and determine the dynamic threshold (i.e., the dynamic reference input) to provide a quantization boundary for parameter tuning. Then, the position following performance is optimized separately to obtain basic parameters that meet operational requirements (i.e., the first set of tuning data). Finally, the spatial proportional gain parameter value is combined to obtain the optimal parameters that balance position accuracy and contact stability (i.e., the second set of tuning data). Compared to the traditional PID tuning method, this method decouples and optimizes "position following performance" and "contact stability," avoiding the contradiction of "one set of parameters governing the whole." Simultaneously, it replaces empirical trial and error with quantified dynamic thresholds, and asynchronous data acquisition ensures no interference, significantly improving the accuracy, reproducibility, and applicability of parameter tuning, thereby enhancing the robot's control parameter tuning effect. Attached Figure Description

[0012] Figure 1 This application provides an illustration of the application environment for a robot control parameter tuning method according to an embodiment of the present application. Figure 2 This is a schematic diagram of the structure of a robot control system provided in an embodiment of this application; Figure 3 A flowchart illustrating a method for tuning control parameters of a robot provided in an embodiment of this application; Figure 4A schematic diagram illustrating the initialization configuration of control parameters provided in an embodiment of this application; Figure 5 A schematic diagram of the architecture of a robot control system provided in an embodiment of this application; Figure 6 This application provides a structural block diagram of a robot control parameter tuning device according to an embodiment of the present application. Figure 7 An internal structural diagram of a computer device provided in an embodiment of this application; Figure 8 An internal structural diagram of another computer device provided in an embodiment of this application; Figure 9 This is an internal structural diagram of a computer-readable storage medium provided in an embodiment of this application. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0014] The following explains some of the technical terms or phrases used in this application: Robot control parameters: These are the core software configuration parameters used to adjust the robot's force-position hybrid control characteristics. They are a key set of parameters for achieving "precise following in free space and smooth interaction in contact space" at the robot's end effector. Specifically, they include impedance control parameters (stiffness parameter K, damping parameter B, virtual mass parameter M, etc.), numerical stability parameters (spatial proportional gain parameter P, spatial differential gain parameter D, etc.), and error limiting parameters (position error limiting, attitude error limiting, etc.). These parameters directly determine the robot's force response characteristics when in contact with the environment and its following performance during manual control.

[0015] Forward dynamics solver: This refers to the numerical calculation module in FDCC based on the robot's dynamics model, running within the real-time control system. Its core function is to solve the robot's forward dynamics equations through numerical integration based on the input virtual force / torque commands and the robot's joint states (position, velocity), outputting the expected motion state (acceleration, velocity, position) of the robot's end effector. It is the core algorithm unit for realizing the dual-loop linkage of "virtual simulation-physical interaction," and the accuracy and speed of the solution directly affect the real-time performance and stability of FDCC.

[0016] Steady-state position following error: This refers to the stable deviation between the actual position of the robot's end effector and the target position when the robot performs manually guided following or trajectory tracking tasks in free space. It is a key quantitative indicator for measuring the robot's position following performance. This error is a constant deviation after dynamic stabilization, and its value directly reflects the "responsiveness" of the robot's manual control. The smaller the error, the higher the position following accuracy and the smoother the operation.

[0017] Forward Dynamics Compliance Control (FDCC) is a control strategy based on forward dynamics simulation models to calculate the compliant motion of a robot. Specifically, FDCC is essentially a double-loop nested nonlinear architecture. The inner loop performs numerical simulations of the kinematics and dynamics of the virtual robot, with the core calculations completed by a forward dynamics solver. The outer loop represents the contact force interaction in the real physical world, with virtual simulation parameters corrected in real time through force sensor feedback. This architecture integrates Cartesian impedance characteristics, force servo adjustment, and forward dynamics integration mechanisms, enabling simultaneous high-precision position following and compliant force control. It is widely used in human-robot collaboration, precision assembly, and other scenarios.

[0018] FDCC Controller: This refers to the core control unit running on a hard real-time system (e.g., Linux Preempt-RT), and is the hardware and software carrier of the FDCC control architecture. Internally, it integrates a forward dynamics solver, maintaining three parameter groups: impedance control, numerical stability, and inner-loop simulation. Its core function is to perform real-time calculation of FDCC control parameters, completing a closed-loop control process of "virtual force calculation - physical force feedback - dynamic parameter adjustment," making it the core hardware module for achieving compliant robot control.

[0019] Linux Preempt-RT is a real-time patch for the Linux operating system, used to convert standard Linux into a hard real-time operating system with deterministic latency.

[0020] IPC (Inter-Process Communication): Refers to the technical mechanisms by which two or more processes in a computer exchange data or signals.

[0021] Please see Figure 1 , Figure 1This diagram illustrates the application environment of a robot control parameter tuning method provided in this embodiment. The terminal 102 communicates with the server 104 via a communication network. A data storage system can store the data that the server 104 needs to process. The data storage system can be integrated onto the server 104 or located on a cloud or other network server. The terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart vehicle devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0022] It should be explained that the terminal 102 or the server 104 can execute any of the implementation methods described in the robot control parameter tuning method provided in the embodiments of this application, which will not be repeated here.

[0023] The computer device described in the embodiments of this application may include at least one of terminal 102 or server 104.

[0024] Please see Figure 2 , Figure 2 This is a schematic diagram of a robot control system provided in an embodiment of this application. The robot control system includes: a robot, an asynchronous interactive tuning unit, and a control unit. The control unit may include an FDCC controller, wherein: The robot is the main executor of the system and the ultimate recipient of compliant control parameter tuning and control commands, specifically undertaking two core tasks: Parameter tuning stage: According to the instructions issued by the control unit, the preset motion is executed in the preset free space. In conjunction with the asynchronous interactive tuning unit, key data such as dynamic reference input and steady-state position following error are collected to provide physical motion state support for parameter calculation. In the actual operation stage: it receives real-time control commands output by the control unit, completes tasks such as high-precision position following in free space and compliant contact force control in constrained space, and realizes target working conditions such as human-machine collaboration and precision assembly.

[0025] The asynchronous interactive tuning unit is the system's interference-free data acquisition and tuning interaction module. Its core function is to complete the detection, transmission, and interaction of data required for parameter tuning without affecting the real-time control flow of the control unit. For example, it can detect the joint position, velocity, and end effector posture during the robot's movement in real time, as well as intermediate variables such as dynamic input and position error within the FDCC controller. Then, it asynchronously transmits the acquired physical quantities and internal controller variables to the control unit, avoiding delays and interference to the real-time control loop. The asynchronous interactive tuning unit can also support parameter distribution and status feedback during the tuning process. It can receive parameter configuration commands from the control unit and display the tuning progress and key data to the operator, assisting in parameter debugging and verification.

[0026] The control unit is the core command and calculation center of the system, responsible for coordinating the parameter tuning process and real-time control tasks. The FDCC controller integrated inside is the core functional carrier. The control unit can control the robot to perform preset movements and receive the data collected by the asynchronous interactive tuning unit. Based on the collected data, it can carry out operations such as dynamic reference input measurement and parameter fusion calculation in order to complete the tuning of compliant control parameters.

