Robot low-speed motion control method and device and computer readable storage medium

Abstract

The invention provides a robot low-speed motion control method and device and a computer readable storage medium. According to the scheme, an original position deviation signal of a robot when executing a low-speed task is obtained, and the original position deviation signal is converted into a frequency domain signal; processing the frequency domain signal by adopting a Chebyshev low-pass filter to obtain a filtered frequency domain signal; the filtered frequency domain signal is converted into a time domain signal through inverse Fourier transform, a filtered time domain signal is obtained, the speed loop gain of the robot is adjusted until a preset stop condition is met, and the adjusted speed loop gain is obtained; and controlling the robot to move by using the filtered time domain signal and the adjusted speed loop gain. According to the scheme, the problem that the stability of low-speed movement of the robot is poor is solved.

Description

Technical Field

[0001] This application relates to the field of robot control technology, and more specifically, to a control method for low-speed robot movement, a control device for low-speed robot movement, and a computer-readable storage medium. Background Technology

[0002] The vibration of a robot originates from its internal mechanical structure and moving parts. When a robot is working, its internal mechanical components such as motors, reducers, and joints generate power, and this power transmission causes vibrations. This vibration is particularly noticeable when the robot performs low-speed, high-precision movements. Furthermore, during movement, factors such as uneven ground, uneven load, or sudden changes in the trajectory can cause momentary imbalances in the overall structure, resulting in vibration. This vibration can not only affect the robot's stability and accuracy but also negatively impact its lifespan.

[0003] Current robot control systems have certain limitations in handling low-speed jitter, resulting in poor stability of the robot during low-speed movement. Summary of the Invention

[0004] The main objective of this application is to provide a control method, a control device, and a computer-readable storage medium for low-speed robot movement, so as to at least solve the problem of poor stability of low-speed robot movement in the prior art.

[0005] To achieve the above objectives, according to one aspect of this application, a method for controlling the low-speed motion of a robot is provided, comprising: acquiring an original position deviation signal of the robot when performing a low-speed task, and converting the original position deviation signal into a frequency domain signal; processing the frequency domain signal using a Chebyshev low-pass filter to obtain a filtered frequency domain signal; converting the filtered frequency domain signal into a time domain signal through an inverse Fourier transform to obtain a filtered time domain signal, and adjusting the robot's velocity loop gain until a preset stopping condition is reached to obtain an adjusted velocity loop gain; and controlling the robot's motion using the filtered time domain signal and the adjusted velocity loop gain.

[0006] Optionally, the frequency domain signal is processed using a Chebyshev low-pass filter to obtain a filtered frequency domain signal, including: adjusting the increment of the filtering time constant using a gradient descent algorithm until a preset convergence criterion is reached to obtain a target filtering time constant; setting the cutoff frequency of the Chebyshev low-pass filter based on the target filtering time constant; comparing the frequency domain signal with the cutoff frequency, and removing the frequency domain signal higher than the cutoff frequency to obtain the filtered frequency domain signal.

[0007] Optionally, adjusting the increment of the filtering time constant using a gradient descent algorithm until a preset convergence criterion is reached to obtain the target filtering time constant includes: calculating the gradient of the objective function under the current filtering time constant using the gradient descent algorithm, wherein the objective function is determined based on the integral of the squared error; adjusting the increment of the filtering time constant using a learning rate based on the gradient under the current filtering time constant until the preset convergence criterion is reached to obtain the target filtering time constant, wherein the preset convergence criterion indicates that the increment of the filtering time constant is less than a preset convergence threshold.

[0008] Optionally, adjusting the robot's velocity loop gain until a preset stopping condition is reached to obtain the adjusted velocity loop gain includes: calculating an objective function for adjusting the velocity loop gain based on the filtered time-domain signal, wherein the objective function for the velocity loop gain is set based on the position deviation and response time of the robot during movement; calculating the gradient of the current velocity loop gain according to the objective function for the velocity loop gain; updating the velocity loop gain according to the gradient of the current velocity loop gain and a preset learning rate until the preset stopping condition is reached to obtain the adjusted velocity loop gain, wherein the preset stopping condition is set based on the difference between two consecutive updates of the velocity loop gain and a preset threshold.

[0009] Optionally, acquiring the original position deviation signal of the robot when performing a low-speed task includes: a control step, in which the robot is controlled to move to a preset test point when performing the low-speed task, the preset test point including a target position within the working range of the robot; an acquisition step, in which the actual position of the robot is acquired when the robot reaches the preset test point; and a calculation step, in which the deviation between the actual position and the target position is calculated to obtain the original position deviation signal.

[0010] Optionally, after calculating the deviation between the actual position and the target position to obtain the original position deviation signal, the method further includes: selecting multiple preset test points within the working range of the robot, repeating the control step, the acquisition step, and the calculation step to obtain multiple original position deviation signals; generating a position deviation distribution map based on the multiple original position deviation signals, and analyzing the positioning accuracy of the robot within the working range according to the position deviation distribution map.

[0011] Optionally, after calculating the deviation between the actual position and the target position to obtain the original position deviation signal, the method further includes: controlling the robot to perform a preset motion to obtain the actual position of the robot under the preset motion, wherein the preset motion includes linear motion and circular trajectory motion; setting the target position of the robot under the preset motion; calculating the deviation between the actual position and the target position of the robot at the target point in the preset motion to obtain the target deviation, wherein the target point includes a starting point, an ending point, and an inflection point; and evaluating the positioning accuracy of the robot within the working range based on the target deviation.

[0012] Optionally, after calculating the deviation between the actual position and the target position to obtain the original position deviation signal, the method further includes: performing preprocessing operations on the original position deviation signal, the preprocessing operations including noise reduction, data smoothing, and format conversion.

[0013] According to another aspect of this application, a control device for low-speed robot movement is provided, comprising: an acquisition unit for acquiring an original position deviation signal of the robot when performing a low-speed task, and converting the original position deviation signal into a frequency domain signal; a processing unit for processing the frequency domain signal using a Chebyshev low-pass filter to obtain a filtered frequency domain signal; an adjustment unit for converting the filtered frequency domain signal into a time domain signal through an inverse Fourier transform to obtain a filtered time domain signal, and adjusting the robot's velocity loop gain until a preset stopping condition is reached to obtain an adjusted velocity loop gain; and a first control unit for controlling the robot's movement using the filtered time domain signal and the adjusted velocity loop gain.

[0014] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform any of the control methods for low-speed robot movement described above.

[0015] This application uses a Chebyshev low-pass filter to obtain the original position deviation signal of a robot performing a low-speed task, and converts the original position deviation signal into a frequency domain signal. A Chebyshev low-pass filter is used to process the frequency domain signal, resulting in a filtered frequency domain signal. The filtered frequency domain signal is then converted into a time domain signal using an inverse Fourier transform, resulting in a filtered time domain signal. The robot's velocity loop gain is adjusted until a preset stopping condition is met, resulting in an adjusted velocity loop gain. The filtered time domain signal and the adjusted velocity loop gain are used to control the robot's motion. In this scheme, by applying a Chebyshev low-pass filter in the frequency domain, high-frequency noise in the position deviation signal during low-speed robot operation is effectively filtered out. Simultaneously, the filter's cutoff frequency and filtering time constant are optimized, improving the signal's smoothness and stability. By adjusting the velocity loop gain to its optimal state, stable response is maintained while increasing bandwidth, avoiding oscillations and abnormal noises caused by over-adjustment. Using the filtered signal and the optimized velocity loop gain to control the robot's motion reduces jitter and improves stability, thus solving the problem of poor stability in low-speed robot motion. Attached Figure Description

[0016] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0017] Figure 1 A hardware structure block diagram of a mobile terminal for performing a control method for low-speed robot movement according to an embodiment of this application is shown.

