Robotic teleoperation method based on flexible manufacturing direct force feedback

Abstract

The present application relates to the technical field of robot control, in particular to a robot remote control method based on flexible manufacturing direct force feedback, comprising: collecting force feedback signals of a robot end and network transmission parameters; determining an initial compression ratio according to the network transmission parameters, and combining time domain variation characteristics of the force feedback signals, compressing the force feedback signals based on the initial compression ratio to obtain compressed force signals; updating the initial compression ratio to obtain a target compression ratio based on signal errors and correlations of restored force signals obtained by restoring the compressed force signals relative to the force feedback signals, and compressing the force feedback signals based on the target compression ratio to obtain target compressed force signals; transmitting the target compressed force signals to a remote operation end, and sending control instructions generated by the remote operation end according to the target compressed force signals to a robot controller to drive an actuator of the robot end to perform work. The present application realizes improved transmission efficiency while ensuring remote control accuracy.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, and specifically to a robot remote control method based on direct force feedback from flexible manufacturing. Background Technology

[0002] In flexible manufacturing scenarios, robot telemanipulation technology enables operators to remotely control robots to perform contact-based tasks such as precision assembly and smoothing. Its core lies in establishing a two-way information loop between the operator's end effector and the robot's end effector through force feedback signals. This allows the operator to obtain force perception of the remote working environment and adjust their actions accordingly. Force feedback signals are typically collected by multi-dimensional force sensors installed at the robot's end effector, containing force and torque components across multiple axes. The signal data volume is large, and high real-time transmission requirements are necessary.

[0003] In existing robot telecontrol force feedback transmission schemes, the raw force feedback signals collected by sensors are typically sent directly to the remote control terminal via the network, or compressed at a fixed compression ratio before transmission. Direct transmission of high-dimensional, high-frequency force feedback signals consumes a large amount of network bandwidth and generates high transmission latency when network conditions are limited, resulting in slow response of the force feedback loop and affecting the real-time performance and stability of telecontrol. While using a fixed compression ratio can reduce the amount of data, it cannot adapt to changes in network conditions and dynamic changes in force signal characteristics during operation. When the compression ratio is set too high, key force feedback details are easily lost; when the compression ratio is set too low, it cannot effectively alleviate bandwidth pressure. It is difficult to balance overall transmission efficiency and control accuracy, resulting in poor overall telecontrol performance. Summary of the Invention

[0004] To address the technical challenge of balancing transmission efficiency and control precision in force feedback telecontrol in dynamic network environments, this invention aims to provide a robot telecontrol method based on direct force feedback from flexible manufacturing. The specific technical solution adopted is as follows: Firstly, a robot teleoperation method based on direct force feedback in flexible manufacturing is provided. This method includes: acquiring force feedback signals and network transmission parameters from the robot's end effector; the force feedback signals include multiple axial force components and multiple circumaxial torque components; the network transmission parameters include real-time available bandwidth and transmission delay; determining an initial compression ratio based on the network transmission parameters, and compressing the force feedback signals based on the time-domain variation characteristics of the force feedback signals to obtain a compression force signal; updating the initial compression ratio to obtain a target compression ratio based on the signal error and correlation between the restored force signal obtained after restoring the compression force signal and the force feedback signal, and compressing the force feedback signals based on the target compression ratio to obtain a target compression force signal; transmitting the target compression force signal to a remote operating terminal, and sending control commands generated by the remote operating terminal based on the target compression force signal to the robot controller to drive the actuator at the robot's end effector to perform the operation.

[0005] In one possible design, the initial compression ratio is determined based on network transmission parameters, including: determining a bandwidth index based on real-time available bandwidth and a preset baseline bandwidth; determining a transmission delay index based on transmission delay and a preset baseline transmission delay; determining a compression ratio adjustment factor based on the bandwidth index and the transmission delay index; and determining the initial compression ratio based on the compression ratio adjustment factor and a preset compression ratio.

[0006] In one possible design, the force feedback signal is compressed based on an initial compression ratio, taking into account its time-domain variation characteristics, to obtain a compressive force signal. This includes: segmenting the acquired force feedback signal into data segments with a preset duration; dividing the data segments into stationary or non-stationary phases based on their time-domain variation characteristics; filtering the force feedback signal within the data segment for the stationary phase and compressing the filtered force feedback signal based on the initial compression ratio to obtain the compressive force signal; and compressing the force feedback signal within the data segment for the non-stationary phase based on the initial compression ratio to obtain the compressive force signal.

[0007] In one possible design, data segments are divided into stationary or non-stationary phases based on their time-domain variation characteristics. This includes: for the first data segment, classifying it as a non-stationary phase; for subsequent data segments, determining a first non-stationarity index based on the force feedback signal within the data segment and the force feedback signal in the adjacent preceding data segment, the first non-stationarity index characterizing the signal fluctuation between adjacent data segments; determining a second non-stationarity index based on the number of amplitude abrupt change sampling points of the force feedback signal within the data segment, the second non-stationarity index characterizing the degree of local signal abrupt change within the data segment, the first and second non-stationarity indices constituting the time-domain variation characteristics; classifying the data segment as a stationary phase when the first non-stationarity index is less than a first non-stationarity threshold and the second non-stationarity index is less than a second non-stationarity threshold; otherwise, classifying the data segment as a non-stationary phase.

[0008] In one possible design, a first non-stationarity index is determined based on the force feedback signal within a data segment and the force feedback signal in the adjacent preceding data segment. This includes: for each component of the force feedback signal, determining the signal variance deviation rate of the component based on the variance of the component in the data segment and the variance of the component in the adjacent preceding data segment; and determining the maximum value among the signal variance deviation rates of all components as the first non-stationarity index. A second non-stationarity index is determined based on the number of amplitude abrupt change sampling points of the force feedback signal within the data segment. This includes: for each component of the force feedback signal, determining the amplitude change of the component at each sampling point within the data segment excluding the first sampling point, and determining the number of sampling points where the amplitude change exceeds a preset amplitude change threshold as the number of amplitude abrupt change sampling points of the component; and determining the maximum value among the number of amplitude abrupt change sampling points of all components as the second non-stationarity index.

[0009] In one possible design, the target compression ratio is obtained by updating the initial compression ratio based on the signal error and correlation between the restored force signal (obtained after restoring the compression force signal) and the force feedback signal. This includes: decompressing the compression force signal to obtain the restored force signal, and determining the signal error based on the restored force signal and the force feedback signal, whereby the signal error characterizes the degree of information loss in the force feedback signal during compression; using a preset correlation analysis algorithm to determine the correlation coefficient between the force feedback signal and the restored force signal; determining the compression effect evaluation value based on the correlation coefficient and the signal error; updating the initial compression ratio based on the compression effect evaluation value if it is lower than a preset effect threshold, and redetermining the compression effect evaluation value based on the updated initial compression ratio until the compression effect evaluation value reaches or exceeds the preset effect threshold; and determining the initial compression ratio where the compression effect evaluation value reaches or exceeds the preset effect threshold as the target compression ratio.