[0027] Please see Figure 3 , Figure 3 This is a flowchart illustrating a robot control parameter tuning method provided in an embodiment of this application. This robot control parameter tuning method can be applied to a control unit in a robot control system (hereinafter referred to as the system). The system also includes: a robot and an asynchronous interactive tuning unit. The robot is located in a preset free space. The method may include the following steps: S101. Control the robot to perform preset movements in a preset free space.

[0028] Both the preset free space and preset motion can be preset in advance or left as default.

[0029] In this embodiment, the control unit may include an FDCC controller; the FDCC controller internally maintains three sets of key parameters to ensure the compliance and stability of the control process. The three sets of key parameters are as follows: Impedance control parameter group: stiffness parameter K, damping parameter B, virtual mass parameter M, position error limit max_error_distance, attitude error limit max_error_angle.

[0030] Numerical stability parameter set: spatial proportional gain parameter P, spatial differential gain parameter D, global dissipation factor damping_factor, global gain parameter error_scale.

[0031] Inner loop simulation parameter group: number of inner loop simulation iterations, simulation time step (simulation_period).

[0032] Before controlling the robot to perform a preset motion in a preset free space, the system needs to be initialized, as follows: Parameter dimensionality reduction strategy: Although parameters such as stiffness parameter K, damping parameter B, and spatial proportional gain parameter P are operated in matrix form within the algorithm, in this embodiment, the dimensionality of these parameters is reduced according to the principles of "isotropy" and "group isomorphism," forcing the translational degrees of freedom (X, Y, Z) parameters to have consistent values, and the rotational degrees of freedom (Rx, Ry, Rz) parameters to have consistent values. In this way, the complex adjustment of dozens of matrix elements can be simplified to the independent adjustment of "translational scalars" and "rotational scalars," reducing the complexity of parameter tuning. Taking stiffness parameter K as an example, in the Cartesian space control of a robot, stiffness parameter K is a 6×6 diagonal matrix (corresponding to 3 translational degrees of freedom X / Y / Z and 3 rotational degrees of freedom Rx / Ry / Rz). The complete form before dimensionality reduction is:

[0033] Among them, the translational degree of freedom parameters are k x 、k y 、k z (Stiffness values ​​in the X, Y, and Z directions, respectively), rotational degree of freedom parameters k rx 、k ry 、k rz (These correspond to stiffness values ​​in the RX, RY, and RZ directions, respectively); Before dimensionality reduction, adjusting the stiffness parameter K requires optimizing six parameters. These parameters are unconstrained, making the tuning process complex and prone to parameter mismatches between degrees of freedom, leading to distortion of the robot's end effector trajectory and fluctuations in contact force. Therefore, the stiffness parameter K can be constrained and dimensionality reduced by following the principles of isotropy and grouping isomorphism: Translational degrees of freedom are isotropic: forced k x = k y = k z = K t ( K t (for translational scalar stiffness). Rotational degrees of freedom are isotropic: forced k rx =kry =k rz = K r ( K r (for rotational scalar stiffness). The stiffness parameter K after dimensionality reduction simplifies to:

[0034] At this point, adjusting the stiffness parameter K requires optimizing two parameters ( K t and K r The parameters were simplified from 6 to 2, which greatly reduced the parameter coupling and tuning complexity, while ensuring that the motion characteristics of the translational and rotational degrees of freedom were consistent, thus avoiding control anomalies caused by parameter differences.

[0035] Simulation environment preset (key prerequisite): Fixed inner loop simulation parameter set and basic physical parameters to avoid parameter fluctuations affecting benchmark measurements. In this embodiment, the configuration is as follows: the number of inner loop simulation iterations is set to 1 to balance real-time control requirements and computational efficiency; the simulation time step is set to 0.02s (one order of magnitude higher than the actual control cycle) to smooth high-frequency noise; the global dissipation factor damping_factor is set to 0.90 to provide basic numerical damping to ensure stability; the virtual mass parameter M is set to 0 because the forward dynamic integral has inherent inertial characteristics and no additional configuration is required.

[0036] Global gain normalization: In the asynchronous interactive tuning unit, the global error scaling factor (ErrorScale) of the FDCC controller is set to 1.0, and the position error limit max_error_distance and attitude error limit max_error_angle are set to sufficiently large values ​​(e.g., 1.0m / 1.0rad) to completely eliminate the interference of nonlinear truncation effect on the reference measurement and ensure data accuracy.

[0037] Velocity term decoupling: Temporarily set the physical damping parameter B and spatial differential gain D to 0 to eliminate the masking effect of velocity-related terms on the dynamic reference input, ensuring that the measured reference value is contributed only by pure stiffness and gain terms, without the superposition of other variables.

[0038] Sensor shielding: Temporarily disable the robot's force sensor feedback function (ForceFeedback=0) to prevent random noise from the force sensor from causing system value divergence under the initial high gain state, and to ensure the stability of the initialization and subsequent benchmark measurement process.

[0039] It should be explained that grouping isomorphism refers to the logic of dividing the overall degrees of freedom, emphasizing "grouping by motion type," and clearly defining translation and rotation as two independent isomorphic units, with the parameters of the two groups being configurable differently. Isotropy refers to isotropy. In the embodiments of this application, isotropy specifically refers to the strategy of forcing the control parameters of different axes within the same group (e.g., X, Y, Z within the translation group) to remain consistent during the parameter tuning process.

[0040] In some embodiments, please refer to Figure 4 , Figure 4 This is a schematic diagram illustrating the initialization configuration of control parameters provided in an embodiment of this application. Figure 4 The specific content is as follows: controller_params: iterations:1# Number of internal iterations of FDCC simulation_period:0.02 # Fixed simulation time step (seconds) max_error_distance:0.05#meters max_error_angle:0.1#radians stiffness:[1000,1000,1000,40,40,40]# Initial stiffness #stiffness:[0.0,0.0,0.0,0.0,0.0,0.0]#Pure force control mode damping:[0,0,0,0,0,0] pd_gains:#6D Spatial PD Controller Gain p:[0.001,0.001,0.001,0.02,0.02,0.02] d:[0.0001,0.0001,0.0001,0.002,0.002,0.002] error_scale:1#Global gain 1 damping_factor:0.90# Global velocity damping (0.0 -> Immediate stop, 1.0 -> No damping) As can be seen, "iterations:1" indicates that the number of simulation iterations for the inner loop is set to 1; "max_error_distance:0.05#meters" indicates that the position error limit max_error_distance is 0.05, in meters. That is, when the position deviation of the robot end exceeds 0.05 meters, the system will trigger safety protection to prevent unexpected large movements.