[0018] Figure 2 A flowchart illustrating a control method for low-speed robot movement according to an embodiment of this application is shown.

[0019] Figure 3 A flowchart is shown illustrating a specific method for controlling low-speed robot motion according to an embodiment of this application;

[0020] Figure 4 The diagram shows a first sampled position deviation waveform of a specific robot low-speed motion control method according to an embodiment of this application;

[0021] Figure 5 The diagram shows a second acquisition of position deviation waveforms for a specific robot low-speed motion control method according to an embodiment of this application;

[0022] Figure 6 The diagram illustrates a third acquisition of position deviation waveforms for a specific robot low-speed motion control method provided according to an embodiment of this application;

[0023] Figure 7 The diagram shows a fourth acquisition of position deviation waveforms for a specific robot low-speed motion control method provided according to an embodiment of this application;

[0024] Figure 8 A structural block diagram of a control device for low-speed robot movement provided according to an embodiment of this application is shown.

[0025] The above figures include the following reference numerals:

[0026] 102. Processor; 104. Memory; 106. Transmission device; 108. Input / output device. Detailed Implementation

[0027] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0028] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0030] As described in the background section, existing robot control systems have certain limitations in handling low-speed jitter. To address the problem of poor stability of robots during low-speed movement, embodiments of this application provide a control method for low-speed robot movement, a control device for low-speed robot movement, and a computer-readable storage medium.

[0031] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0032] The methods and embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for a robot low-speed motion control method according to an embodiment of the present invention. Figure 1 As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0033] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the robot low-speed motion control method in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0034] This embodiment provides a control method for low-speed robot movement that runs on a mobile terminal, computer terminal, or similar computing device. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Also, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0035] Figure 2 This is a flowchart illustrating a control method for low-speed robot movement according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0036] Step S201: Obtain the original position deviation signal of the robot when performing a low-speed task, and convert the original position deviation signal into a frequency domain signal;

[0037] Specifically, the robot in this embodiment is an industrial robot. When the robot performs low-speed tasks, the inherent characteristics of its mechanical components can cause vibrations and positional deviations, which affect the robot's accuracy and stability. Therefore, it is first necessary to acquire the original signal reflecting the deviation between the robot's actual position and the desired (target) position, i.e., the original position deviation signal, for subsequent analysis and optimization. Low-speed tasks often involve high precision requirements. For example, in scenarios such as precision assembly, electronic component welding, or handling of tiny objects, the robot needs to perform precise operations within a micrometer-level or smaller range. In these cases, the robot's speed needs to be sufficiently low to ensure the accuracy and stability of position control. In human-robot collaborative environments or situations with strict safety requirements for the surrounding environment, such as medical surgery assistance or nuclear facility maintenance, low-speed operation can reduce the risk of accidental collisions and improve operational safety. Low-speed tasks may also involve special requirements for dynamic response. In certain sensitive or rapidly changing environments, robot control needs to respond quickly and accurately to environmental changes; lower speeds can provide more controllable dynamic performance. The specific numerical range of "low speed" for a low-speed task needs to consider the robot type, application field, maximum speed capability, and task requirements. For example, for an industrial robot with a maximum speed of 500 mm / s, the speed for performing low-speed tasks is defined as 10% to 30% of the maximum speed, i.e., between 50 mm / s and 150 mm / s. This range is generally considered low-speed operation, designed to ensure the robot's stability when performing high-precision operations. In scenarios with more delicate or safety requirements, the speed definition for low-speed tasks is even lower; for example, in precision electronic assembly or operations in sensitive environments, the speed is limited to 1% to 10% of the maximum speed, or even lower.

[0038] The original position deviation signal is a time-domain signal, directly reflecting the robot's position error at a specific moment. Converting the original time-domain position deviation signal into a frequency-domain signal allows for a more intuitive understanding and manipulation of the different frequency components within the signal, which is crucial for identifying and suppressing high-frequency noise that causes jitter. Frequency-domain analysis enables focused processing of signals within a specific frequency range, enhancing the understanding of the signal's essential characteristics. The Fast Fourier Transform (FFT) is used to convert the time-domain signal into a frequency-domain signal. The FFT reveals the frequency components and amplitude of the signal, making it possible to identify which frequency components significantly contribute to jitter. In the frequency domain, the signal is decomposed into a superposition of its constituent frequencies, allowing for filtering within a specific frequency range to suppress unwanted high-frequency noise while retaining low-frequency signals effective for robot motion.

[0039] Step S201 not only acquires the original position deviation signal of the robot performing low-speed tasks, but also converts it to the frequency domain. This facilitates subsequent targeted filtering using a Chebyshev low-pass filter, ensuring that the signal contains only frequency components beneficial to robot motion control, thereby improving the robot's stability and accuracy under low-speed conditions. This process ensures that the robot can complete the predetermined task more smoothly and accurately, reducing position deviations caused by mechanical vibration.

[0040] Step S202: The frequency domain signal is processed using a Chebyshev low-pass filter to obtain the filtered frequency domain signal.

[0041] Specifically, a Chebyshev low-pass filter is a filter that exhibits equal ripple characteristics in the passband and steep attenuation in the stopband. Its main advantage lies in its ability to provide a very steep transition band under limited passband ripple conditions. This means that low-frequency signals can be effectively distinguished and preserved, while high-frequency signals are strongly suppressed, enabling precise signal separation in the frequency domain.

[0042] Applying a Chebyshev low-pass filter to a frequency domain signal attenuates or removes signal components above the cutoff frequency. This process essentially "slices" the signal in the frequency spectrum, retaining the low-frequency components and discarding the high-frequency components. After filtering, the resulting frequency domain signal will only contain signal components below the preset cutoff frequency. These signals are smoother in the frequency domain, and high-frequency noise and jitter are significantly suppressed.

[0043] The key to step S202 lies in utilizing the characteristics of a Chebyshev low-pass filter to effectively reduce the high-frequency components in the original position deviation signal. These high-frequency components are often the cause of low-speed jitter in robots. Filtering not only improves signal quality but also indirectly enhances the robot's position accuracy and motion stability. In short, using a Chebyshev low-pass filter to process the frequency domain signal optimizes the position deviation control of industrial robots performing low-speed tasks from a frequency domain perspective through signal separation and noise suppression, reducing errors caused by high-frequency jitter and enhancing the smoothness and accuracy of robot motion.

[0044] Step S203: Convert the filtered frequency domain signal into a time domain signal through inverse Fourier transform to obtain the filtered time domain signal, and adjust the speed loop gain of the robot until the preset stopping condition is reached to obtain the adjusted speed loop gain.

[0045] Specifically, after applying a Chebyshev low-pass filter, a filtered frequency domain signal is obtained, in which high-frequency noise and jitter components have been effectively suppressed, the signal is smoother, and it contains only low-frequency components. To reuse the filtered frequency domain signal in the time domain (i.e., in actual robot operation), an inverse Fourier transform is used to convert the frequency domain signal back to the time domain. This process converts the signal from a frequency perspective to a time perspective, allowing direct observation and utilization of the signal's changes along the time axis. The resulting inverse Fourier transform yields a filtered time domain signal that contains only the low-frequency components filtered out by the Chebyshev low-pass filter. These components are more relevant to actual robot motion control, helping to improve positional accuracy and reduce jitter.