[0010] In one possible design, the force feedback signal is compressed based on the target compression ratio to obtain the target compression force signal, including: for data segments divided into stationary phases, the force feedback signal within the data segment is filtered, and the filtered force feedback signal is compressed based on the target compression ratio to obtain the target compression force signal; for data segments divided into non-stationary phases, the force feedback signal within the data segment is compressed based on the target compression ratio to obtain the target compression force signal.

[0011] In one possible design, after obtaining the data segment, the above method further includes: acquiring the maximum force amplitude of each axial force component and the maximum torque amplitude of each axial torque component within the current data segment; determining a first safety proximity based on the maximum force amplitude and a preset force safety threshold; determining a second safety proximity based on the maximum torque amplitude and the preset torque safety threshold; and performing lossless compression processing on the force feedback signal within the current data segment if the first safety proximity is greater than a preset safety ratio or the second safety proximity is greater than a preset safety ratio, and using the lossless compressed data as a compression force signal.

[0012] In one possible design, the target compression force signal is transmitted to a remote control terminal, and the control command generated by the remote control terminal based on the target compression force signal is sent to the robot controller to drive the actuator at the robot end to perform the operation. This includes: decompressing and restoring the target compression force signal at the remote control terminal, and generating a feedback force from the decompressed and restored force signal through a force feedback device; collecting motion data generated in response to the operation action through an input device, and generating control commands based on the motion data; and transmitting the control commands to the robot controller to drive the actuator at the robot end to perform the operation action corresponding to the control command.

[0013] In one possible design, force feedback signals and network transmission parameters of the robot end effector are collected, including: acquiring multiple axial force components and multiple axial torque components of the robot end effector at a preset sampling frequency using a six-dimensional force sensor installed at the robot end effector; and acquiring the real-time available bandwidth and transmission delay between the robot and the remote control terminal through a network monitoring tool at a preset update frequency.

[0014] The present invention has the following beneficial effects: In the robot telecontrol method based on direct force feedback in flexible manufacturing provided by this invention, force feedback signals from the robot end effector and network transmission parameters are collected synchronously. The force feedback signals are then compressed based on the network transmission parameters and the temporal variation characteristics of the force feedback signals. This allows the compression level to be dynamically adjusted according to changes in network conditions and operating states. When real-time available bandwidth is sufficient and transmission latency is good, a baseline compression level is used to balance data integrity and transmission efficiency. When network conditions deteriorate, the compression level is adaptively adjusted to ensure real-time transmission. When the operation is in a stable phase, high-frequency transient components are filtered out to further reduce the data volume. Simultaneously, based on the restored force signal phase... The initial compression ratio is updated iteratively in a closed loop based on the signal error and correlation of the original force feedback signal. This allows the compression ratio to automatically reduce the compression level to preserve force information when the compression is too high and force information is lost. When severe distortions such as signal direction reversal occur, the evaluation value naturally decreases and triggers a corresponding adjustment of the compression ratio. Ultimately, a target compression force signal that balances force fidelity and real-time transmission is obtained. This enables the processing of the force feedback signal to continuously adapt to changes in network status and signal characteristics throughout the teleoperation process. While ensuring the operator's sense of presence and operational accuracy, it effectively reduces data transmission volume, alleviates network burden, and improves the real-time performance and stability of teleoperation. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating a robot teleoperation method based on direct force feedback in flexible manufacturing, as provided in one embodiment of the present invention. Detailed Implementation

[0016] The following description, in conjunction with the accompanying drawings, details a specific scheme for a robot remote control method based on direct force feedback in flexible manufacturing provided by the present invention.

[0017] Please see Figure 1 The diagram illustrates a flowchart of a robot teleoperation method based on direct force feedback in flexible manufacturing, provided by an embodiment of the present invention, including the following steps S101-S105.

[0018] S101. Collect the force feedback signal and network transmission parameters of the robot's end effector.

[0019] The force feedback signal includes multiple axial force components and multiple axial torque components, and the network transmission parameters include real-time available bandwidth and transmission delay.

[0020] As one possible implementation, in terms of force feedback signal acquisition, a six-dimensional force sensor installed at the end of the robot is used to acquire multiple axial force components and multiple axial torque components at a preset sampling frequency (e.g., 1000Hz).

[0021] Optionally, the six-dimensional force sensor can be installed at the end flange of the robot and collect axial force components along the three orthogonal directions of the robot end coordinate system (X-axis, Y-axis, Z-axis) to obtain three axial force components along the X-axis, Y-axis, and Z-axis respectively; and collect axial torque components around the three orthogonal axes (X-axis, Y-axis, Z-axis) to obtain three axial torque components around the X-axis, Y-axis, and Z-axis respectively. Each sampling point collected yields a force feedback signal containing six dimensions, which completely records the force interaction information between the robot end and the workpiece.

[0022] Regarding the acquisition of network transmission parameters, the real-time available bandwidth and transmission delay between the robot and the remote control terminal are collected through network monitoring tools at a preset update frequency (e.g., 20Hz). The real-time available bandwidth is used to characterize the data transmission capability of the current communication link; the transmission delay is used to characterize the time taken for the force feedback signal to be transmitted between the robot and the remote control terminal.

[0023] It should be noted that the preset sampling frequency and the preset update frequency are set independently. Because force feedback signal sampling requires capturing rapid changes in contact force, the preset sampling frequency is set relatively high; while network transmission parameters change relatively slowly, so the preset update frequency is set relatively low. By independently setting these two frequencies, the accuracy requirements for force feedback signal acquisition can be met while avoiding unnecessary and frequent queries to the network status, thus saving computational resources.

[0024] S102. Determine the initial compression ratio based on the network transmission parameters.