[0041] "stiffness:[1000,1000,1000,40,40,40]" indicates the initial stiffness configuration for six degrees of freedom. The first three values ​​correspond to the X / Y / Z translational axes (stiffness of 1000 N / m), and the last three values ​​correspond to the Rx / Ry / Rz rotational axes (stiffness of 40 N / m). (m / rad), used to balance position following accuracy and force control compliance; "#stiffness:[0.0,0.0,0.0,0.0,0.0,0.0] #pure force control mode" is a comment line, indicating that when all stiffnesses are set to 0, the system enters pure force control mode, adjusting the output only through force feedback, without position stiffness constraints; "damping:[0,0,0,0,0,0]" represents the initial value of the six-degree-of-freedom damping coefficient, currently set to 0 to indicate that damping compensation is not enabled, and can be adjusted later according to the oscillation suppression requirements.

[0042] "p:[0.001,0.001,0.001,0.02,0.02,0.02]" represents the proportional gain (P), with the first three values ​​corresponding to the translational axis and the last three values ​​corresponding to the rotational axis. Its function is to convert position / attitude error into control output. "d:[0.0001,0.0001,0.0001,0.002,0.002,0.002]" represents the differential gain (D), with the first three values ​​corresponding to the translational axis and the last three values ​​corresponding to the rotational axis. It is used to suppress system oscillation. "error_scale:1" indicates that the error amplification factor is set to 1. 1 means that no additional scaling is applied to the error signal. If you need to enhance sensitivity, you can increase this value.

[0043] "damping_factor:0.90" means that the global velocity damping factor is set to 0.9. 0.9 means that 90% of the velocity response is retained and high-frequency vibration is moderately suppressed. The smaller the value, the stronger the damping (0.0 means immediate stop).

[0044] Specifically, after the system completes the initialization operation, the control unit can send parameter commands to the FDCC controller, setting the spatial proportional gain parameter P to 1.0 and the stiffness parameter K to 1.0; at the same time, it maintains the parameter state during the initialization phase to ensure that no additional variables interfere with data acquisition during the motion. Then, the control unit can generate a preset motion trajectory command according to the tuning requirements, specifying the motion parameters. For example, if the preset motion is sinusoidal motion, the trajectory command may include: specifying the motion axis (e.g., X-axis), amplitude (e.g., ±0.05m), frequency (e.g., 0.5Hz), and target working speed (e.g., 0.1m / s), ensuring that the motion amplitude and speed match the robot's actual working range; as another example, if the preset motion is reciprocating motion, the trajectory command may include: specifying the start and end point coordinates (e.g., X-axis 0m→0.1m→0m), the constant speed segment speed (i.e., the target working speed), and acceleration and deceleration time to avoid motion impact.

[0045] Then, the control unit can send the preset motion trajectory command to the robot, driving the robot to perform preset motion in the preset free space. During the motion, the control unit can continuously monitor the force sensor feedback status of the robot (since the force sensor feedback has been shielded during the initialization phase, it is necessary to confirm that there is no external force interference to trigger the feedback), ensuring that the robot is always in a non-contact free space working condition, and avoiding contact force from interfering with subsequent measurements.

[0046] S102. During the process of the robot performing the preset motion, the robot is detected by the asynchronous interactive tuning unit to obtain the dynamic reference input.

[0047] The asynchronous interaction tuning unit adopts a multi-process architecture design, including a data acquisition process and a UI rendering process, to achieve interference-free data acquisition and visual interaction. Data acquisition process: Key data such as dynamic input module length, end position error, and external contact force are extracted from the FDCC controller at a frequency of 100Hz via shared memory or high-speed IPC queue; UI rendering process: Build a visualization panel to draw the time-domain waveforms of the above variables in real time, and assist in the monitoring and verification of the parameter tuning process.

[0048] In some embodiments, the control unit includes an FDCC controller, which includes a forward dynamics solver; the robot is detected through an asynchronous interactive tuning unit to obtain dynamic reference input quantities, including: S21. The robot is detected by the asynchronous interactive tuning unit to obtain the magnitude data of the input vector of the forward dynamics solver; S22. Determine the dynamic reference input quantity based on the magnitude data of the input vector.

[0049] Specifically, the data acquisition process of the asynchronous interactive tuning unit can be initiated. A real-time data transmission channel is established with the FDCC controller through shared memory or a high-speed IPC queue. At the same time, the input monitoring window of the asynchronous interactive tuning unit is opened, and the monitoring parameter is configured as the input vector magnitude |U| of the forward dynamics solver. The data acquisition frequency is set to be synchronized with the robot's preset motion cycle (e.g., 100Hz). During the robot's execution of the preset motion, the forward dynamics solver will continuously receive the virtual force-related input vector U and calculate its magnitude |U| in real time (the magnitude reflects the overall magnitude of the input vector, avoiding the one-sidedness of a single-dimensional component). Meanwhile, the input monitoring window can collect the magnitude of the input vector U in real time, thereby obtaining the magnitude data of the input vector U. Then, the time-domain waveform of |U| is plotted based on the magnitude data to intuitively show its changing trend with the motion process. After the robot's motion state stabilizes (e.g., after completing 3-5 sinusoidal motion cycles or 2-3 reciprocating motions), the steady-state mean or peak value of the time-domain waveform of |U| is extracted and recorded as the dynamics reference input U_base.

[0050] It should be explained that the dynamic baseline input U_base represents the normalized "virtual force" level that the system can withstand, provided that the forward dynamic solver does not experience numerical divergence.

[0051] As can be seen, the core advantage of directly collecting the magnitude data of the input vector and extracting the threshold (i.e., the steady-state mean or peak value of the time-domain waveform of |U|) is that it replaces the empirical judgment of traditional PID parameter tuning with a quantified dynamic benchmark input, which defines a clear stable boundary for the forward dynamic solver to prevent numerical divergence in subsequent parameter tuning. This avoids the risk of system oscillation caused by blind parameter adjustment and ensures the reproducibility of the data acquisition and threshold determination process.

[0052] In some embodiments, please refer to Figure 5 , Figure 5 This is a schematic diagram of the architecture of a robot control system provided in an embodiment of this application. As can be seen, the robot control system includes a three-layer architecture, namely: Hardware layer: Includes the robot body and a six-dimensional force sensor; the robot body is responsible for performing motion and feeding back the state, the six-dimensional force sensor collects the force / torque signals at the time of contact to obtain the robot state, and then the robot state can be fed back to the shared memory in the real-time controller.

[0053] Real-time controller (i.e., control unit): includes control parameters (K, B, P, D), FDCC algorithm, and shared memory (IPC); the real-time controller can run the FDCC algorithm at a high frequency (greater than or equal to 500Hz), and in combination with the control parameters, complete high-speed data interaction through IPC and output joint commands to the robot body.

[0054] The asynchronous interaction system (i.e., the asynchronous interaction tuning unit) includes a visualization panel, a user interface rendering process, and a data acquisition process. The asynchronous interaction system can acquire and cache data in the IPC at a low frequency (less than or equal to 100Hz) through the data acquisition process. Then, the user interface rendering process draws waveforms on the visualization panel based on the cached data, showing the force control waveform and system status to the operator, thereby realizing human-machine monitoring and interaction.

[0055] S103. Tune the robot's position following performance to obtain the first set of tuning data.