[0046] Velocity loop gain is a key parameter in robot motion control systems, determining the system's sensitivity to speed commands and control accuracy. A higher velocity loop gain results in a faster response to speed changes, but may also increase the likelihood of oscillations. Based on the filtered time-domain signal, the robot's velocity loop gain is adjusted in an iterative process involving the optimization of the objective function (e.g., the integral of the squared error of position deviation). The gain value is gradually modified using gradient descent or similar algorithms until a preset stopping condition is met. Through repeated adjustments and optimizations, an adjusted velocity loop gain is obtained. This adjusted velocity loop gain, together with the filtered time-domain signal, provides more stable and accurate robot motion control, especially in low-speed, high-precision operation scenarios, significantly reducing jitter and improving the stability of position control.

[0047] By converting the filtered frequency domain signal into a time domain signal through inverse Fourier transform and combining it with the adjusted velocity loop gain, precise control of positional deviations during low-speed industrial robot movement can be achieved. This process not only removes high-frequency noise from the signal but also optimizes the control system's response to speed commands, thereby improving the robot's overall motion stability and positional accuracy.

[0048] Step S204: Use the filtered time-domain signal and the adjusted velocity loop gain to control the robot's motion.

[0049] Specifically, through the Chebyshev low-pass filter processing and inverse Fourier transform described above, the filtered time-domain signal has removed high-frequency noise and jitter components, retaining only the low-frequency signals relevant to the robot's actual motion control. This means the signal is cleaner and more suitable for guiding the robot's precise motion. The filtered time-domain signal is used as part of the position control command to adjust the robot's motion in real time. Because the filtered time-domain signal contains fewer useless high-frequency components, the robot can more accurately follow commands during control, avoiding unnecessary jitter and position deviations, and achieving higher-quality motion trajectory control.

[0050] Adjusting the velocity loop gain is crucial for balancing the sensitivity and stability of the system response. An adjusted velocity loop gain maximizes velocity response while ensuring system stability, meaning the robot can respond more quickly and directly to changes in position control commands. Applying the adjusted velocity loop gain to velocity loop control significantly improves the robot's motion quality during low-speed operations. Higher gain translates to less hysteresis and more precise velocity tracking, thereby reducing positional deviations and improving the smoothness and repeatability of motion.

[0051] A new control strategy is formed by combining the filtered time-domain signal with the adjusted velocity loop gain. Based on this strategy, a cleaner signal and optimized gain parameters are used to guide the robot's motion. In actual motion control, the robot continuously receives and processes the filtered control signal and responds according to the adjusted velocity loop gain. Position deviations are monitored in real time, and the speed is continuously adjusted to compensate for the deviations, ensuring that the robot maintains high accuracy and stability even at low speeds.

[0052] By utilizing filtered time-domain signals and adjusted velocity loop gain, more precise and stable control of industrial robot motion can be achieved. This strategy not only eliminates noise interference common in high-speed operation but also optimizes the response to speed commands, thereby significantly improving the robot's positional accuracy and smoothness in low-speed operation, providing strong technical support for achieving high-precision, low-jitter robot motion.

[0053] In this embodiment, a Chebyshev low-pass filter is applied in the frequency domain to effectively filter out high-frequency noise in the position deviation signal when the robot is running at low speed. At the same time, the cutoff frequency and filtering time constant of the filter are optimized to improve the smoothness and stability of the signal. By adjusting the speed loop gain to the optimal state, it is ensured that stable response can be maintained while increasing bandwidth, avoiding oscillation and abnormal noise caused by over-adjustment. The robot's motion is controlled by the filtered signal and the optimized speed loop gain, which reduces jitter and improves stability, thereby solving the problem of poor stability of robot movement at low speed.

[0054] In the specific implementation process, the Chebyshev low-pass filter is used to process the above frequency domain signal to obtain the filtered frequency domain signal. This includes: adjusting the increment of the filtering time constant using a gradient descent algorithm until a preset convergence criterion is reached to obtain the target filtering time constant; setting the cutoff frequency of the Chebyshev low-pass filter based on the target filtering time constant; comparing the above frequency domain signal with the cutoff frequency, and removing the above frequency domain signal that is higher than the cutoff frequency to obtain the filtered frequency domain signal.

[0055] Furthermore, the gradient descent algorithm is used to adjust the increment of the filtering time constant until a preset convergence criterion is reached to obtain the target filtering time constant. This includes: calculating the gradient of the objective function under the current filtering time constant using the gradient descent algorithm, wherein the objective function is determined based on the integral of the squared error; and adjusting the increment of the filtering time constant using the learning rate based on the gradient under the current filtering time constant until the preset convergence criterion is reached to obtain the target filtering time constant. The preset convergence criterion indicates that the increment of the filtering time constant is less than a preset convergence threshold.

[0056] Specifically, the objective function is primarily based on the integral of the squared error, a crucial indicator for measuring the degree of error accumulation in control performance. In this embodiment, the error refers to the deviation between the robot's actual position and the target position, quantified by integrating the squared deviations at all time points. The gradient is the derivative vector of the objective function at a point, indicating the direction of the fastest increase in the function. The gradient descent algorithm requires calculating the gradient of the objective function at the current filter time constant, involving mathematical operations and differential calculations. The gradient calculation results help understand the changing trend of the objective function value when adjusting the filter time constant, providing a basis for further adjustments.

[0057] The filter time constant is a crucial parameter determining the filter's cutoff frequency, directly impacting the suppression of high-frequency components in the signal. The initial filter time constant may not be optimal, thus requiring gradual adjustment. Gradient descent is an optimization algorithm used to find the minimum point of the loss function (in this example, the integral of the squared error of the positional deviation). By calculating the gradient (i.e., derivative) of the loss function under the current filter time constant, the direction and magnitude of the time constant adjustment can be determined, thereby gradually approaching the optimal filter time constant.

[0058] The learning rate is a hyperparameter that controls the step size of the filter time constant adjustment. A high learning rate may lead to overly rapid adjustments, easily skipping the optimal solution; a low learning rate may slow down the algorithm's convergence speed. Setting a reasonable learning rate can balance adjustment speed and accuracy, helping the gradient descent algorithm efficiently find the minimum point of the objective function.

[0059] Based on the gradient and learning rate at the current filtering time constant, the increment of the filtering time constant is adjusted, i.e., determining the direction and magnitude of the adjustment. Through continuous iteration, the optimal filtering time constant is gradually approached. A preset convergence criterion is one of the conditions for terminating the gradient descent algorithm, determined by whether the increment of the filtering time constant is less than a preset convergence threshold. When it is detected that the adjustment of the filtering time constant is very small in several consecutive iterations, i.e., the increment is less than the preset convergence threshold, it indicates that the optimal solution has been approached or reached. At this point, the iteration can be stopped, and the obtained filtering time constant is the target filtering time constant.

[0060] Adjusting the filter time constant using the gradient descent algorithm enables automatic optimization in complex control environments, reducing the complexity and uncertainty of manual parameter tuning. Specific steps include calculating the gradient of the objective function, adjusting the time constant using the learning rate, and setting reasonable convergence criteria to ensure timely termination after finding a filter time constant that effectively improves robot motion stability and positional accuracy, avoiding over-optimization or getting trapped in local optima. This not only provides a quantitative and scientific method for parameter optimization but also adapts to changes in different working environments, enabling automatic optimization and improvement of the low-speed motion control performance of industrial robots.

[0061] Once the target filtering time constant is determined, the cutoff frequency of the Chebyshev low-pass filter can be calculated. The selection of the cutoff frequency ensures that signal components below this frequency are preserved, while high-frequency components above it are suppressed, which is crucial for reducing robot jitter. Based on the calculated cutoff frequency, the Chebyshev low-pass filter is designed and configured to effectively filter out unwanted high-frequency signal components, while allowing low-frequency signals to pass through, maintaining the basic signal structure and the information required for control. The original position deviation signal, converted to frequency domain form, is compared with the set filter cutoff frequency. Any signal components with frequencies higher than the cutoff frequency will be identified and removed by the filter to prevent these high-frequency components from interfering with the robot's stability.