[0025] During robot teleoperation, force feedback signals need to be frequently transmitted between the robot and the remote operator. Network transmission conditions directly determine the efficiency and reliability of signal transmission. Assuming a relatively constant force feedback signal data generation rate, when the available real-time bandwidth is low, the network data throughput is weak, and the amount of data that can be transmitted per unit time is limited. If transmission is still done with a low compression level, it will lead to data queue backlog, exacerbating transmission delays and severely affecting the real-time performance and stability of the feedback loop. In this case, a greater compression level is needed to reduce data transmission volume and alleviate network burden. Conversely, when the available real-time bandwidth is ample, the network throughput is strong, and data can be sent promptly. In this case, the compression level can be reduced to retain richer force sensory details and improve the sense of force presence. When the transmission delay is high, network transmission is slow, and the real-time performance of the force feedback signal is difficult to guarantee. A greater compression level is needed to reduce data transmission volume and alleviate network burden, thereby maintaining the real-time performance of the force feedback loop. Therefore, it is necessary to adaptively determine a reasonable initial compression ratio based on the real-time available bandwidth and transmission delay in the network transmission parameters. When the network conditions are poor, the compression level should be actively increased to ensure real-time performance, and when the network conditions are good, the compression level should be appropriately reduced to ensure sensory fidelity.

[0026] As one possible implementation, the bandwidth index is first determined based on the real-time available bandwidth and a preset baseline bandwidth. The preset baseline bandwidth is a pre-selected reference bandwidth value, representing the typical available bandwidth under good network conditions; for example, an empirical value of 10 Mbps can be used. The bandwidth index measures the degree of constraint of the currently available real-time bandwidth. The lower the bandwidth, the weaker the data throughput, the higher the required compression level, and the larger the bandwidth index. To avoid an excessively high bandwidth index when bandwidth is extremely insufficient, an upper limit is set. For example, the formula for calculating the bandwidth index is as follows: In the formula, For bandwidth index, This refers to the real-time available bandwidth in the network transmission parameters. For the preset baseline bandwidth, This sets the upper limit for the preset bandwidth index, for example, a value of 2. This indicates that the smaller of the two values ​​is taken. When the real-time available bandwidth is greater than or equal to the preset baseline bandwidth, the ratio does not exceed 1, and the bandwidth index is taken as this ratio or less than 1, indicating that the bandwidth is sufficient and the demand for compression is low. When the real-time available bandwidth is less than the preset baseline bandwidth, the ratio increases, the bandwidth index increases, and a higher compression ratio is required, but it does not exceed the upper limit of the preset bandwidth index.

[0027] Secondly, the transmission delay index is determined based on the transmission delay and a preset baseline transmission delay. The preset baseline transmission delay is a pre-selected reference delay value, representing the typical transmission delay under good network conditions; for example, an empirical value of 10ms can be used. The transmission delay index measures the degree to which the current transmission delay, relative to the preset baseline transmission delay, drives compression demand. When the transmission delay is at an excellent level, the transmission delay index is 1, indicating that no additional compression demand is generated due to the delay. When the transmission delay exceeds the preset baseline transmission delay, the transmission delay index increases with the increase in delay to reflect the gradually increasing real-time compression demand. To avoid an excessively large transmission delay index leading to an excessive increase in the compression ratio, an upper limit is set on it. For example, the formula for calculating the transmission delay index is as follows: In the formula, The transmission delay index, For transmission delay in network transmission parameters, The preset baseline transmission delay, To preset the upper limit of the latency index, an empirical value of 2 can be used. This means taking the larger of the two values. This indicates that the smaller of the two values ​​is taken. When the transmission delay is less than or equal to the preset baseline transmission delay... The function sets the transmission delay exponent to 1, indicating excellent current latency and no additional compression requirements. When the transmission delay exceeds a preset baseline transmission delay, the transmission delay exponent is the ratio of the two, and does not exceed the preset upper limit of the delay exponent. The greater the latency, the greater the transmission latency exponent, and the higher the compression requirement.

[0028] Then, the compression ratio adjustment factor is determined based on the bandwidth index and the transmission delay index. The bandwidth index reflects the compression requirements caused by insufficient transmission capacity from the perspective of limited data throughput; a larger value indicates more scarce bandwidth and a higher degree of compression required. The transmission delay index reflects the compression required for real-time performance from the perspective of response speed; both are positively correlated with compression requirements. The compression ratio adjustment factor is determined by averaging the two factors to synthesize the compression requirements from both aspects, and a preset adjustment upper limit is set. The preset adjustment upper limit is greater than 1, so that the initial compression ratio can be amplified based on the preset compression ratio when network conditions deteriorate. For example, the formula for calculating the compression ratio adjustment factor is as follows: In the formula, This is the compression ratio adjustment factor, with a value ranging from 0 to 1; For bandwidth index, The transmission delay index, To preset the upper limit, for example, an empirical value of 2 can be taken. When network conditions are good, bandwidth is sufficient, and latency is excellent, and Both are close to or equal to 1, and their average is close to 1. A value close to 1 indicates a small adjustment range to the preset compression ratio; however, when network conditions deteriorate, bandwidth is insufficient, or latency is high, and As the number of elements increases, the average value of both elements increases. Increase accordingly, but not exceeding the preset adjustment limit. .

[0029] Finally, the initial compression ratio is determined based on the compression ratio adjustment factor and the preset compression ratio. The preset compression ratio is a baseline compression ratio that balances data integrity and transmission efficiency under good network conditions. It represents the system's default compression level under ideal network conditions with ample bandwidth and excellent transmission latency. For example, an empirical value of 7 can be used, indicating that the amount of data before compression is 7 times the amount of data after compression. At this compression level, the force feedback signal transmission can achieve a balance between force fidelity and real-time transmission. When network conditions deteriorate, the initial compression ratio is amplified based on the preset compression ratio using the compression ratio adjustment factor to reflect the additional compression requirements caused by the worsening network conditions. For example, the formula for determining the initial compression ratio is as follows: In the formula, This is the initial compression ratio. This is the compression ratio adjustment factor. This is the preset compression ratio, also known as the baseline compression ratio. The compression ratio adjustment factor is used when network conditions are good. When the initial compression ratio is close to or equal to 1, it is close to or equal to the preset compression ratio. At this point, a baseline compression level is used to achieve a balance between data integrity and transmission efficiency. When network conditions deteriorate, the compression ratio adjustment factor is adjusted. The initial compression ratio is increased to a higher level than the preset compression ratio, thus improving the compression level. This sacrifices some data integrity to ensure real-time transmission. Under extreme network conditions, the compression ratio adjustment factor... Reaching the preset adjustment limit The initial compression ratio reaches the preset compression ratio. This achieves the highest level of compression.

[0030] Through the above steps, the initial compression ratio is dynamically adjusted according to changes in real-time available bandwidth and transmission delay: when real-time available bandwidth is sufficient, the bandwidth exponent is small, and the compression requirement is low; when the transmission delay is excellent, the transmission delay exponent is 1, and the compression requirement does not increase additionally. At this time, the compression ratio adjustment coefficient is small, the initial compression ratio is small, and the compression degree is low, in order to retain richer sensory details. When real-time available bandwidth is insufficient, the bandwidth exponent increases; when the transmission delay exceeds the baseline, the transmission delay exponent increases, both of which drive the compression ratio adjustment coefficient to increase, the initial compression ratio to increase, and the compression degree to increase, in order to ensure the real-time performance of data transmission.