[0056] In some embodiments, the robot's position-following performance is tuned to obtain a first set of tuning data, including: S31. Obtain *a* stiffness parameter values; *a* is an integer greater than 1. S32. The stiffness parameters of the system are adjusted sequentially to the stiffness parameter values ​​among a stiffness parameter values. Under each stiffness parameter value, test data of the robot's position following performance experiment are obtained, resulting in a set of test data. S33. Based on the test data of group a, determine the steady-state position following error value and the values ​​of a position following performance parameters; S34. Determine the first set of tuning data based on a position following performance parameter values, a stiffness parameter values, and steady-state position following error values.

[0057] The position tracking performance parameters may include at least one of the following: tracking lag time, root mean square error of position tracking, velocity response bandwidth, and trajectory tracking overlap.

[0058] Specifically, the target robot model can be obtained first, and the target stiffness parameter range can be determined based on the target robot model. For example, a pre-stored mapping relationship between machine models and stiffness parameter ranges can be used to determine the target stiffness parameter range corresponding to the target machine model. Then, values ​​can be selected within the target stiffness parameter range to obtain 'a' stiffness parameter values. For example, assuming the target stiffness parameter range is 0.2 to 2.0, and using an equal gradient interval method, if the gradient step size is 0.5, then 4 stiffness parameter values ​​can be obtained: 0.2, 0.7, 1.2, and 1.7; that is, 'a' equals 4.

[0059] Next, the spatial proportional gain parameter P can be kept at 1.0. Then, the stiffness parameter of the system is adjusted to the stiffness parameter value among a stiffness parameter values ​​in sequence. Under each stiffness parameter value, the test data of the robot's position following performance experiment is obtained, resulting in a set of test data. The position following performance experiment can include one of the following: manual guided following experiment or preset trajectory tracking experiment.

[0060] To illustrate, suppose the position following performance experiment is a manually guided following experiment. Keep the spatial proportional gain parameter P at 1.0 and other parameters unchanged during the initialization phase. Adjust the system stiffness parameter to the first stiffness parameter value, which is any one of a stiffness parameter values. Then, the operator, wearing anti-collision gloves, holds the robot's end effector and performs a combination of movements at a uniform and stable speed along the X / Y / Z translational and Rx / Ry / Rz rotational directions. The movement trajectory covers the commonly used working range (e.g., translational amplitude ±0.1m, rotational amplitude ±10°) to guide the robot to follow the movement. During the movement, the actual position coordinates of the robot's end effector and the target position coordinates guided by the operator (which can be obtained through a visual positioning device) can be collected in real time at a frequency of 100Hz through the asynchronous interactive tuning unit. The timestamp of each sampling moment is recorded synchronously. Repeat the above guided movement 3 to 5 times to ensure that the data is statistically significant and to avoid the randomness of a single movement. Thus, a set of test data is obtained. Repeating this a times will yield a sets of test data.

[0061] It needs to be explained that the manual guided following experiment simulates human-machine collaboration scenarios (e.g., drag-and-drop teaching, manual assembly), where the operator directly guides the robot's end effector movement to test the system's "responsiveness"; the preset trajectory tracking experiment simulates automated operation scenarios (e.g., precise trajectory reproduction), where the control unit issues fixed trajectory commands to test the robot's trajectory reproduction accuracy.

[0062] Next, based on the test data set 'a', the steady-state position following error value and 'a' position following performance parameter values ​​can be determined. Specifically, the position following performance parameter can be the following lag time. For each set of test data, the effective motion segment data can be screened first, eliminating the acceleration segment during the motion start-up phase and the deceleration segment during the stopping phase, retaining only the steady-state data of uniform guided motion to avoid interference from the non-steady-state phase on the index calculation. Then, based on the timestamps recorded by the asynchronous interactive tuning unit, the actual position coordinate sequence of the robot end effector is matched with the target position coordinate sequence guided by the operator time-by-time to ensure that the sampling times of the two sets of data correspond one-to-one, obtaining time-aligned position data pairs. x actual ( t n ) ，x target ( t n )),in, t n Indicates the first n Each sampling time, x actual ( t n ) indicates the first nThe actual position coordinates of the robot's end effector at each sampling time. x target ( t n ) indicates the first n The target position coordinates at each sampling time; for already aligned position data pairs, their corresponding position following error values ​​can be calculated, and the specific calculation formula is as follows:

[0063] in, This indicates the position following error value; N This indicates the number of sampled data points for steady-state data.

[0064] It should be explained that if the test data in group a contains multiple degrees of freedom of translation and rotation, then the steady-state error of each degree of freedom needs to be calculated separately, and then the average value is taken as the comprehensive steady-state position following error value (i.e., position following error value) of the data in group a.

[0065] Next, in the aligned position data pairs, identify the moments when the target position changes significantly, and record them as abrupt change points, such as the moment when the X-axis target position jumps from 0m to 0.05m. Then, find the earliest moment when the actual position coordinates change from the value before the abrupt change to when the target position deviation is ≤ a preset deviation threshold. Determine the following lag time based on this earliest moment and the abrupt change point. For example, assuming the sampling frequency of the asynchronous interactive tuning unit is 100Hz, i.e., the sampling time interval is 0.01s, the preset deviation threshold can be preset in advance or defaulted, for example, the preset deviation threshold can be set to 5%. In the time-series aligned data of the X-axis manually guided following experiment, identify the abrupt change point t. change =10.00s, at which point the target position on the X-axis jumps from 0m to 0.05m; the steady-state value of the actual position on the X-axis before the sudden change is 0.001m (minor fluctuations are ignored). Then, the earliest moment when the actual position satisfies "deviation from target position ≤ 5%" is found: the 5% deviation threshold for the target position of 0.05m is: 0.05 × 5% = 0.0025m; that is, the actual position needs to fall within 0.0475m. Within the 0.0525m interval, after searching the data sequence, the earliest time t was found. response At 10.08s, the actual position of the X-axis reaches 0.048m, meeting the requirement that the target position deviation is ≤5%. At this point, the position can be determined based on the earliest time t. response and the mutation time point t change The calculation is performed using the following formula: T delay =t response -t change =10.08-10.00=0.08 (s); Among them, T delay Indicates the following lag time; It should be explained that if a set of test data contains multiple movements, the following lag time for each movement can be calculated to obtain multiple following lag times. The average of these multiple following lag times is taken as the following lag time for the set of test data.

[0066] Thus, by analyzing each set of test data in group a, we can obtain a position following performance parameter values ​​and a position following error values. The average of these a position following error values ​​is taken as the steady-state position following error value.

[0067] Finally, based on *a* position following performance parameter values, *a* stiffness parameter values, and steady-state position following error values, the first set of tuning data can be determined. Specifically, the optimal position following performance parameter value among the *a* position following performance parameter values ​​can be determined. For example, assuming the *a* position following performance parameter values ​​are *a* following lag times, the minimum value among these *a* position following performance parameter values ​​can be determined, and this minimum value is the optimal position following performance parameter value. Alternatively, assuming the *a* position following performance parameter values ​​are *a* trajectory tracking overlap degrees, the maximum value among these *a* position following performance parameter values ​​can be determined, and this maximum value is the optimal position following performance parameter value. Next, the optimal stiffness parameter value (i.e., the reference stiffness parameter value) corresponding to the optimal position following performance parameter value among the *a* stiffness parameter values ​​can be determined. The first set of tuning data is composed of the reference stiffness parameter value and the steady-state position following error value.

