A method and system for online compensation of aging in ion-conductive continuous dielectric skin

Abstract

This invention discloses an online compensation method and system for aging of ion-conducting continuous dielectric skin. The method includes the following steps: S1, determining whether the system has entered a non-interactive idle state; S2, constructing a complex impedance spectrum matrix containing amplitude and phase information; S3, outputting an aging drift vector including electrode contact impedance; S4, based on the aging drift vector, updating the mapping weights or sensitivity matrix in the tactile inversion model in real time to compensate for tactile perception deviations caused by material aging; S5, monitoring based on the output electrode contact impedance, and when the contact impedance of any electrode exceeds a predetermined threshold, cutting off the failed electrode channel, and reconstructing the failed area using virtual electrode mapping based on the electric field distribution formed by the remaining electrodes to maintain the integrity of tactile perception. This invention aims to solve the problem of tactile perception drift caused by the evolution of materials and interfaces over time in ion-conducting skin, achieving low-cost online health assessment and adaptive compensation.

Description

Technical Field

[0001] This invention relates to the field of intelligent skin technology for service robots / humanoid robots, specifically to an online aging compensation method and system for ion-conducting continuous dielectric skin. Background Technology

[0002] Ionic skin belongs to an electrochemical system, and its material properties and interface states evolve non-stationarily over time. During long-term use, factors such as moisture penetration / escape, ion migration, polymer microcrack propagation, and electrode-dielectric interface contact degradation can cause the factory-calibrated tactile mapping matrix to gradually fail, thereby affecting the accuracy and stability of tactile perception.

[0003] Currently, tactile inversion of ion-conducting continuous dielectric skin mainly relies on the traditional method of "fixed parameters + one-time calibration". However, due to the combined effects of material aging and environmental disturbances, tactile response shows significant drift within a timescale of minutes to days, leading to the continuous accumulation of inversion errors and the current situation where products are "peak at the factory, but fail after prolonged use".

[0004] In existing technologies, electrochemical impedance spectroscopy can be used to characterize ion-derived skin in a laboratory setting to assess its aging level. However, this method relies on expensive and bulky specialized equipment, making it impossible to directly embed low-cost edge modules, and further hindering online, real-time health monitoring and closed-loop compensation. Therefore, there is an urgent need for a method and system capable of online assessment of the aging status of ion-derived skin and adaptive compensation of the tactile inversion model. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, one of the objectives of this invention is to provide an online compensation method for aging of ion-conductive continuous medium skin, so as to solve the problem of tactile perception drift caused by the evolution of materials and interfaces over time in ion-conductive skin, and to achieve low-cost online health assessment and adaptive compensation.

[0006] The second objective of this invention is to provide an online aging compensation system for ion-conductive continuous medium skin, in order to solve the problem of tactile perception drift caused by the evolution of materials and interfaces over time in ion-conductive skin, and to achieve low-cost online health assessment and adaptive compensation.

[0007] To achieve one of the objectives of this invention, the following solution is adopted: An online aging compensation method for ion-conducting continuous dielectric skin includes the following steps: S1. Synchronously monitor proximity signals, tactile change rate, and load characteristics of the drive unit to determine whether the system has entered a non-interactive idle state; S2. In response to the determination of entering the idle state, multi-frequency excitation is applied to the sparse edge electrode array and complex impedance response is acquired to construct a complex impedance spectrum matrix containing amplitude and phase information. S3. Compare the complex impedance spectrum matrix with the pre-stored healthy baseline, and use the feature decoupling algorithm to separate the offset into reversible environmental disturbance components and irreversible structural aging components, and output the aging drift vector including electrode contact impedance. S4. Based on the aging drift vector, update the mapping weight or sensitivity matrix in the tactile inversion model in real time to compensate for the tactile perception deviation caused by material aging. S5. Based on the output electrode contact impedance, the contact impedance of any electrode exceeds a predetermined threshold. The failed electrode channel is cut off, and the failed area is reconstructed by virtual electrode mapping based on the electric field distribution formed by the remaining electrodes, so as to maintain the integrity of tactile perception.

[0008] Further, step S1 includes: determining that the robot is in a non-interactive idle state by using the back electromotive force or current ripple of the drive unit.