[0031] S103. Combining the time-domain variation characteristics of the force feedback signal, the force feedback signal is compressed based on the initial compression ratio to obtain the compression force signal.

[0032] In robot telecontrol, the stable state of the force feedback signal directly determines the compression strategy. During the stable phase, signal changes are slow, and the operator primarily perceives the low-frequency trend; high-frequency components are mostly noise or non-critical transient signals. After filtering out high-frequency noise, compression can be performed at the initial compression ratio to significantly reduce data volume. In the non-stationary phase (including abrupt operation, fine operation, and transition phases), signal changes are drastic. The operator needs to perceive these rapid changes to make real-time control decisions; therefore, the complete details of the signal must be preserved, and compression should be performed directly at the initial compression ratio. Furthermore, all compression operations are predicated on safe operation. For data segments in the force feedback signal where the force or torque amplitude is close to the safety threshold, which may be dangerous signals caused by collisions, overloads, etc., lossless compression must be used to preserve and transmit them completely. Therefore, it is necessary to classify data segments into stable or non-stationary phases based on their time-domain variation characteristics, and simultaneously, in conjunction with a safety check mechanism, perform differentiated compression on data segments in different states.

[0033] As one possible approach, the collected force feedback signal is first segmented into data segments with a preset duration.

[0034] Optionally, the preset duration can be set according to the sampling frequency of the force feedback signal and the timeliness requirements of force perception. For example, an empirical value of 10ms can be taken. Every 10ms of force feedback signal collected constitutes a data segment. Each data segment contains multiple sampling points continuously collected within the preset duration. Each sampling point contains three axial force components and three about-axis torque components, completely recording the force interaction information between the robot end effector and the workpiece within this time period.

[0035] Secondly, for each data segment, a safety check needs to be performed before performing time-domain feature analysis and classification to determine whether the force feedback signal of the current data segment contains dangerous signals that are close to the safety threshold.

[0036] Optionally, the maximum force amplitude of each axial force component and the maximum torque amplitude of each about-axis torque component within the current data segment are obtained. A first safety proximity is determined based on the maximum force amplitude and a preset force safety threshold, and a second safety proximity is determined based on the maximum torque amplitude and the preset torque safety threshold. The first safety proximity characterizes the degree to which the force signal in the current data segment approaches the safety limit, and the second safety proximity characterizes the degree to which the torque signal in the current data segment approaches the safety limit. For example, the calculation formulas for determining the first and second safety proximity are as follows: In the formula, For the first level of safe proximity, This represents the maximum force amplitude among the three axial force components at each sampling point within the current data segment. A preset force safety threshold can be set, for example, an empirical value of 50N can be taken; For the second level of security proximity, This represents the maximum torque amplitude among the three about-axis torque components at each sampling point within the current data segment. A preset torque safety threshold can be set, for example, an empirical value of 10 Nm can be taken. or The larger the value, the closer the corresponding signal in the current data segment is to the safety limit.

[0037] If either the first security proximity or the second security proximity is greater than a preset security ratio, the force feedback signal within the current data segment is compressed using a lossless compression algorithm to ensure completely distortion-free transmission. The losslessly compressed data is then used as the compression force signal, and subsequent time-domain feature analysis and classification steps are skipped. For example, the preset security ratio can be 90% (i.e., 0.9). If neither the first nor the second security proximity exceeds the preset security ratio, the data segment does not involve security risks, and subsequent time-domain feature analysis continues.

[0038] Then, for the data segments that pass the security check, it is necessary to determine the stage to which the data segment belongs based on the time-domain variation characteristics of the force feedback signal. The time-domain variation characteristics include two dimensions: a first non-stationarity index and a second non-stationarity index. The first non-stationarity index characterizes the degree of signal fluctuation between adjacent data segments, reflecting the signal's trend over a continuous time period. It can be obtained by comparing the variance differences of each force component and torque component in the current data segment and the adjacent previous data segment. The greater the difference, the more obvious the change in the overall signal state, and the higher the degree of non-stationarity. The second non-stationarity index characterizes the degree of local signal abrupt changes within a data segment, reflecting the drastic changes in the signal within a single data segment. It can be obtained by statistically analyzing the frequency of large value jumps between adjacent sampling points of each component within the current data segment. The higher the frequency of abrupt changes, the closer the force interaction is to a non-stationary state such as fine manipulation or collision.

[0039] For the first data segment, since there is no adjacent preceding data segment, the first non-stationarity index cannot be calculated. Therefore, this data segment is directly divided into a non-stationary stage to preserve the complete force perception information of the initial operation stage and ensure safety.

[0040] For data segments that are not the first, the first and second non-stationary indices are determined using the following method.

[0041] The first nonstationarity index is determined based on the signal variance deviation rate. For each axial force component and each about-axis moment component included in the force feedback signal, the variance of that component in the current data segment and the variance of that component in the adjacent previous data segment are calculated. Then, the ratio of the absolute value of the difference between the two to the variance of the adjacent previous data segment is calculated to obtain the signal variance deviation rate of that component. The larger the variance deviation rate, the more significant the overall fluctuation of that component changes between two adjacent time periods. Simultaneously, to avoid the denominator being zero, a very small positive number is introduced into the denominator. The maximum value among the signal variance deviation rates of all components is then taken as the first nonstationarity index. For example, the formula for calculating the signal variance deviation rate is as follows: In the formula, For the first Signal variance deviation rate of each component Use component indices to iterate through the three axial force components and the three about-axis moment components; For the first Each component in the current data segment within variance; For the first Each component is in the adjacent preceding data segment within variance; This is a preset, extremely small positive number used to prevent the denominator from being zero; for example, it can be 0.001.

[0042] Furthermore, after determining the signal variance deviation rate of each component, the maximum value among all the signal variance deviation rates is determined as the first non-stationarity index. The larger the value of the first non-stationarity index, the more severe the signal fluctuation between adjacent data segments.