[0068] In some embodiments, this application introduces a dimensionless dynamic reference input as a quantization anchor for numerical stability, and constructs the following parameter equilibrium equation: U_base≈P*K*Δx≈P*F_contact; Wherein, U_base represents the dynamic reference input; P represents the spatial proportional gain parameter; K represents the stiffness parameter; Δx represents the robot's position following error (in practical applications, it can be measured, or Δx_ref can be used directly to replace Δx, where Δx_ref represents the above steady-state position following error value); F_contact represents the robot's contact interaction force (which can be measured by a force sensor).

[0069] This parameter equilibrium equation quantitatively reveals the intrinsic transformation relationship between stiffness parameter, spatial proportional gain parameter, follow-up error (i.e., position following error), and contact interaction force. It breaks the traditional "black box" debugging mode of parameter tuning, transforms the complex nonlinear performance trade-off problem into an algebraic calculation problem based on physical benchmark values, and provides solid theoretical support for parameter decoupling and optimization.

[0070] Thus, by abandoning the traditional parameter adjustment method that relies on manual feel, and instead adopting a systematic gradient selection of stiffness parameters and conducting standardized position following performance experiments, a quantitative mapping relationship between stiffness parameters and steady-state following error and following lag time is established, so that parameter tuning has clear data support.

[0071] In addition, by selecting the parameter range that meets the position following requirements based on the test data of group a, the first set of tuning data can be directly used as the optimization benchmark for subsequent fusion contact stability, which greatly shortens the overall parameter tuning cycle.

[0072] S104. Determine the corresponding spatial proportional gain parameter value for the robot.

[0073] In some embodiments, determining the spatial proportional gain parameter value corresponding to the robot includes: A1. Obtain the rated contact force corresponding to the robot; A2. Determine the spatial proportional gain parameter value based on the preset dynamic input equilibrium equation, rated contact force, and dynamic reference input.

[0074] The preset dynamic input equilibrium equation can be preset in advance or set to default.

[0075] Specifically, the target application scenario of the robot can be obtained, and the rated contact force for normal operation of the robot can be determined based on the target application scenario. Specifically, a pre-stored mapping relationship between application scenarios and contact forces can be used to determine the rated contact force corresponding to the target application scenario. For example, assuming the target application scenario is a human-robot collaborative dragging scenario, referring to ergonomic standards, the contact force must meet the requirement of comfortable and pressure-free human contact, and the value range of the contact force is 20 Newtons. 40 Newtons, therefore, the rated contact force can be set to the middle value, that is, 30 Newtons.

[0076] Next, the rated contact force and dynamic reference input can be substituted into the preset dynamic input equilibrium equation to calculate the spatial proportional gain parameter value. The dynamic input equilibrium equation is as follows: P_stable=U_base / F_normal; Where P_stable represents the spatial proportional gain parameter value; U_base represents the dynamic reference input; and F_normal represents the rated contact force.

[0077] Thus, by relying on the dynamic input equilibrium equation, and using the rated contact force and the dynamic reference input, two quantitative indicators with clear physical meaning, as the basis for calculation, the spatial proportional gain parameter value is directly derived. This replaces the traditional gain parameter tuning method that relies on repeated trial and error. It ensures that the parameter value is strictly anchored to the system stability boundary (not exceeding the numerical divergence threshold corresponding to the dynamic reference input), and also ensures that the robot's interactive contact force accurately matches the rated operation requirements under the final parameter configuration. This achieves precise control of the contact force performance within the stability boundary, and greatly improves the scientificity and efficiency of parameter tuning.

[0078] In some embodiments, the robot includes a sensor module; after determining the spatial proportional gain parameter value based on a preset dynamic input equilibrium equation, a rated contact force, and a dynamic reference input, the method further includes: B1. Set the system stiffness parameter to 0, set the spatial proportional gain parameter to the spatial proportional gain parameter value, and enable the feedback of the sensor module. B2. Conduct contact tests on the robot to obtain target test data; B3. Determine the test results based on the target test data; the test results include one of the following: the spatial proportional gain parameter value is qualified, or the spatial proportional gain parameter value is unqualified. B4. When the test results include an unqualified spatial proportional gain parameter value, update the spatial proportional gain parameter value to obtain a new spatial proportional gain parameter value.

[0079] The sensor module can include the robot's force sensor. Of course, the sensor module can also include other sensors, such as vision sensors, position sensors, speed sensors, and attitude sensors.

[0080] Specifically, the system stiffness parameter is set to 0 (pure force control mode), the spatial proportional gain parameter is set to the spatial proportional gain parameter value, and other parameters remain unchanged. Simultaneously, feedback from the sensor module is enabled; for example, the sensor module may include a force sensor, and its feedback function is activated. Then, a contact test can be performed on the robot to obtain target test data. Specifically, the contact test can include at least one of the following: static contact force stability test and dynamic contact following test (e.g., dragging, assembly, grinding, etc.). For example, assuming the contact test is a static contact force stability test, the control unit can control the robot to drive the robot's end effector close to the contact platform at a low speed (e.g., 0.02 m / s) until the force sensor detects the interaction force rising from 0. The robot's force control output is then adjusted to stabilize the real-time interaction force at the rated contact force F_normal, maintaining this force value for 10-20 seconds to ensure the system enters a steady state. During this test, the interaction force waveform is observed in real time through the force monitoring window of the asynchronous interactive tuning unit. Simultaneously, the control unit can also record the system's vibration amplitude and frequency, thereby obtaining the target test data.

[0081] Next, the test results can be determined based on the target test data. Specifically, the test data after the real-time interaction force stabilizes at the rated contact force F_normal can be extracted from the target test data to obtain steady-state test data. Then, the maximum system vibration amplitude and the maximum fluctuation amplitude of the real-time interaction force can be determined based on the steady-state test data. If the maximum fluctuation amplitude is less than the preset fluctuation amplitude, and the maximum system vibration amplitude is less than the preset vibration amplitude, then the test results, including the space proportional gain parameter value, are deemed qualified. Conversely, if the maximum fluctuation amplitude is not less than the preset fluctuation amplitude, and / or the maximum system vibration amplitude is not less than the preset vibration amplitude, then the test results, including the space proportional gain parameter value, are deemed unqualified. The preset fluctuation amplitude and the preset vibration amplitude can both be preset in advance or defaulted.