[0062] After the above steps, a filtered frequency domain signal is obtained, retaining only low-frequency components while effectively suppressing high-frequency jitter and noise. These filtered signals will be transformed into more stable and accurate time-domain control signals in the subsequent inverse Fourier transform process, used to guide and optimize the robot's motion control. In summary, by precisely adjusting the filtering time constant using the gradient descent algorithm to find the parameter value that best reduces robot jitter, and then setting the cutoff frequency of the Chebyshev low-pass filter based on the target filtering time constant, high-frequency components in the frequency domain signal are effectively removed. The resulting filtered frequency domain signal will be used to generate subsequent time-domain signals, providing cleaner and more stable position control commands, thereby significantly improving the motion accuracy and stability of industrial robots under low-speed conditions.

[0063] In some embodiments of this application, an adaptive filter parameter adjustment algorithm is introduced. This algorithm can dynamically monitor the robot's position deviation and vibration during low-speed movement, and automatically adjust the filter time constant and cutoff frequency to adapt to the current working conditions. The core of the algorithm includes: Real-time vibration monitoring: Using sensors to collect vibration signals from the robot in real time during task execution and quickly analyzing the vibration spectrum. Dynamic parameter adjustment: Based on the spectral characteristics of the vibration signal, the algorithm automatically calculates the optimal filter time constant and cutoff frequency, and updates the filter parameters in real time. Feedback control mechanism: Establishing a feedback control loop, comparing the adjusted filter output with the position deviation signal in real time, further optimizing the parameters, and ensuring minimal jitter. The adaptive filter parameter adjustment algorithm can significantly improve the robot's control stability in changing environments, reduce jitter and position deviation caused by changes in working conditions, and improve the accuracy and reliability of robot operations.

[0064] In some embodiments of this application, adjusting the speed loop gain of the robot until a preset stopping condition is reached to obtain the adjusted speed loop gain includes: calculating an objective function for adjusting the speed loop gain based on the filtered time-domain signal, wherein the objective function for the speed loop gain is set based on the position deviation and response time of the robot during movement; calculating the gradient of the current speed loop gain according to the objective function for the speed loop gain; updating the speed loop gain according to the gradient of the current speed loop gain and a preset learning rate until the preset stopping condition is reached to obtain the adjusted speed loop gain, wherein the preset stopping condition is set based on the difference between two consecutive updates of the speed loop gain and a preset threshold.

[0065] Specifically, when adjusting the velocity loop gain, an objective function needs to be defined, based on the robot's position deviation and response time during motion. Position deviation reflects the difference between the robot's actual position and the target position, while response time indicates how quickly the robot responds to control commands. Optimizing the objective function aims to minimize position deviation and ensure a fast response, thereby improving motion control performance. The calculation of the objective function relies on the filtered time-domain signal. Since high-frequency noise and jitter have been effectively removed, these cleaner signals more accurately reflect the position deviation and response characteristics in robot motion control, providing an accurate data basis for subsequent gain adjustments.

[0066] The gradient is the derivative vector of the objective function with respect to the current velocity loop gain, indicating the rate of change of the objective function value in the direction of gain adjustment. Calculating the gradient reveals how gain adjustment affects position deviation and response time, providing a basis for decision-making. Gradient calculation is derived by differentiating the objective function relative to the velocity loop gain. This involves mathematical analysis and algorithm implementation, including numerical or analytical differentiation methods, depending on the complexity and differentiability of the objective function. The preset learning rate controls the step size of the velocity loop gain adjustment. The preset learning rate is a hyperparameter used to balance adjustment speed and accuracy, ensuring the gain approaches the optimal value at an appropriate rate while avoiding system oscillations caused by over-adjustment.

[0067] Based on the gradient at the current velocity loop gain and a preset learning rate, the gain value can be updated using gradient descent. Each update adjusts the gain in the opposite direction of the gradient to reduce the objective function value, thus reducing positional bias and accelerating the response. A preset stopping condition is used to determine whether the gain adjustment should terminate. The preset stopping condition is based on the difference between two consecutive velocity loop gain updates being less than a preset threshold. When the difference is sufficiently small, indicating a significant slowdown in the adjustment speed and the gain approaching its optimal value, the iteration stops, and the current gain value is determined as the adjusted velocity loop gain.

[0068] Through the above process, the objective function is calculated using the filtered time-domain signal, and then the velocity loop gain is adjusted using a gradient descent algorithm until a preset stopping condition is met. This enables automatic optimization and dynamic calibration of the robot's motion control performance. The resulting adjusted velocity loop gain not only significantly reduces positional deviations during low-speed motion but also ensures stability even at high gains, avoiding abnormal noises and current hum, thus achieving high precision and stability in motion control. In practical applications, it can adapt to different working conditions and loads, providing strong technical support for industrial robots operating in complex environments.

[0069] In some other embodiments of this application, obtaining the original position deviation signal of a robot when performing a low-speed task includes: a control step, in which the robot is controlled to move to a preset test point when performing the low-speed task, the preset test point including a target position within the working range of the robot; an acquisition step, in which the actual position of the robot is acquired when the robot reaches the preset test point; and a calculation step, in which the deviation between the actual position and the target position is calculated to obtain the original position deviation signal.

[0070] Specifically, the control steps ensure the robot is performing a low-speed task to collect real-world performance data under these conditions, particularly regarding positional deviations. Pre-set test points are distributed across several target locations within the robot's working range. These test points should cover different areas of the robot's operating space to comprehensively evaluate its performance during low-speed movement. Control commands guide the robot from its current position to each pre-set test point, simulating low-speed movement in real-world conditions, such as welding or assembly. The control steps prepare the necessary conditions for the acquisition steps: the robot is performing a low-speed task and has accurately reached the pre-set test points, laying the foundation for further data acquisition and performance analysis.

[0071] After the robot successfully reaches the preset test point, the data acquisition system is immediately activated to record the robot's actual position data at that location. This data comes from the robot's position sensors and can accurately reflect the robot's actual position coordinates in space. The acquisition step is a crucial step in collecting the raw position deviation signal, directly affecting the quality and accuracy of the deviation signal in subsequent calculation steps. Through this step, the most realistic position information of the robot under low-speed motion can be obtained for subsequent performance evaluation and optimization.

[0072] Based on the acquired actual position data and the coordinate information of the preset test points (i.e., target positions), the position deviation between the two is calculated. The deviation is calculated by determining the Euclidean distance between the two sets of coordinates to obtain the robot's position deviation value. The calculated deviation value is then serialized to form the raw position deviation signal. This signal contains the position deviation information of all preset test points throughout the robot's low-speed task execution and is a crucial data source for subsequent filtering and gain adjustment.

[0073] The calculation of the raw position deviation signal provides an intuitive quantitative indicator of robot performance, especially positional accuracy during low-speed motion. It serves as the starting point for control strategy optimization. By analyzing and processing the raw position deviation signal, filter parameters and velocity loop gain can be adjusted to effectively reduce low-speed jitter and improve the robot's position control accuracy and stability when performing low-speed tasks. In summary, the above steps accurately capture the robot's true position deviation performance during low-speed task execution, thereby guiding system optimization. This series of operations not only provides solid data support for the subsequent application of Chebyshev low-pass filters and velocity loop gain adjustment but also enhances the intelligence and precision of the robot's low-speed motion control, effectively improving its performance in low-speed, high-precision operation scenarios and promoting overall robot performance improvement.