[0009] Furthermore, in step S2, the set of frequency points for the multi-frequency excitation covers logarithmically spaced frequency points from 10 Hz to 100 kHz, and each scan includes no less than 8 frequency points.

[0010] Further, in step S3, the feature decoupling algorithm includes performing equivalent circuit fitting on the complex impedance spectrum matrix to extract a parameter vector including bulk resistance, charge transfer resistance, interface capacitance, and electrode contact impedance.

[0011] Furthermore, in step S3, the reversible environmental disturbance component is driven by temperature or humidity variables and estimated through characteristic quantities in the mid-frequency range or an independent temperature sensing channel.

[0012] Furthermore, in step S4, the real-time update includes stability constraints, which include at least one of step size upper limit, hysteresis threshold or confidence threshold. When the amplitude of the aging drift vector is lower than a preset threshold, the weight update is paused to avoid short-term noise interference.

[0013] Furthermore, in step S5, the predetermined threshold is set based on the drift rate or absolute value of the electrode contact impedance extracted in the state decoupling step.

[0014] Furthermore, in step S5, the virtual electrode mapping reconstruction includes recalculating or updating the sensitivity matrix, and performing spatial interpolation or regularized reconstruction based on the physical forward model on the failure region in tactile inversion.

[0015] Furthermore, it also includes step S6: based on the historical change trajectory of the aging drift vector accumulated by executing step S3 multiple times, predict the time when it reaches the failure threshold, and introduce the Arrhenius model to calculate the acceleration factor based on the temperature history to correct the prediction result.

[0016] To achieve the second objective of this invention, the following solution is adopted: An online aging compensation system for ion-conducting continuous dielectric skin includes: The idle state determination module is used to synchronously monitor proximity signals, tactile change rate, and load characteristics of the drive unit to determine whether the system has entered a non-interactive idle state. The self-excited scanning module is used to apply multi-frequency excitation to the sparse edge electrode array and collect complex impedance response in response to the determination of entering the idle state, and construct a complex impedance spectrum matrix containing amplitude and phase information. The fingerprint construction and decoupling module is used to compare the complex impedance spectrum matrix with the pre-stored health baseline, and to separate the offset into reversible environmental disturbance components and irreversible structural aging components through a feature decoupling algorithm, and output the aging drift vector including electrode contact impedance. An online compensation module is used to update the mapping weights or sensitivity matrix in the tactile inversion model in real time based on the aging drift vector, so as to compensate for the tactile perception deviation caused by material aging. The failure reconstruction module is used to monitor the contact impedance of the output electrodes. When the contact impedance of any electrode exceeds a predetermined threshold, the failure electrode channel is cut off, and the failure area is reconstructed by virtual electrode mapping based on the electric field distribution formed by the remaining electrodes, so as to maintain the integrity of tactile perception.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention achieves non-invasive online monitoring. By performing a self-excited scan during the system's non-interactive idle state, and utilizing the idle window to complete health fingerprint collection, this invention neither affects normal tactile interaction nor requires additional dedicated detection equipment, thus achieving low-cost, embedded online observation of the electrochemical state of ionic skin.

[0018] 2. This invention achieves decoupling between environmental disturbances and aging drift. Through a feature decoupling algorithm, this invention separates the offset of the complex impedance spectrum matrix into reversible environmental disturbance components and irreversible structural aging components, avoiding miscompensation caused by the mutual confusion between environmental factors such as temperature and humidity and material aging, thus ensuring the accuracy and stability of the tactile inversion model update.

[0019] 3. This invention realizes the dynamic evolution of the tactile inversion model. Based on the irreversible aging drift vector, this invention updates the mapping weight or sensitivity matrix of the tactile inversion model in real time, upgrading tactile perception from the traditional "static factory calibration" to "real-time state-driven dynamic mapping", which significantly suppresses the tactile accuracy deviation caused by material aging and interface degradation, and extends the effective service life of ion skin.