[0082] For example, suppose F_normal is 30 Newtons; the preset fluctuation amplitude is 10% of F_normal, i.e., 3 Newtons; the preset vibration amplitude is 0.05mm (to ensure no obvious mechanical vibration), control the robot end effector to adhere to the contact platform, adjust the interaction force to stabilize at 30N and maintain it for 30 seconds, and collect steady-state test data during this period: Real-time interactive force data (unit: N): 29.2, 29.8, 30.1, 29.5, 30.3, 29.7, 30.0, 29.9; System vibration amplitude data (unit: mm): 0.02, 0.03, 0.025, 0.018, 0.032; The steady-state test data shows that the maximum real-time interaction force is 30.3N, the minimum is 29.2N, and the maximum fluctuation amplitude is 30.3N - 29.2N = 1.1N. The maximum fluctuation amplitude of 1.1N is less than the preset fluctuation amplitude of 3N. In addition, the maximum system vibration amplitude is 0.03mm, which is less than the preset vibration amplitude of 0.05mm. Both indicators are less than the corresponding thresholds. Therefore, the test results indicate that the spatial proportional gain parameter value is qualified and can be used as the benchmark value for subsequent parameter optimization.

[0083] If the test results include an unqualified spatial proportional gain parameter value, the spatial proportional gain parameter value can be updated to obtain a new spatial proportional gain parameter value. For example, suppose the original spatial proportional gain parameter value was set to 1.2, but after testing, it was found that the system vibration amplitude exceeded the standard, and the parameter was determined to be unqualified. At this time, the parameter can be adjusted to 1.0 or 0.8 to obtain a new spatial proportional gain value, and then the test can be repeated to verify whether it meets the standard.

[0084] In this way, by first fixing the stiffness parameter to 0 and then enabling sensor feedback, the interference of stiffness on force control performance is eliminated, and the rationality of the spatial proportional gain parameter can be accurately verified. At the same time, the qualification of the parameter is determined based on the actual data of the contact test, avoiding the blindness of subjective experience in parameter tuning. When the parameter is not qualified, the gain value is updated in time, which can quickly find the parameters that are suitable for the stable operation of the system, ensuring the response accuracy and stability of the robot in the force control interaction scenario, and greatly improving the efficiency and scientific nature of parameter tuning.

[0085] S105. Determine the second set of tuning data based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input.

[0086] In some embodiments, the first set of tuning data includes: a reference stiffness parameter value and a steady-state position following error value; the reference stiffness parameter value is a stiffness parameter value among a stiffness parameter values; the second set of tuning data is determined based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input, including: C1. Determine the target stiffness parameter value based on the preset stiffness calculation formula, dynamic reference input, spatial proportional gain parameter value, and steady-state position following error value; the specific stiffness calculation formula is as follows: K_final=U_base / (P_stable*Δx_ref); Wherein, K_final represents the target stiffness parameter value; U_base represents the dynamic reference input; P_stable represents the spatial proportional gain parameter value; and Δx_ref represents the steady-state position following error value. C2. Determine the second set of tuning data based on the target stiffness parameter value, dynamic reference input, spatial proportional gain parameter value, and steady-state position following error value.

[0087] The preset stiffness calculation formula can be preset in advance or used as a default value.

[0088] It needs to be explained that the target stiffness parameter value K_final, obtained through the preset stiffness calculation formula, ensures mathematically that the system: In free space, the first parameter equation is satisfied: P_stable*K_final*Δx≈U_base (preserving the original responsiveness); Substituting the preset stiffness calculation formula into the first parameter equation and simplifying, we get: Δx / Δx_ref≈1; This means that the actual position following error Δx will be stable near the steady-state position following error value Δx_ref, and will not cause motion lag or overshoot due to improper stiffness settings, ensuring sensitive operation response.

[0089] In the contact space, the second parameter equation is satisfied: P_stable*F_normal≈U_base (which guarantees stability under rated operating conditions).

[0090] When the robot comes into contact with the environment and applies the rated contact force F_normal, this second parameter equation provides the mathematical guarantee for force control stability. Combined with the aforementioned parameter balance equation, it can be seen that the system's dynamic baseline input U_base matches the force control output, avoiding force oscillations caused by excessive gain or stiffness, while also preventing sluggish force response due to excessively low parameters, ensuring that force control accuracy and stability meet standards under rated operating conditions.

[0091] Thus, by relying on quantitative stiffness calculation formulas, the target stiffness parameter value is directly derived by using three parameters with clear physical meaning and experimental support—dynamic reference input, spatial proportional gain parameter value, and steady-state position following error value—as the basis for calculation, replacing the traditional stiffness tuning method that relies on empirical trial and error.

[0092] In addition, the second set of tuning data is determined based on the target stiffness parameter value and the above core parameters. This ensures that the parameter values ​​are strictly anchored to the numerical stability boundary of the system (limited by U_base), and also ensures that the robot's position following accuracy (limited by Δx_ref) and force control performance (related by P_stable) under the final stiffness configuration are optimally matched. This greatly improves the scientific nature and accuracy of parameter tuning and provides a reproducible and quantifiable parameter benchmark for robot compliant control.

[0093] In some embodiments, after determining the second set of tuning data based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input, the method further includes: D1. Obtain the maximum safe contact force corresponding to the robot; D2. Determine the spatial differential gain value based on the spatial proportional gain parameter value; D3. Obtain the tool arm and rated torque constraints corresponding to the robot; D4. Determine the equivalent force constraint based on the tool lever arm and rated torque constraint; D5. Based on equivalent force constraints, the rotational domain parameters of the robot are tuned to obtain the third set of tuning data; D6. Determine the steady-state position following error amplitude based on the steady-state position following error value, the target stiffness parameter value, and the maximum safe contact force.

[0094] Among them, attitude parameters are key parameters that describe the position and attitude state of the robot's end effector (or tool) in three-dimensional space. Their core function is to establish the mapping relationship between joint motion, end effector pose, and force / torque constraints, so as to ensure the stability of the end effector's attitude and the accuracy of interaction during force control.

[0095] Specifically, the target machine model of the robot can be obtained first, and the maximum safe contact force can be determined based on the target machine model. For example, a preset mapping relationship between machine models and safe contact forces can be stored in advance, and the maximum safe contact force corresponding to the target machine model can be determined based on the mapping relationship. For example, assuming that the threshold for emergency stop of the robot's robotic arm or environmental damage is determined to be 50 Newtons based on the target machine model, the maximum safe contact force can be set to 50 Newtons.

[0096] Next, the spatial differential gain value can be determined based on the spatial proportional gain parameter value. The specific calculation formula is as follows: D≈d*P_stable; Where D represents the spatial differential gain value; d represents the preset scaling factor; P_stable represents the spatial scaling gain parameter value, and the preset scaling factor d can be 0.1.

[0097] Next, the corresponding tool arm and rated torque constraints of the robot can be obtained. Specifically, conventional TCP calibration methods for robots (e.g., the four-point method or the six-point method) can be used to determine the precise three-dimensional coordinates of the tool center point (TCP) of the robot's end effector in the robot's base coordinate system, thus obtaining the TCP position coordinates. Then, the robot's technical manual or sensor installation drawings can be consulted to obtain the sensor center coordinates of the force sensor in the robot's base coordinate system. The distance between the TCP position coordinates and the sensor center coordinates can be calculated using the distance formula between two points in space, thus obtaining the target distance, i.e., the tool arm. Next, the target application scenario of the robot can be obtained, and the actual rotational operation conditions of the robot in the target application scenario (e.g., tightening bolts, flipping workpieces, grinding curved surfaces) can be determined, thus determining the allowable rated rotational torque range of the end effector, i.e., the rated torque constraints.