[0074] In some embodiments of this application, deep learning, particularly Long Short-Term Memory (LSTM) networks or Convolutional Neural Networks (CNNs), is used to predict and compensate for positional deviations of the robot during low-speed motion. Specifically, a training set is constructed using a large amount of historical positional deviation signals and working condition data, covering different loads, speeds, and environmental factors. LSTM or CNNs are used to perform deep learning on the training set to identify potential patterns of positional deviations and their correlation with working conditions. While the robot is performing a task, the current working conditions are input into the trained deep learning model in real time to predict potential future positional deviations. Based on the predicted positional deviations, the robot's control commands are adjusted in advance, and compensation strategies are applied to minimize the positional deviations during actual motion. Deep learning-based positional deviation prediction and compensation can effectively handle complex problems in nonlinear dynamic models, reduce positional deviations caused by changes in system dynamic characteristics, and further improve the stability and accuracy of the robot in low-speed, high-precision task execution. It is particularly suitable for situations where control parameters are difficult to predict accurately using traditional mathematical models.

[0075] Furthermore, after calculating the deviation between the actual position and the target position to obtain the original position deviation signal, the method further includes: selecting multiple preset test points within the working range of the robot, repeating the control steps, the acquisition steps, and the calculation steps to obtain multiple original position deviation signals; generating a position deviation distribution map based on the multiple original position deviation signals, and analyzing the positioning accuracy of the robot within the working range based on the position deviation distribution map.

[0076] Specifically, multiple pre-selected test points are carefully chosen within the robot's working range. These points should be evenly distributed or focus on areas prone to positioning deviations during robot operation to ensure that the collected data comprehensively reflects its performance. For each pre-selected test point, the control steps (i.e., controlling the robot to move to the test point), acquisition steps (i.e., recording the robot's actual position at the test point), and calculation steps (i.e., calculating the deviation between the actual position and the target position) are repeatedly executed to generate raw position deviation signals corresponding to each test point. Through multi-point testing, multiple raw position deviation signals are accumulated. These raw position deviation signals collectively depict the overall positioning accuracy of the robot within its working range, providing rich and detailed data support for subsequent analysis.

[0077] Based on multiple collected raw position deviation signals, statistical analysis methods are used to construct a position deviation distribution map. The distribution map, presented in the form of a chart, heatmap, or scatter plot, shows the robot's position deviation at different test points. The position deviation distribution map transforms abstract deviation values ​​into visual graphical information, making the distribution of the robot's positioning error within its working range readily apparent. This helps to quickly identify areas with large deviations, providing direction for subsequent performance optimization.

[0078] A detailed analysis of the position deviation distribution map allows for the assessment of the robot's positioning accuracy in different areas, identifying performance bottlenecks and strengths. For example, test points with large deviations indicate control difficulties at specific locations, while points with smaller deviations reflect the robot's strong positioning capabilities in those areas. Analyzing positioning accuracy not only helps in understanding the robot's current performance state but also guides subsequent adjustments to filter parameters and velocity loop gain. By implementing targeted optimization measures for areas with large positioning deviations, the overall positioning accuracy and stability of the robot can be significantly improved.

[0079] By precisely controlling the robot to move to a preset test point and recording the deviation between its actual and target positions, a raw position deviation signal is generated. This process is repeated to collect multiple raw position deviation signals covering key locations within the working range, and a position deviation distribution map is constructed based on these signals. The position deviation distribution map visually demonstrates the robot's positioning accuracy at different positions, providing precise guidance for performance optimization. By analyzing the position deviation distribution map, areas with large positioning deviations can be identified, allowing for corresponding measures, such as optimizing filtering parameters or adjusting the velocity loop gain, to reduce position deviations and improve overall positioning accuracy and stability. This process ensures that the robot achieves higher control precision even at low speeds.

[0080] In some embodiments of this application, after calculating the deviation between the actual position and the target position to obtain the original position deviation signal, the method further includes: controlling the robot to perform a preset movement to obtain the actual position of the robot under the preset movement, wherein the preset movement includes linear movement and circular trajectory movement; setting the target position of the robot under the preset movement; calculating the deviation between the actual position and the target position of the robot at the target point in the preset movement to obtain the target deviation, wherein the target point includes a start point, an end point, and an inflection point; and evaluating the positioning accuracy of the robot within the working range based on the target deviation.

[0081] Specifically, firstly, a series of preset movements are set for the robot, including linear and circular trajectories. These preset movement paths cover different directions and areas within the robot's working range, thus comprehensively testing its positioning capabilities under dynamic conditions. While the robot executes the preset movement paths, sensors or a positioning system record its actual position during the movement. In particular, the coordinates of key positions such as the starting point, ending point, and inflection points along the path are recorded, which helps to accurately analyze the robot's positional deviation under dynamic conditions. Target positions are set for each key point (target point) on the preset movement path. These target positions represent the theoretically ideal positions, i.e., the accurate coordinates that the robot should reach. The deviation between the robot's actual position and the target position on the preset movement path is calculated to obtain the target deviation for each key point (target point). This target deviation is an important indicator for evaluating the robot's positioning accuracy, especially during dynamic movement.

[0082] By analyzing all target deviations, the robot's dynamic positioning accuracy across its entire working range can be evaluated, including its performance in linear and circular trajectories. This evaluation reflects not only the accuracy of static positioning but also the control stability under dynamic conditions. The target deviation data provides direct evidence for improving the robot's motion control performance. In-depth analysis of points with significant deviations can identify potential problems in motion control, such as deficiencies in the control algorithm or limitations in the mechanical structure, allowing for appropriate optimization measures to be taken.

[0083] By controlling the robot to execute preset straight and circular trajectories, position deviation signals during dynamic processes can be accurately acquired, allowing for a comprehensive evaluation of its positioning accuracy within its working range. This approach not only covers deviation measurements at static points but also incorporates analysis of dynamic path motion deviations, resulting in a more holistic assessment. By setting target positions at key points along the motion path and calculating the deviation between the robot's actual position and the target position, a series of target deviation data are obtained. These data reflect the robot's positioning accuracy when executing complex motion commands. The evaluation based on target deviations reveals performance bottlenecks in the robot control system under dynamic conditions, providing specific guidance for subsequent parameter optimization (such as filter adjustment and velocity loop gain settings), thereby significantly improving the robot's positioning capabilities in low-speed, high-precision task execution.

[0084] Furthermore, after calculating the deviation between the actual position and the target position to obtain the original position deviation signal, the method further includes: performing preprocessing operations on the original position deviation signal, including noise reduction, data smoothing, and format conversion.

[0085] Specifically, the purpose of denoising is to eliminate random fluctuations in the signal introduced by factors such as sensor noise, environmental interference, or system instability. These fluctuations may mask the true positional deviation information, leading to distorted subsequent analysis results. Signal processing techniques, such as low-pass filters, median filters, or wavelet denoising, can be applied to filter out high-frequency noise and retain the basic trend of the original positional deviation signal. Data smoothing aims to further reduce short-term fluctuations in the signal, making the data more continuous and smooth. This is crucial for subsequent analysis, especially trend-based analysis and parameter tuning. Techniques such as moving averages, Savitzky-Golay filters, or exponentially weighted moving averages are used to process the original positional deviation signal to smooth the signal curve, making it closer to the true trend. Format conversion aims to ensure that data is stored and processed in appropriate formats and units, facilitating subsequent mathematical operations and system integration. Format standardization is particularly important for cross-system data exchange and analysis. The original deviation signal is converted from its original format (such as sensor-specific format) to a standardized data format, while adjusting the units (such as converting from non-standard units to millimeters or micrometers) to meet the needs of subsequent calculations and evaluations.