[0043] The second non-stationarity index is determined based on the number of amplitude abrupt change sampling points between adjacent sampling points within a data segment. For each component of the force feedback signal, starting from the second sampling point within the data segment, the amplitude change of force or torque between each sampling point and the previous sampling point is calculated sequentially. If the amplitude change exceeds a preset amplitude change threshold, the sampling point is marked as an amplitude abrupt change sampling point, and the number of amplitude abrupt change sampling points for that component throughout the entire data segment is counted. The preset amplitude change threshold is set according to the accuracy of the force feedback sensor and the operational requirements; for example, it can be set to 0.5N or an equivalent torque value to filter out small amplitude fluctuations. The maximum value among the number of amplitude abrupt change sampling points for all components is taken as the second non-stationarity index. The larger the value of the second non-stationarity index, the more local large amplitude jumps exist within the data segment, and the more the operation phase tends towards a fine contact or abrupt change state.

[0044] After obtaining the first and second non-stationarity indices, they are compared with preset first and second non-stationarity thresholds, respectively. When both the first and second non-stationarity indices are less than the first non-stationarity threshold and the second non-stationarity indices are less than the second non-stationarity threshold, it indicates that the force feedback signal within the current data segment is generally stable and without significant local abrupt changes, and the data segment is classified as a stationary phase; otherwise, the data segment is classified as a non-stationary phase. The first and second non-stationarity thresholds can be preset according to the sensitivity requirements of the actual remote control task and the noise characteristics of the force sensor. For example, the first non-stationarity threshold can be set to 5% (i.e., 0.05), and the second non-stationarity threshold can be set to 2 to distinguish between stable contact and fine operation states.

[0045] After the phase division is completed, different compression processing methods are applied to the force feedback signals within the data segments based on the division results.

[0046] For data segments classified as stable phases, the force feedback signal within these segments is filtered to remove high-frequency transient components and retain low-frequency steady-state force components. This filtering can be achieved using a low-pass filter with a preset cutoff frequency. The preset cutoff frequency can be set according to the characteristics of human perception of force signals, for example, 20Hz, retaining only mid-to-low frequency force information meaningful to the operator's perception of environmental stiffness and contact state. The filtered force feedback signal is then compressed based on the initial compression ratio to obtain the compression force signal.

[0047] For data segments divided into non-stationary phases, no filtering is performed. Instead, the original force feedback signal within the data segment is directly compressed based on the initial compression ratio to obtain a compression force signal, so as to fully preserve the force details of abrupt changes or fine operation phases.

[0048] It should be noted that, through the above adaptive compression mechanism, the data segment in the stable phase is compressed after filtering out high-frequency noise, which can significantly reduce the amount of data while retaining the core force information needed by the operator to perceive the environment; the data segment in the non-stationary phase is compressed while retaining the complete signal details, which can ensure that the operator can perceive rapid changes in signals and make real-time control decisions; the safety critical data segment is processed using a lossless compression method, which can ensure that safety-critical signals such as collisions and overloads are transmitted to the remote operation terminal completely and without distortion.

[0049] S104. Based on the signal error and correlation between the restored force signal obtained after restoring the compression force signal and the force feedback signal, update the initial compression ratio to obtain the target compression ratio, and perform compression processing on the force feedback signal based on the target compression ratio to obtain the target compression force signal.

[0050] In actual robot telemanipulation, the compressed force signal is decompressed and restored to obtain the restored force signal, which is the force information actually perceived by the operator. If the waveform direction consistency of the force feedback signal is disrupted or the numerical accuracy is insufficient during compression, the operator will not be able to accurately perceive the force interaction state of the robot's end effector, thus affecting the operational accuracy. Therefore, it is necessary to perform closed-loop verification and iterative optimization of the initial compression ratio based on the waveform direction consistency and numerical error between the restored force signal and the original force feedback signal until the compression effect meets the requirements. The optimized compression ratio is then determined as the final target compression ratio, and the force feedback signal is recompressed based on the target compression ratio.

[0051] As one possible approach, the compression force signal is first decompressed to obtain the restored force signal.

[0052] The compression force signal is the force feedback signal after compression processing, with a reduced data volume compared to the original force feedback signal. Decompression and restoration is the reverse process of compression processing, restoring the compression force signal to a force signal with the same sampling rate and data format as the original force feedback signal, so that it can be compared with the original force feedback signal in the same dimension. The specific method of decompression and restoration corresponds to the method of compression processing. For example, if the compression processing used a downsampling method, the sampling points are restored by interpolation during decompression and restoration; if the compression processing used a quantization level adjustment method, inverse quantization is performed during decompression and restoration.

[0053] It should be noted that when the compression force signal is obtained through lossless compression processing—that is, when the data segment undergoes lossless compression processing due to a safety proximity trigger—the decompressed and restored force signal is completely consistent with the original force feedback signal. The signal error of each component is zero, the correlation coefficient is 1, and the compression effect evaluation value is 1, reaching the theoretical maximum value. In this case, the compression effect evaluation value will inevitably exceed the preset effect threshold, automatically skipping the initial compression ratio update without incurring additional computational overhead. Furthermore, the compression force signal obtained through lossless compression is the target compression force signal subsequently transmitted to the remote operating terminal.

[0054] Secondly, based on the restored force signal and the original force feedback signal, the signal error for each component is determined. Since the force feedback signal contains three axial force components and three about-axis torque components, and these components have different physical dimensions and numerical ranges, directly comparing the signal errors of each component would result in the force component error dominating the evaluation due to its larger amplitude, thus masking the distortion of the torque component. Therefore, for each axial force component and each about-axis torque component, the mean square error between the restored force signal and the original force feedback signal is calculated, and then the mean square errors of the force and torque components are normalized. The purpose of normalization is to map the mean square errors of different dimensions and amplitude ranges to the same scale, making the distortion levels of the force and torque components comparable. Normalization can be achieved by dividing the mean square error of each component by the square of the variance or peak value of the original force feedback signal for that component, thus eliminating differences in dimensions and amplitude. The maximum value among the normalized mean square errors of all components is taken as the signal error of that data segment, ensuring that it is most sensitive to distortion in the component with the worst force fidelity. For example, the formula for calculating the mean square error of each component is as follows: In the formula, For the first The mean square error of the nth component is used to characterize the nth component during compression. The degree of information loss for each component is indicated by its value; a larger value indicates a more severe information loss. Use component indices to iterate through the three axial force components and the three about-axis moment components; The total number of sampling points involved in the calculation is the number of sampling points within the current data segment; The sampling point number; The first force feedback signal The sampling point of the first sampling point The amplitude of each component; The first in the reducing force signal The sampling point of the first sampling point The amplitude of each component.

[0055] After obtaining the mean square error of each component, it is normalized based on the root mean square value or peak value of the original force feedback signal of that component in the data segment to eliminate the difference in the dimension and amplitude range of that component; then the maximum value of the normalized mean square error is determined as the signal error of that data segment.