[0098] Then, the equivalent force constraint can be determined based on the tool lever arm and rated torque constraint. The specific calculation formula is as follows: F_eq=tau_normal / L; Where F_eq represents the equivalent force constraint; tau_normal represents the rated torque constraint; L represents the tool arm; then, the rotational domain parameters of the robot can be tuned based on the equivalent force constraint to obtain the third set of tuning data. Specifically, the equivalent force constraint can be used as the equivalent rated contact force in the rotational domain to replace the rated contact force F_normal used in the translational domain tuning. Steps S101 to S105 are repeated to tune the rotational domain parameters of the robot to obtain the stiffness parameter value and the spatial proportional gain parameter value in the rotational domain. These two parameter values ​​constitute the third set of tuning data.

[0099] It needs to be explained that the parameters in the translational domain and the parameters in the rotational domain are in one-to-one correspondence. This correspondence stems from their shared tuning logic and equivalent control objectives. The rotational domain achieves the equivalent conversion of "torque-force" through the tool arm L, completely reusing the parameter tuning model of the translational domain. Therefore, each parameter in the translational domain can find a functionally matching parameter in the rotational domain. For example, if there is a spatial proportional gain parameter in the translational domain, there is a corresponding rotational proportional gain parameter in the rotational domain, and so on. These will not be elaborated further here.

[0100] In this embodiment, the parameters in the second set of tuning data are parameters of the translational domain. The parameters of the translational domain are responsible for position following and force control stability in the X / Y / Z axis directions, ensuring responsiveness in free space and force accuracy in contact space. Correspondingly, the parameters in the third set of tuning data are parameters of the rotational domain. The parameters of the rotational domain are responsible for attitude following and torque control stability in the Rx / Ry / Rz directions, ensuring attitude accuracy during rotational motion and stability during torque interaction.

[0101] In some embodiments, the third set of tuning data may include, in addition to the stiffness parameter value and spatial proportional gain parameter value in the rotation domain, the differential gain parameter, steady-state attitude error, and safety constraint parameters (e.g., maximum safety torque, attitude error limit, etc.) in the rotation domain.

[0102] Finally, the steady-state position following error amplitude can be determined based on the steady-state position following error value, the target stiffness parameter value, and the maximum safe contact force. Specifically, the first ratio can be determined based on the target stiffness parameter value and the maximum safe contact force, and the specific calculation formula is as follows: Q = F_safe / K_final; Where Q represents the first ratio; F_safe represents the maximum safe contact force; K_final represents the target stiffness parameter value; then, the steady-state position following error amplitude can be determined based on the first ratio and the steady-state position following error value. Specifically, the smaller value between the first ratio and the steady-state position following error value can be determined as the steady-state position following error amplitude, or the smaller value can be appropriately reduced to obtain the steady-state position following error amplitude.

[0103] It should be explained that if the steady-state position following error amplitude is taken as the steady-state position following error value, it can prevent the robot from making unexpected large movements due to excessive position deviation, thus ensuring that the motion trajectory is controllable. If the steady-state position following error amplitude is taken as the first ratio, it can ensure that when the robot collides rigidly with the environment, the maximum static force generated by the deformation of the environment is strictly limited to within F_safe, avoiding overload damage to the equipment or the environment.

[0104] The aforementioned robot control parameter tuning method involves controlling the robot to execute preset movements, using an asynchronous interactive tuning unit to collect data and determine dynamic thresholds (i.e., dynamic reference inputs) to provide quantization boundaries for parameter tuning. Then, position following performance is optimized separately to obtain basic parameters that meet operational requirements (i.e., the first set of tuning data). Finally, the spatial proportional gain parameter value is combined to obtain the optimal parameters that balance position accuracy and contact stability (i.e., the second set of tuning data). Compared to traditional PID parameter tuning, this method decouples and optimizes "position following performance" and "contact stability," avoiding the contradiction of "one set of parameters governing the whole." Simultaneously, it replaces empirical trial and error with quantified dynamic thresholds, and asynchronous data acquisition ensures interference-free operation, significantly improving the accuracy, reproducibility, and applicability of parameter tuning, thereby enhancing the robot's control parameter tuning effect.

[0105] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps.

[0106] Based on the same inventive concept, this application also provides a robot control parameter tuning device. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more robot control parameter tuning device embodiments provided below can be found in the limitations of the robot control parameter tuning method described above, and will not be repeated here.

[0107] like Figure 6 As shown, this application embodiment provides a robot control parameter tuning device 600, applied to the control unit in a robot control system. The system further includes: a robot, an asynchronous interactive tuning unit, and the robot is located in a preset free space. The robot control parameter tuning device 600 includes: Control module 601 is used to control the robot to perform preset movements in a preset free space; The tuning module 602 is used to detect the robot through an asynchronous interactive tuning unit during the robot's execution of a preset motion, obtain the dynamic reference input quantity, tune the robot's position following performance to obtain the first set of tuning data, determine the corresponding spatial proportional gain parameter value of the robot, and determine the second set of tuning data based on the spatial proportional gain parameter value, the first set of tuning data and the dynamic reference input quantity.

[0108] In some embodiments, the control unit includes an FDCC controller, which includes a forward dynamics solver; in terms of detecting the robot through an asynchronous interactive tuning unit to obtain dynamics reference inputs, the tuning module 602 is specifically used for: The robot is detected by the asynchronous interactive tuning unit, and the magnitude data of the input vector of the forward dynamics solver is obtained. Based on the magnitude data of the input vector, determine the dynamic reference input quantity.

[0109] In some embodiments, in tuning the robot's position-following performance to obtain a first set of tuning data, the tuning module 602 is specifically used for: Obtain *a* stiffness parameter values; *a* is an integer greater than 1. The system stiffness parameters are sequentially adjusted to the stiffness parameter values ​​among a stiffness parameter values. Under each stiffness parameter value, test data of the robot's position following performance experiment are obtained, resulting in a set of test data. Based on the test data of group a, determine the steady-state position following error value and the values ​​of a position following performance parameters; The first set of tuning data is determined based on a position following performance parameter values, a stiffness parameter values, and steady-state position following error values.

[0110] In some embodiments, the tuning module 602 is specifically used for determining the spatial proportional gain parameter value corresponding to the robot as follows: Obtain the robot's rated contact force; The spatial proportional gain parameter value is determined based on the preset dynamic input equilibrium equation, rated contact force, and dynamic reference input.

[0111] In some embodiments, the robot includes a sensor module, and the tuning module 602 is further specifically used for: Set the system stiffness parameter to 0, set the spatial proportional gain parameter to the spatial proportional gain parameter value, and enable the feedback of the sensor module. Conduct contact tests on the robot to obtain target test data; The test results are determined based on the target test data; the test results include one of the following: the spatial proportional gain parameter value is qualified, or the spatial proportional gain parameter value is unqualified. If the test results include an unqualified spatial proportional gain parameter value, the spatial proportional gain parameter value is updated to obtain a new spatial proportional gain parameter value.