[0086] Preprocessing significantly reduces the impact of noise, smooths the signal curve, and makes the signal more clearly reflect the robot's true position deviation. This high-quality signal data provides a solid foundation for subsequent position deviation analysis, filter parameter optimization, and velocity loop gain settings, ensuring the accuracy and effectiveness of robot control system adjustments and ultimately improving the robot's positioning accuracy and operational stability. Preprocessing not only enhances signal usability but also simplifies the complexity of subsequent processing.

[0087] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of the robot low-speed motion control method of this application will be described in detail below with reference to specific embodiments.

[0088] This embodiment relates to a specific control method for low-speed robot movement, such as... Figure 3 As shown, it includes signal acquisition, signal conversion, cutoff frequency setting, frequency domain processing, signal conversion back to time domain, and signal output.

[0089] Working principle: The core function of a Chebyshev low-pass filter is to allow low-frequency signals to pass through while suppressing high-frequency signals. A steeper transition band is achieved by introducing controllable ripples in the passband or stopband, thus providing higher performance in specific applications. A suitable filter amplitude square function is... Here, ε is a positive number less than 1, and the value of ε represents the size of the passband ripple. Is ω to ω P The normalized frequency, ωP Here is the passband cutoff frequency. Based on the above equation and the properties of the Chebyshev polynomial, the Nth-order normalized low-pass Chebyshev filter has the following fundamental characteristics:

[0090] When ω = 0 and N is even... When N is odd, |H(j0)|=1.

[0091] When ω=ω P At that time, that is | At that time, all amplitude response curves passed through Point, therefore ω P is the passband cutoff frequency of the Chebyshev filter.

[0092] When ω≤ω P At that time, The undulations are like waves.

[0093] When ω≥ω P hour, As ω increases, |H(jω)| will rapidly and monotonically decrease to zero.

[0094] The filter time constant is inversely proportional to the cutoff frequency; a smaller filter time constant results in a larger cutoff frequency. Conversely, a larger filter time constant leads to slower capacitor charging and discharging, and a slower filter response. Therefore, reducing robot jitter requires reducing the filter time constant. Reducing the filter time constant increases the cutoff frequency, thereby increasing the speed loop bandwidth. This is because the filter time constant determines the filter's response speed and cutoff frequency. Speed ​​loop bandwidth refers to the frequency range within which the system can effectively track speed commands. Increasing the filter's cutoff frequency means the system can respond to higher frequency speed changes, thus expanding the speed loop bandwidth. It is precisely because the speed loop bandwidth is expanded that adjusting the speed loop gain has a significant effect.

[0095] Filtering steps:

[0096] (1) Signal acquisition: Acquiring raw signals from sensors or other sources. This signal may contain multiple frequency components, including low and high frequencies.

[0097] (2) Signal conversion: Converting the time-domain signal to the frequency domain, usually using Fourier transform (such as Fast Fourier Transform, FFT). This step allows us to analyze and manipulate the frequency components of the signal.

[0098] (3) Filter design: Design a low-pass filter according to application requirements and set a cutoff frequency. Signals below this frequency will be retained, and signals above this frequency will be attenuated or suppressed.

[0099] (4) Frequency domain processing: A low-pass filter is applied in the frequency domain to attenuate or reduce components above the cutoff frequency to zero. This step ensures that only low-frequency signals pass through, while high-frequency noise or interference is removed.

[0100] (5) Signal conversion back to time domain: The processed frequency domain signal is converted back to the time domain through inverse Fourier transform to obtain the filtered signal.

[0101] (6) Signal Output: The filtered signal is output for subsequent processing or applications, such as improving the stability of the control system or removing noise in audio processing. Establish Data Buffer: First, a fixed-length data buffer is established in memory (RAM) to store the N most recently acquired data samples.

[0102] With other parameters remaining constant and under the same operating conditions (20% full load speed), and the filter time constant starting from the default value of 200, the waveform of the first type of acquisition position deviation is shown below. Figure 4 With a filter time constant of 200, the position deviation reached 341 pulses. After continuous trials, the position deviation was minimized when the filter time constant was 10. (See [link to relevant documentation]). Figure 5 The position deviation was 230 pulses, and the robot jitter was reduced. The filter time constant was determined and reduced to a suitable value. Increasing the speed loop bandwidth was necessary to adjust the speed loop gain; the next step was to adjust the speed loop gain. Increasing the bandwidth increases the speed loop gain, allowing the system to detect and correct speed errors more quickly, thus supporting higher gain without causing system oscillations. However, the speed loop gain should not be too high, otherwise abnormal noises and current hum may occur. Figure 6 Although the positional deviation was reduced to 47 pulses, the current waveform oscillated significantly and the robot emitted abnormal noises. The optimal gain was finally determined to be 1000. Figure 7 The positional deviation was reduced to 107 pulses, the jitter was basically eliminated, the current waveform oscillation was small, and the robot did not make any abnormal noises.

[0103] Each adjustment increment uses an automatic adjustment method, employing gradient descent. Assume the objective function is the squared integral of the error J(τ), and the gradient is... Initialize τ0 = 0.05 and learning rate α = 0.01, obtained through simulation or mathematical derivation. Iterative update When |τ k+1 -τ k |<10 -6 Stop when the time comes.

[0104] Laser tracker testing steps:

[0105] 1. Calibrate the robot's end effector position:

[0106] Install reflectors: Install spherical reflectors on the robot's end effector (such as a welding torch or fixture), ensuring their stable position and alignment with the tool center point (TCP). Record theoretical position: Record the theoretical coordinates (such as X, Y, Z coordinates) of the target point in the robot control system. These coordinates are provided by the robot programming or CAD model.

[0107] 2. Measure the actual location:

[0108] Moving the robot to the target point: The robot controller moves the end effector to the preset test point. Laser tracker measurement: The laser tracker illuminates the reflector with a laser beam and measures its three-dimensional coordinates (X, Y, Z) in space. The measurement is repeated multiple times (e.g., 3 times), and the average value is taken to reduce random errors. Data recording: The actual coordinates measured by the laser tracker are compared with the theoretical coordinates in the robot control system.

[0109] 3. Calculate the positional deviation:

[0110] Deviation formula:

[0111] The deviation value is the Euclidean distance between the actual position and the theoretical position in three-dimensional space. Split-axis analysis: Deviations along the X, Y, and Z axes can be calculated separately to analyze the robot's positioning accuracy in different directions.

[0112] 4. Multi-point testing and trajectory analysis:

[0113] Multi-point testing: Select multiple test points (e.g., 10-20) within the robot's working range, repeat the above steps, and generate a position deviation distribution map. Trajectory testing: Perform continuous motion on the robot (e.g., straight line or circular trajectory), record the deviations of key points along the path, and evaluate the dynamic positioning accuracy.

[0114] Among them, the high-precision laser tracker Leica AT960 typically has an accuracy of ±0.01mm (short distance) to ±0.05mm (long distance).

[0115] Before the improvement, the following was used: Figure 4 The overall deviation was a maximum of 341 pulses, which is quite large. Improving the filter and gain parameters, and using appropriate filter and gain parameters, yielded the following waveform: Figure 7 The maximum deviation was only 107 pulses, a reduction of 234 pulses compared to before adjustment. From Figure 4 , Figure 6 As can be seen from the waveform, the amplitude of the oscillation is large, the unstable position deviation is large, and the positional accuracy is low, which seriously affects the accuracy of the robot; from the waveform Figure 7 As you can see, the positional deviation has decreased, and the repositioning accuracy has improved.

[0116] When using this filtering and debugging method, the robot will not experience jitter during operation, whether manually or automatically, and can move smoothly. This improves the robot's positional accuracy and reduces position command deviation.

[0117] This application also provides a control device for low-speed robot movement. It should be noted that the control device for low-speed robot movement in this application can be used to execute the control method for low-speed robot movement provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0118] The following describes the control device for low-speed robot movement provided in the embodiments of this application.