[0056] Meanwhile, to evaluate the fidelity of the compression on the force feedback signal waveform structure from the signal morphology perspective, a pre-defined correlation analysis algorithm is used to determine the correlation coefficient between the original force feedback signal and the restored force signal for each component. Different components may have different waveform change patterns; if correlation is calculated by mixing them, the amplitude differences will lead to evaluation distortion. Therefore, the correlation coefficient is calculated separately for each component, and the minimum value among all component correlation coefficients is taken as the correlation coefficient for that data segment. The minimum value is chosen because a lower correlation coefficient indicates worse waveform fidelity, and the minimum value reflects the waveform distortion of the worst component. The correlation coefficient itself is a dimensionless indicator, unaffected by differences in the amplitude range of each component, allowing for direct comparison among the components.

[0057] For example, the preset correlation analysis algorithm can be cosine similarity, and the calculated correlation coefficient ranges from -1 to 1. When the value approaches 1, it means that the waveform direction of the restored force signal of this component is consistent with that of the original force feedback signal, and the shape is highly similar, and the compression has a high fidelity to the signal structure. When the value approaches 0, it means that there is almost no correlation between the waveform shape of the two, and the compression has seriously damaged the time domain structure of the signal. When the value approaches -1, it means that the waveform directions of the two are completely opposite, and the direction of the force is reversed, which is a serious and dangerous signal error.

[0058] After obtaining the signal error and the processed correlation coefficient, a compression effect evaluation value is generated based on both. The signal error measures the degree of information loss caused by compression at the amplitude level; a smaller error indicates higher amplitude fidelity. Since the signal error has already been normalized and its value ranges from 0 to 1, it is converted into a positive index in the same direction as the correlation coefficient, using the formula "..." The amplitude fidelity is expressed as a parameter in the form of "". When the signal error is close to 0, the amplitude fidelity is close to 1, indicating almost no loss of amplitude information; when the signal error is close to 1, the amplitude fidelity is close to 0, indicating severe distortion of amplitude information. The correlation coefficient measures the degree of fidelity of the compression on the signal waveform structure at the morphological level. Its value is itself a positive indicator. When the correlation coefficient is close to 1, it indicates that the waveforms are highly similar; when it is close to 0, it indicates almost no waveform correlation; and when it is negative, it indicates that the waveforms are in opposite directions, which is a serious and dangerous signal error. The amplitude fidelity and morphological fidelity are weighted and summed to comprehensively evaluate the compression effect. The weights of the two can be adjusted according to the different emphases on amplitude accuracy and waveform morphology in the actual teleoperation task. The closer the compression effect evaluation value is to 1, the better the overall compression effect. For example, the formula for calculating the compression effect evaluation value is as follows: In the formula, The value is used to evaluate the compression effect, and its range is [value range missing]. Up to 1; The correlation coefficient is the minimum value among all component correlation coefficients, and its value ranges from -1 to 1. This is the normalized signal error, with a value ranging from 0 to 1; For amplitude fidelity, the error is converted into a positive index, and the larger the value, the higher the amplitude fidelity. The preset weighting coefficients range from 0 to 1. For example, a value of 0.5 can represent equal weights for amplitude fidelity and waveform fidelity. If the remote control task requires higher waveform fidelity, the value can be increased. If higher amplitude accuracy is required, the value can be reduced. .when Close to 1 and When approaching 0, A value close to 1 indicates excellent amplitude and shape fidelity for each component; when... Reduce or When it increases, A corresponding decrease indicates a worse compression effect; when the correlation coefficient is negative, Using negative values ​​in the weighted summation makes Significantly reduced or even negative, and The smaller the value, the lower the compression effect evaluation value, making it easier for the compression effect evaluation value to fall below the preset effect threshold, thus triggering an update of the compression ratio.

[0059] After obtaining the compression effect evaluation value, determine whether the initial compression ratio needs to be updated.

[0060] Specifically, when the compression effect evaluation value is lower than the preset effect threshold (e.g., an empirical value of 0.7), it indicates that the current compression level is too high, causing the loss of force information to exceed the acceptable range. Therefore, the compression level needs to be reduced, i.e., the compression ratio needs to be decreased. The update magnitude of the compression ratio is determined based on the difference between the compression effect evaluation value and the preset effect threshold; the worse the compression effect, the larger the update magnitude. The updated initial compression ratio is obtained by multiplying the initial compression ratio before the update by the compression effect evaluation value and rounding down. To ensure that the compression ratio is always an integer not less than 1, the larger of the rounded value and 1 is taken as the final update result. For example, the formula for determining the updated initial compression ratio is as follows: In the formula, This is the updated initial compression ratio; This represents the initial compression ratio before the update. This is the evaluation value for compression effect; Indicates rounding down; This indicates that the larger of the two values ​​should be taken to ensure that the initial compression ratio after the update is not less than 1. When If the value is greater than or equal to the preset effect threshold, no update is triggered, and the initial compression ratio remains unchanged. (Updated initial compression ratio) It is used to replace the original initial compression ratio and to re-perform compression processing and compression effect evaluation based on the updated initial compression ratio.

[0061] It needs to be explained that, in When the value is 1, it indicates that the current network conditions and compression effect require no lossy compression of the force feedback signal. In this case, lossless compression processing is directly applied to the force feedback signal, and the data after lossless compression is used as the target compression force signal. Subsequent filtering and compression steps based on the target compression ratio are no longer performed, because when the compression ratio is 1, the compression processing is equivalent to lossless transmission, and lossless compression processing can reduce the data volume to a certain extent while maintaining signal integrity.

[0062] Repeat the above process: after each update, re-compress the force feedback signal based on the updated initial compression ratio to obtain a new compression force signal, then decompress and restore the new compression force signal, calculate the new signal error, correlation coefficient and compression effect evaluation value, and determine whether to continue updating until the compression effect evaluation value reaches or exceeds the preset effect threshold.

[0063] When the compression effect evaluation value reaches or exceeds the preset effect threshold, it means that the current compression ratio has achieved an acceptable balance between the force fidelity and the degree of compression of each component of the compression force signal. At this time, the current initial compression ratio is determined as the target compression ratio.

[0064] After obtaining the target compression ratio, the force feedback signal is compressed based on the target compression ratio to obtain the target compression force signal.

[0065] Specifically, for data segments classified as stationary phases, the force feedback signal within the data segment is filtered to remove high-frequency transient components, and the filtered force feedback signal is compressed based on the target compression ratio to obtain the target compression force signal; for data segments classified as non-stationary phases, the force feedback signal within the data segment is compressed based on the target compression ratio to obtain the target compression force signal.