[0112] In some embodiments, the first set of tuning data includes: a reference stiffness parameter value and a steady-state position following error value; the reference stiffness parameter value is a stiffness parameter value among a stiffness parameter values; in determining the second set of tuning data based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input, the tuning module 602 is specifically used for: Based on the preset stiffness calculation formula, dynamic reference input, spatial proportional gain parameter value, and steady-state position following error value, the target stiffness parameter value is determined; the specific stiffness calculation formula is as follows: K_final=U_base / (P_stable*Δx_ref); Wherein, K_final represents the target stiffness parameter value; U_base represents the dynamic reference input; P_stable represents the spatial proportional gain parameter value; and Δx_ref represents the steady-state position following error value. The second set of tuning data is determined based on the target stiffness parameter value, dynamic reference input, spatial proportional gain parameter value, and steady-state position following error value.

[0113] In some embodiments, the setting module 602 is further specifically used for: Obtain the maximum safe contact force corresponding to the robot; Determine the spatial differential gain value based on the spatial proportional gain parameter value; Obtain the tool arm and rated torque constraints corresponding to the robot; Determine the equivalent force constraint based on the tool lever arm and rated torque constraint; Based on equivalent force constraints, the rotational domain parameters of the robot are tuned to obtain the third set of tuning data. The steady-state position following error amplitude is determined based on the steady-state position following error value, the target stiffness parameter value, and the maximum safe contact force.

[0114] Each module in the aforementioned robot control parameter tuning device 600 can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0115] In some embodiments, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 7 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data related to the robot's control parameter tuning method. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the steps in the robot's control parameter tuning method described above.

[0116] In some embodiments, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements the steps in the robot control parameter tuning method described above. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen; the input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs or touchpads set on the casing of the computer device, or external keyboards, touchpads or mice, etc.

[0117] Those skilled in the art will understand that Figure 7 or Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0118] In some embodiments, a computer device is provided, the computer device including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps in the above method embodiments.

[0119] In some embodiments, such as Figure 9 The diagram shows the internal structure of a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the above-described method embodiments.

[0120] In some embodiments, a computer program product is provided, which includes a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0121] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0122] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0123] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0124] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for tuning control parameters of a robot, characterized in that, A control unit applied in a robot control system, the system further comprising: a robot and an asynchronous interactive tuning unit, the robot being situated in a preset free space; the method comprising: The robot is controlled to perform a preset movement within the preset free space; During the process of the robot performing the preset motion, the robot is detected by the asynchronous interactive tuning unit to obtain the dynamic reference input. The robot's position-following performance is tuned to obtain the first set of tuning data; Determine the spatial proportional gain parameter value corresponding to the robot; The second set of tuning data is determined based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input.

2. The method according to claim 1, characterized in that, The control unit includes an FDCC controller, and the FDCC controller includes a forward dynamics solver; The step of detecting the robot through the asynchronous interactive tuning unit to obtain the dynamic reference input includes: The robot is detected by the asynchronous interactive tuning unit to obtain the magnitude data of the input vector of the forward dynamics solver; Based on the magnitude data of the input vector, the dynamic reference input quantity is determined.

3. The method according to claim 1 or 2, characterized in that, The tuning of the robot's position-following performance yields a first set of tuning data, including: Obtain *a* stiffness parameter values; where *a* is an integer greater than 1. The stiffness parameters of the system are sequentially adjusted to the stiffness parameter values ​​among the a stiffness parameter values. Under each stiffness parameter value, test data of the robot's position following performance experiment are obtained, resulting in a set of test data. Based on the test data of group a, determine the steady-state position following error value and the values ​​of a position following performance parameters; The first set of tuning data is determined based on the a position following performance parameter values, the a stiffness parameter values, and the steady-state position following error value.

4. The method according to claim 3, characterized in that, Determining the spatial proportional gain parameter value corresponding to the robot includes: Obtain the rated contact force corresponding to the robot; The spatial proportional gain parameter value is determined based on the preset dynamic input equilibrium equation, the rated contact force, and the dynamic reference input.

5. The method according to claim 4, characterized in that, The robot includes a sensor module; After determining the spatial proportional gain parameter value based on the preset dynamic input equilibrium equation, the rated contact force, and the dynamic reference input, the method further includes: Set the stiffness parameter of the system to 0, set the spatial proportional gain parameter to the value of the spatial proportional gain parameter, and simultaneously enable the feedback of the sensor module. The robot was subjected to a contact test to obtain target test data; Based on the target test data, the test result is determined; the test result includes one of the following: the spatial ratio gain parameter value is qualified, or the spatial ratio gain parameter value is unqualified; When the test result includes a non-compliant spatial proportional gain parameter value, the spatial proportional gain parameter value is updated to obtain a new spatial proportional gain parameter value.

6. The method according to claim 4, characterized in that, The first set of tuning data includes: a reference stiffness parameter value and the steady-state position following error value; the reference stiffness parameter value is the stiffness parameter value among the a stiffness parameter values; The step of determining the second set of tuning data based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input includes: The target stiffness parameter value is determined based on the preset stiffness calculation formula, the dynamic reference input, the spatial proportional gain parameter value, and the steady-state position following error value; the specific stiffness calculation formula is as follows: K_final=U_base / (P_stable*Δx_ref); Wherein, K_final represents the target stiffness parameter value; U_base represents the dynamic reference input; P_stable represents the spatial proportional gain parameter value; and Δx_ref represents the steady-state position following error value. The second set of tuning data is determined based on the target stiffness parameter value, the dynamic reference input, the spatial proportional gain parameter value, and the steady-state position following error value.

7. The method according to claim 6, characterized in that, After determining the second set of tuning data based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input, the method further includes: Obtain the maximum safe contact force corresponding to the robot; Based on the spatial proportional gain parameter value, determine the spatial differential gain value; Obtain the tool arm and rated torque constraints corresponding to the robot; Based on the tool lever arm and the rated torque constraint, determine the equivalent force constraint; Based on the equivalent force constraint, the rotational domain parameters of the robot are tuned to obtain the third set of tuning data; The steady-state position following error amplitude is determined based on the steady-state position following error value, the target stiffness parameter value, and the maximum safe contact force.

8. A control parameter tuning device for a robot, characterized in that, A control unit applied in a robot control system, the system further comprising: a robot and an asynchronous interactive tuning unit, the robot being situated in a preset free space; the device comprising: The control module is used to control the robot to perform preset movements in the preset free space; The tuning module is used to detect the robot through the asynchronous interactive tuning unit during the robot's execution of the preset motion, obtain the dynamic reference input quantity, tune the robot's position following performance to obtain a first set of tuning data, determine the corresponding spatial proportional gain parameter value of the robot, and determine a second set of tuning data based on the spatial proportional gain parameter value, the first set of tuning data, and the dynamic reference input quantity.

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

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 steps of the method according to any one of claims 1 to 7.

Images