[0119] Figure 8 This is a structural block diagram of a robot low-speed movement control device according to an embodiment of this application. Figure 8 As shown, the device includes an acquisition unit 10, a processing unit 20, an adjustment unit 30, and a first control unit 40. The acquisition unit acquires the original position deviation signal of the robot when performing a low-speed task and converts it into a frequency domain signal. The processing unit processes the frequency domain signal using a Chebyshev low-pass filter to obtain a filtered frequency domain signal. The adjustment unit converts the filtered frequency domain signal into a time domain signal using an inverse Fourier transform to obtain a filtered time domain signal, and adjusts the robot's velocity loop gain until a preset stopping condition is reached, obtaining an adjusted velocity loop gain. The first control unit uses the filtered time domain signal and the adjusted velocity loop gain to control the robot's movement.

[0120] In this embodiment, a Chebyshev low-pass filter is applied in the frequency domain to effectively filter out high-frequency noise in the position deviation signal when the robot is running at low speed. At the same time, the cutoff frequency and filtering time constant of the filter are optimized to improve the smoothness and stability of the signal. By adjusting the speed loop gain to the optimal state, it is ensured that stable response can be maintained while increasing bandwidth, avoiding oscillation and abnormal noise caused by over-adjustment. The robot's motion is controlled by the filtered signal and the optimized speed loop gain, which reduces jitter and improves stability, thereby solving the problem of poor stability of robot movement at low speed.

[0121] In the specific implementation process, the above processing unit includes an adjustment module, a setting module, and a comparison module. The adjustment module is used to adjust the increment of the filtering time constant using a gradient descent algorithm until a preset convergence criterion is reached to obtain the target filtering time constant; the setting module is used to set the cutoff frequency of the Chebyshev low-pass filter based on the target filtering time constant; the comparison module compares the frequency domain signal with the cutoff frequency and removes the frequency domain signal above the cutoff frequency to obtain the filtered frequency domain signal.

[0122] Furthermore, the aforementioned adjustment module includes a calculation submodule and an adjustment submodule. The calculation submodule is used to calculate the gradient of the objective function at the current filtering time constant using the aforementioned gradient descent algorithm, wherein the objective function is determined based on the integral of the squared error. The adjustment submodule is used to adjust the increment of the filtering time constant using the learning rate based on the gradient at the current filtering time constant until the aforementioned preset convergence criterion is reached, thereby obtaining the aforementioned target filtering time constant, wherein the aforementioned preset convergence criterion indicates that the increment of the filtering time constant is less than a preset convergence threshold.

[0123] Adjusting the filter time constant using the gradient descent algorithm enables automatic optimization in complex control environments, reducing the complexity and uncertainty of manual parameter tuning. Specific steps include calculating the gradient of the objective function, adjusting the time constant using the learning rate, and setting reasonable convergence criteria to ensure timely termination after finding a filter time constant that effectively improves robot motion stability and positional accuracy, avoiding over-optimization or getting trapped in local optima. This not only provides a quantitative and scientific method for parameter optimization but also adapts to changes in different working environments, enabling automatic optimization and improvement of the low-speed motion control performance of industrial robots.

[0124] Through the above embodiments, a filtered frequency domain signal was obtained, in which only low-frequency components were retained, and high-frequency jitter and noise were effectively suppressed. These filtered signals will be transformed into more stable and accurate time-domain control signals in the subsequent inverse Fourier transform process, used to guide and optimize the robot's motion control. In summary, by precisely adjusting the filtering time constant using the gradient descent algorithm to find the parameter value that best reduces robot jitter, and then setting the cutoff frequency of the Chebyshev low-pass filter according to the target filtering time constant, high-frequency components in the frequency domain signal are effectively removed. The processed filtered frequency domain signal will be used to generate the subsequent time-domain signal to provide cleaner and more stable position control commands, thereby significantly improving the motion accuracy and stability of the industrial robot under low-speed conditions.

[0125] In some embodiments of this application, the adjustment unit includes a first calculation module, a second calculation module, and an update module. The first calculation module is used to calculate an objective function for adjusting the velocity loop gain based on the filtered time-domain signal. The objective function for the velocity loop gain is set based on the position deviation and response time of the robot during movement. The second calculation module is used to calculate the gradient of the current velocity loop gain according to the objective function of the velocity loop gain. The update module is used to update the velocity loop gain according to the gradient of the current velocity loop gain and a preset learning rate until the preset stopping condition is reached to obtain the adjusted velocity loop gain. The preset stopping condition is set based on the difference between two consecutive updates of the velocity loop gain and a preset threshold.

[0126] Through the above embodiments, the objective function is calculated using the filtered time-domain signal, and then the velocity loop gain is adjusted using a gradient descent algorithm until a preset stopping condition is met. This enables automatic optimization and dynamic calibration of the robot's motion control performance. The resulting adjusted velocity loop gain not only significantly reduces positional deviations during low-speed motion but also ensures stability even at high gain, avoiding abnormal noises and current hum, thus achieving high precision and stability in motion control. In practical applications, it can adapt to different working conditions and loads, providing strong technical support for industrial robots operating in complex environments.

[0127] In some other embodiments of this application, the acquisition unit includes a control module, an acquisition module, and a third calculation module. The control module executes the control steps, specifically controlling the robot to move to a preset test point when the robot is performing the low-speed task. The preset test point includes a target position within the robot's working range. The acquisition module executes the acquisition steps, specifically acquiring the robot's actual position when the robot reaches the preset test point. The third calculation module executes the calculation steps, specifically calculating the deviation between the actual position and the target position to obtain the original position deviation signal.

[0128] The calculation of the raw position deviation signal provides an intuitive quantitative indicator of robot performance, especially positional accuracy during low-speed motion. It serves as the starting point for control strategy optimization. By analyzing and processing the raw position deviation signal, filter parameters and velocity loop gain can be adjusted to effectively reduce low-speed jitter and improve the robot's position control accuracy and stability when performing low-speed tasks. In summary, the above steps accurately capture the robot's true position deviation performance during low-speed task execution, thereby guiding system optimization. This series of operations not only provides solid data support for the subsequent application of Chebyshev low-pass filters and velocity loop gain adjustment but also enhances the intelligence and precision of the robot's low-speed motion control, effectively improving its performance in low-speed, high-precision operation scenarios and promoting overall robot performance improvement.

[0129] Furthermore, the aforementioned device also includes a repeating unit and a generating unit. The repeating unit is used to select multiple preset test points within the working range of the robot after calculating the deviation between the actual position and the target position to obtain the original position deviation signal, and repeating the control steps, the acquisition steps, and the calculation steps to obtain multiple original position deviation signals. The generating unit is used to generate a position deviation distribution map based on the multiple original position deviation signals, and analyze the positioning accuracy of the robot within the working range based on the position deviation distribution map.

[0130] By precisely controlling the robot to move to a preset test point and recording the deviation between its actual and target positions, a raw position deviation signal is generated. This process is repeated to collect multiple raw position deviation signals covering key locations within the working range, and a position deviation distribution map is constructed based on these signals. The position deviation distribution map visually demonstrates the robot's positioning accuracy at different positions, providing precise guidance for performance optimization. By analyzing the position deviation distribution map, areas with large positioning deviations can be identified, allowing for corresponding measures, such as optimizing filtering parameters or adjusting the velocity loop gain, to reduce position deviations and improve overall positioning accuracy and stability. This process ensures that the robot achieves higher control precision even at low speeds.