[0066] It should be noted that the division between the stationary and non-stationary phases here follows the results of the phase division based on the time-domain variation characteristics in the previous steps. The same phase division results are used for the same data segment in the initial compression and target compression phases, only the compression ratio is replaced by the target compression ratio instead of the initial compression ratio.

[0067] Understandably, through the aforementioned closed-loop iterative update process, the initial compression ratio is further optimized and adjusted based on network transmission parameters and temporal domain variation characteristics, combined with feedback on the actual fidelity of the compressed signal of each component. When the compression degree is too high, resulting in excessive loss of force information or destruction of waveform structure in any component, the correlation coefficient of that component decreases or the signal error increases, leading to a decrease in the compression effect evaluation value. This triggers an update of the compression ratio, reducing the compression ratio and the degree of compression to retain more force details. When the signal direction of any component reverses, the correlation coefficient of that component becomes negative. After taking the minimum value, the overall correlation coefficient becomes negative, further reducing the compression effect evaluation value, even to a negative value. The compression ratio update amplitude is larger, and the degree of compression decreases rapidly. When the compression degree is appropriate, the compression effect evaluation value reaches or exceeds the preset effect threshold. The current compression ratio is determined as the target compression ratio, and subsequent continuous compression processing is performed using this target compression ratio. This ensures the operator's sense of force presence while adapting the data transmission volume to changes in network transmission conditions.

[0068] S105. Transmit the target compression force signal to the remote control terminal, and send the control command generated by the remote control terminal according to the target compression force signal to the robot controller to drive the actuator at the end of the robot to perform the operation.

[0069] The target compression force signal obtained through the above steps achieves an optimal balance between compression efficiency and signal fidelity, enabling it to carry complete force interaction information with a relatively small amount of data. In the closed-loop control process of robot remote operation, the target compression force signal needs to be transmitted to the remote operator via a communication link. The operator then generates control commands based on force perception and sends these commands back to the robot to drive the actuators to complete the contact operation, thus realizing a complete closed loop from compression transmission to remote execution of the force feedback signal.

[0070] As one possible approach, the received target compression force signal is first decompressed and restored at the remote operating terminal to obtain the restored force signal.

[0071] It should be noted that the specific decompression and restoration method corresponds to the compression processing method. If the compression processing uses a combination of downsampling and interpolation restoration, the sampling points are restored through the corresponding interpolation algorithm during decompression and restoration. If the compression processing uses quantization level adjustment, inverse quantization restoration is performed during decompression and restoration to restore the amplitude of each sampling point to the original quantization precision.

[0072] Then, the restored force signal is used to generate a feedback force through a force feedback device.

[0073] Optionally, the force feedback device is a force perception device installed at the remote operating end, capable of generating corresponding force or torque outputs in each degree of freedom direction based on the input force signal. The force feedback device receives the axial force components and the about-axis torque components from the restored force signal, converting them into feedback forces perceptible to the operator, allowing the operator to obtain a sense of force presence corresponding to the actual contact state of the robot's end effector. The operator judges the current working state based on the perceived feedback force, including the direction, amplitude, and trend of the contact force, and then decides on the next operational action.

[0074] Furthermore, motion data generated by the input device in response to the operator's actions is collected, and control commands are generated based on the motion data.

[0075] Optionally, the input device is a motion acquisition device installed at the remote operating end, capable of detecting the displacement, velocity, and applied force of the operator's hand in each degree of freedom in real time, and converting these physical quantities into motion data. The motion data includes the target position and orientation that the operator expects the robot's end effector to reach, as well as the desired contact force or torque. Control commands are then generated based on the motion data. These commands carry the force and torque parameters required to drive the robot's end effector to perform the corresponding task, ensuring that the control commands correspond to the operator's applied actions in both direction and amplitude.

[0076] Finally, the control commands are transmitted to the robot controller via the network to drive the actuators at the robot's end effector to perform the work actions corresponding to the control commands.

[0077] Optionally, the robot controller is a motion control unit located at the robot's end, responsible for receiving control commands and converting them into drive signals for each joint motor. Based on the force and torque parameters in the control commands, the robot controller controls the actuators at the robot's end effector to generate corresponding contact forces and torques with the workpiece, driving the actuators to perform contact-based operational actions corresponding to the manipulation. New force interaction information generated by the actuators during operation is collected by a six-dimensional force sensor installed at the end flange, forming new force feedback signals, which then enter the next cycle of compression, transmission, and remote control. This forms a complete closed loop from force feedback signal acquisition, compression processing, transmission, force perception reproduction, and operation action acquisition to control command execution. The operator can then use a sense of force presence to guide smooth and precise remote control of the robot.

[0078] Understandably, in the robot teleoperation method based on direct force feedback in flexible manufacturing provided in this embodiment of the invention, the force feedback signal from the robot end effector and network transmission parameters are collected synchronously. The force feedback signal is then compressed based on the network transmission parameters and the temporal variation characteristics of the force feedback signal. This allows the compression level to be dynamically adjusted according to changes in network conditions and operating states. When real-time available bandwidth is sufficient and transmission latency is good, a baseline compression level is used to balance data integrity and transmission efficiency. When network conditions deteriorate, the compression level is adaptively adjusted to ensure real-time transmission. When the operation is in a stable phase, high-frequency transient components are filtered out to further reduce the data volume. Simultaneously, based on… The signal error and correlation of the original force signal relative to the original force feedback signal are used to perform closed-loop iterative updates on the initial compression ratio. This allows the compression ratio to automatically reduce the compression degree to preserve force information when the compression degree is too high and force information is lost. When severe distortions such as signal direction reversal occur, the evaluation value naturally decreases and triggers a corresponding adjustment of the compression ratio. Ultimately, a target compression force signal that achieves a balance between force fidelity and real-time transmission is obtained. This enables the processing of the force feedback signal to continuously adapt to changes in network status and signal characteristics throughout the teleoperation process. While ensuring the operator's sense of force presence and operational accuracy, it effectively reduces data transmission volume, alleviates network burden, and improves the real-time performance and stability of teleoperation.

Claims

1. A robot remote control method based on direct force feedback in flexible manufacturing, characterized in that, The method includes: The force feedback signal and network transmission parameters of the robot end are collected. The force feedback signal includes multiple axial force components and multiple about-axis torque components. The network transmission parameters include real-time available bandwidth and transmission delay. The initial compression ratio is determined based on the network transmission parameters, and the force feedback signal is compressed based on the initial compression ratio, taking into account the time-domain variation characteristics of the force feedback signal, to obtain a compression force signal. Based on the signal error and correlation between the restored force signal obtained after restoring the compression force signal and the force feedback signal, the initial compression ratio is updated to obtain the target compression ratio, and the force feedback signal is compressed based on the target compression ratio to obtain the target compression force signal. The target compression force signal is transmitted to the remote control terminal, and the control command generated by the remote control terminal based on the target compression force signal is sent to the robot controller to drive the actuator at the end of the robot to perform the operation.