[0131] In some embodiments of this application, the above-mentioned device further includes a second control unit, a setting unit, a calculation unit, and an evaluation unit. The second control unit is used to control the robot to perform a preset movement after calculating the deviation between the actual position and the target position to obtain the original position deviation signal, thereby acquiring the actual position of the robot under the preset movement, wherein the preset movement includes linear movement and circular trajectory movement; the setting unit is used to set the target position of the robot under the preset movement; the calculation unit is used to calculate the deviation between the actual position and the target position of the robot at a target point in the preset movement to obtain the target deviation, wherein the target point includes a start point, an end point, and an inflection point; the evaluation unit is used to evaluate the positioning accuracy of the robot within the working range based on the target deviation.

[0132] By controlling the robot to execute preset straight and circular trajectories, position deviation signals during dynamic processes can be accurately acquired, allowing for a comprehensive evaluation of its positioning accuracy within its working range. This approach not only covers deviation measurements at static points but also incorporates analysis of dynamic path motion deviations, resulting in a more holistic assessment. By setting target positions at key points along the motion path and calculating the deviation between the robot's actual position and the target position, a series of target deviation data are obtained. These data reflect the robot's positioning accuracy when executing complex motion commands. The evaluation based on target deviations reveals performance bottlenecks in the robot control system under dynamic conditions, providing specific guidance for subsequent parameter optimization (such as filter adjustment and velocity loop gain settings), thereby significantly improving the robot's positioning capabilities in low-speed, high-precision task execution.

[0133] Furthermore, the above-mentioned device also includes a preprocessing unit, which performs preprocessing operations on the original position deviation signal after calculating the deviation between the actual position and the target position to obtain the original position deviation signal. The preprocessing operations include noise reduction, data smoothing, and format conversion.

[0134] Preprocessing significantly reduces the impact of noise, smooths the signal curve, and makes the signal more clearly reflect the robot's true position deviation. This high-quality signal data provides a solid foundation for subsequent position deviation analysis, filter parameter optimization, and velocity loop gain settings, ensuring the accuracy and effectiveness of robot control system adjustments and ultimately improving the robot's positioning accuracy and operational stability. Preprocessing not only enhances signal usability but also simplifies the complexity of subsequent processing.

[0135] The aforementioned control device for low-speed robot movement includes a processor and a memory. The acquisition unit, processing unit, adjustment unit, and first control unit are all stored as program units in the memory. The processor executes these program units stored in the memory to achieve the corresponding functions. All of the above modules reside in the same processor; alternatively, the modules may be located in different processors in any combination.

[0136] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0137] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the control method for low-speed robot movement.

[0138] This invention provides a processor for running a program, wherein the program executes the control method for low-speed robot movement.

[0139] This invention provides an electronic device, including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of the aforementioned control method for low-speed robot movement. The device described herein can be a server, PC, PAD, mobile phone, etc.

[0140] This application also provides a computer program product that, when executed on a data processing device, is adapted to perform the steps of initializing the control method for the low-speed movement of the robot described above.

[0141] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0142] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0143] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0144] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0145] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0146] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0147] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0148] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0149] 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.

[0150] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0151] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for controlling the low-speed motion of a robot, characterized in that, include: Acquire the original position deviation signal of the robot when performing a low-speed task, and convert the original position deviation signal into a frequency domain signal; The frequency domain signal is processed using a Chebyshev low-pass filter to obtain a filtered frequency domain signal. The filtered frequency domain signal is converted into a time domain signal through inverse Fourier transform to obtain the filtered time domain signal. The speed loop gain of the robot is then adjusted until a preset stopping condition is reached to obtain the adjusted speed loop gain. The robot motion is controlled using the filtered time-domain signal and the adjusted velocity loop gain.

2. The method according to claim 1, characterized in that, The frequency domain signal is processed using a Chebyshev low-pass filter to obtain a filtered frequency domain signal, including: The gradient descent algorithm is used to adjust the increment of the filtering time constant until the preset convergence criterion is reached, thus obtaining the target filtering time constant; The cutoff frequency of the Chebyshev low-pass filter is set based on the target filtering time constant; The frequency domain signal is compared with the cutoff frequency, and the frequency domain signal above the cutoff frequency is removed to obtain the filtered frequency domain signal.

3. The method according to claim 2, characterized in that, The gradient descent algorithm is used to adjust the increment of the filter time constant until a preset convergence criterion is reached, resulting in the target filter time constant, including: The gradient of the objective function under the current filtering time constant is calculated using the gradient descent algorithm, wherein the objective function is determined based on the integral of the squared error. Based on the gradient under the current filtering time constant, the increment of the filtering time constant is adjusted using the learning rate until the preset convergence criterion is reached to obtain the target filtering time constant, wherein the preset convergence criterion indicates that the increment of the filtering time constant is less than a preset convergence threshold.

4. The method according to claim 1, characterized in that, Adjusting the robot's velocity loop gain until a preset stopping condition is reached, to obtain the adjusted velocity loop gain, includes: The objective function for adjusting the velocity loop gain is calculated based on the filtered time-domain signal. The objective function for the velocity loop gain is set based on the position deviation and response time of the robot during its movement. Calculate the gradient of the current velocity loop gain based on the objective function of the velocity loop gain; The velocity loop gain is updated based on the gradient of the current velocity loop gain and the preset learning rate until the preset stopping condition is reached, thereby obtaining the adjusted velocity loop gain. The preset stopping condition is set based on the difference between two consecutive updates of the velocity loop gain and a preset threshold.

5. The method according to claim 1, characterized in that, Acquire the robot's raw position deviation signal when performing low-speed tasks, including: The control step involves controlling the robot to move to a preset test point while the robot is performing the low-speed task. The preset test point includes a target location within the robot's working range. The acquisition step involves obtaining the actual position of the robot when it reaches the preset test point. The calculation step involves calculating the deviation between the actual position and the target position to obtain the original position deviation signal.

6. The method according to claim 5, characterized in that, After calculating the deviation between the actual position and the target position to obtain the original position deviation signal, the method further includes: Within the robot's working range, select multiple preset test points, repeat the control steps, the acquisition steps, and the calculation steps to obtain multiple original position deviation signals; A position deviation distribution map is generated based on multiple original position deviation signals, and the positioning accuracy of the robot within the working range is analyzed based on the position deviation distribution map.

7. The method according to claim 5, characterized in that, After calculating the deviation between the actual position and the target position to obtain the original position deviation signal, the method further includes: The robot is controlled to perform a preset motion, and the actual position of the robot under the preset motion is obtained, wherein the preset motion includes linear motion and circular trajectory motion; Set the target position of the robot under the preset motion; The deviation between the robot's actual position and the target position at the target point in the preset motion is calculated to obtain the target deviation, wherein the target point includes a start point, an end point, and an inflection point; The positioning accuracy of the robot within the working range is evaluated based on the target deviation.

8. The method according to claim 5, characterized in that, After calculating the deviation between the actual position and the target position to obtain the original position deviation signal, the method further includes: The original position deviation signal is preprocessed, including noise reduction, data smoothing, and format conversion.

9. A control device for low-speed robot movement, characterized in that, include: The acquisition unit is used to acquire the original position deviation signal of the robot when performing a low-speed task, and convert the original position deviation signal into a frequency domain signal. The processing unit is used to process the frequency domain signal using a Chebyshev low-pass filter to obtain a filtered frequency domain signal. The adjustment unit is used to convert the filtered frequency domain signal into a time domain signal through inverse Fourier transform to obtain the filtered time domain signal, and adjust the speed loop gain of the robot until a preset stopping condition is reached to obtain the adjusted speed loop gain. A first control unit is used to control the robot's motion using the filtered time-domain signal and the adjusted velocity loop gain.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the control method for low-speed robot movement as described in any one of claims 1 to 8.

Images