2. The robot remote control method based on direct force feedback in flexible manufacturing according to claim 1, characterized in that, Determining the initial compression ratio based on the network transmission parameters includes: The bandwidth index is determined based on the real-time available bandwidth and the preset baseline bandwidth. The transmission delay index is determined based on the transmission delay and the preset baseline transmission delay. The compression ratio adjustment coefficient is determined based on the bandwidth index and the transmission delay index. The initial compression ratio is determined based on the compression ratio adjustment coefficient and the preset compression ratio.

3. The robot remote control method based on direct force feedback in flexible manufacturing according to claim 1, characterized in that, Based on the time-domain variation characteristics of the force feedback signal, and using the initial compression ratio, the force feedback signal is compressed to obtain a compression force signal, including: The collected force feedback signal is segmented into data segments with a preset duration as the period; Based on the time-domain variation characteristics, the data segment is divided into a stationary phase or a non-stationary phase; For the data segment divided into the stable phase, the force feedback signal within the data segment is filtered, and the filtered force feedback signal is compressed based on the initial compression ratio to obtain the compression force signal. For the data segment divided into the non-stationary stage, the force feedback signal within the data segment is compressed based on the initial compression ratio to obtain the compression force signal.

4. The robot remote control method based on direct force feedback in flexible manufacturing according to claim 3, characterized in that, Based on the aforementioned time-domain variation characteristics, each data segment is divided into stationary or non-stationary phases, including: For the first data segment, the data segment is divided into non-stationary phases; For data segments other than the first one, a first nonstationarity index is determined based on the force feedback signal within the data segment and the force feedback signal within the adjacent preceding data segment. The first nonstationarity index is used to characterize the degree of signal fluctuation between adjacent data segments. Based on the number of sampling points of amplitude abrupt change in the force feedback signal within the data segment, a second nonstationarity index is determined. The second nonstationarity index is used to characterize the degree of local signal abrupt change within the data segment. The first nonstationarity index and the second nonstationarity index constitute the time-domain variation characteristics. When the first nonstationarity index is less than the first nonstationarity threshold and the second nonstationarity index is less than the second nonstationarity threshold, the data segment is divided into the stationary phase; otherwise, the data segment is divided into the nonstationary phase.

5. The robot remote control method based on direct force feedback in flexible manufacturing according to claim 4, characterized in that, Based on the force feedback signal within the data segment and the force feedback signal within the adjacent preceding data segment, a first non-stationarity index is determined, including: For each component of the force feedback signal, the signal variance deviation rate of the component is determined based on the variance of the component in the data segment and the variance of the component in the adjacent preceding data segment. The maximum value among the signal variance deviation rates of all components is determined as the first nonstationarity index; The second non-stationarity index is determined based on the number of sampling points where the amplitude of the force feedback signal changes abruptly within the data segment, including: For each component included in the force feedback signal, the amplitude change of the component at each sampling point other than the first sampling point in the data segment is determined, and the number of sampling points where the amplitude change exceeds a preset amplitude change threshold is determined as the number of amplitude change abrupt sampling points of the component. The maximum value among the number of sampling points for amplitude abrupt changes in all components is determined as the second nonstationarity index.

6. The robot remote control method based on direct force feedback in flexible manufacturing according to claim 1, characterized in that, Based on the signal error and correlation between the restored force signal obtained after restoring the compression force signal and the force feedback signal, the initial compression ratio is updated to obtain the target compression ratio, including: The compression force signal is decompressed and restored to obtain the restored force signal. Based on the restored force signal and the force feedback signal, the signal error is determined. The signal error is used to characterize the degree of information loss of the force feedback signal during the compression process. A preset correlation analysis algorithm is used to determine the correlation coefficient between the force feedback signal and the restoration force signal; The compression performance evaluation value is determined based on the correlation coefficient and the signal error. If the compression effect evaluation value is lower than the preset effect threshold, the initial compression ratio is updated according to the compression effect evaluation value, and the compression effect evaluation value is re-determined based on the updated initial compression ratio, until the compression effect evaluation value reaches or exceeds the preset effect threshold. The initial compression ratio that reaches or exceeds the preset effect threshold is determined as the target compression ratio.

7. The robot remote control method based on direct force feedback in flexible manufacturing according to claim 3, characterized in that, Based on the target compression ratio, the force feedback signal is compressed to obtain a target compression force signal, including: For the data segment divided into the stable phase, the force feedback signal within the data segment is filtered, and the filtered force feedback signal is compressed based on the target compression ratio to obtain the target compression force signal; For the data segment divided into the non-stationary stage, the force feedback signal within the data segment is compressed based on the target compression ratio to obtain the target compression force signal.

8. The robot remote control method based on direct force feedback in flexible manufacturing according to claim 3, characterized in that, After obtaining the data segment, the method further includes: Obtain the maximum force amplitude of each axial force component and the maximum torque amplitude of each about-axis torque component within the current data segment; The first safe proximity is determined based on the maximum force amplitude and the preset force safety threshold; The second safe proximity is determined based on the maximum torque amplitude and the preset torque safety threshold. If the first security proximity is greater than a preset security ratio or the second security proximity is greater than the preset security ratio, the force feedback signal in the current data segment is subjected to lossless compression processing, and the data after lossless compression processing is used as the compression force signal.

9. The robot remote control method based on direct force feedback in flexible manufacturing according to claim 1, characterized in that, The target compression force signal is transmitted to a remote control terminal, and the control command generated by the remote control terminal based on the target compression force signal is sent to the robot controller to drive the actuator at the robot's end effector to perform the operation, including: The target compression force signal is decompressed and restored at the remote operation terminal, and the decompressed and restored force signal is used to generate a feedback force through a force feedback device. Collect motion data generated by the input device in response to the operation action, and generate the control command based on the motion data; The control command is transmitted to the robot controller to drive the actuator at the robot end to perform the operation corresponding to the control command.

10. The robot remote control method based on direct force feedback in flexible manufacturing according to claim 1, characterized in that, Collect force feedback signals and network transmission parameters from the robot's end effector, including: The robot end effector uses a six-dimensional force sensor installed at its end to collect multiple axial force components and multiple axial torque components at a preset sampling frequency. The system collects real-time available bandwidth and transmission latency between the robot and the remote control terminal using a network monitoring tool at a preset update frequency.

Images