[0020] 4. This invention achieves self-healing capability for electrode failure. Based on real-time monitoring of the decoupled output electrode contact impedance, when a local electrode contact impedance exceeds the limit, the invention utilizes the redundancy of the electric field distribution in the continuous medium to perform spatial interpolation or regularized reconstruction of the failure area through virtual electrode mapping reconstruction. This avoids the defect of traditional matrix sensors where "one defective point means an entire area is blind," thus improving the system's fault tolerance and maintainability. Attached Figure Description

[0021] Figure 1 This is a block diagram of the online aging compensation system for ion-conducting continuous dielectric skin in an embodiment of the present invention. Figure 2 This is a schematic diagram of the decoupling of complex impedance spectrum characteristics in an embodiment of the present invention; wherein, L1 is the factory baseline, L2 is the reversible environmental disturbance curve, and L3 is the irreversible aging curve; Figure 3 This is a schematic diagram of remaining lifetime prediction in an embodiment of the present invention; Figure 4 This is a flowchart of the online aging compensation method for ion-conducting continuous dielectric skin in an embodiment of the present invention. Detailed Implementation

[0022] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0023] Example 1 like Figure 1-3 As shown, this embodiment of the invention provides an online aging compensation system for ion-conducting continuous dielectric skin, comprising an ion-conducting continuous dielectric skin layer and a sparse edge electrode array disposed at its edges. The skin layer is a single-layer continuous dielectric with no internal wiring and no discrete sensing units. The system also includes an edge electronics and algorithm module, the sparse edge electrode array being electrically connected to the edge electronics and algorithm module, which integrates the following: The idle state determination module is used to synchronously monitor proximity signals, tactile change rate, and load characteristics of the drive unit to determine whether the system has entered a non-interactive idle state. The self-excited scanning module is used to apply multi-frequency excitation to the sparse edge electrode array and collect complex impedance response in response to the determination of entering the idle state, and construct a complex impedance spectrum matrix containing amplitude and phase information. The fingerprint construction and decoupling module is used to compare the complex impedance spectrum matrix with the pre-stored health baseline, and to separate the offset into reversible environmental disturbance components and irreversible structural aging components through a feature decoupling algorithm, and output the aging drift vector including electrode contact impedance. An online compensation module is used to update the mapping weights or sensitivity matrix in the tactile inversion model in real time based on the aging drift vector, so as to compensate for the tactile perception deviation caused by material aging. The failure reconstruction module is used to monitor the contact impedance of the output electrodes. When the contact impedance of any electrode exceeds a predetermined threshold, the failure electrode channel is cut off, and the failure area is reconstructed by virtual electrode mapping based on the electric field distribution formed by the remaining electrodes, so as to maintain the integrity of tactile perception.

[0024] In this embodiment, the edge electronics and algorithm module includes an orthogonal demodulator for extracting real and imaginary features from the acquired complex impedance response.

[0025] In this embodiment, the specific content of the multi-criteria idle detection is as follows: To perform self-excited scanning without affecting normal tactile interaction, the system first needs to accurately determine whether it is currently in a non-interactive idle state. This invention uses a multi-criteria synchronous monitoring method for determination: 1. Proximity signal: Monitoring high-frequency capacitance components; when it is below a preset threshold and the duration exceeds a preset window, it indicates that no external object is approaching; 2. Tactile change rate: Monitoring the first-order difference or energy of low-frequency tactile features; when it is below a preset threshold, it indicates that there is no tactile interaction event; 3. Drive load: Monitoring the motor torque, current ripple, or back electromotive force of the robot drive unit; when it indicates that the robot is in a static, non-interactive state, it is determined that the idle condition is met; 4. Auxiliary confirmation: Optionally, auxiliary confirmation can be performed by combining external modal information such as vision and inertial measurement units to improve the reliability of the determination. When the above conditions are simultaneously met, the system determines that it has entered a non-interactive idle state and triggers the subsequent scanning process.

[0026] In this embodiment, the idle-time self-excited scanning and healthy fingerprint construction are as follows: In response to the determination of entering the idle state, the system performs multi-frequency self-excited scanning on the sparse edge electrode array. The scanning adopts a multi-injection mode, including but not limited to electrode pair injection, three-terminal injection of electrode groups, etc., and applies a low-amplitude excitation signal on a predetermined set of frequency points. The set of frequency points preferably covers logarithmically spaced frequency points from 10 Hz to 100 kHz, and each scan includes no less than 8 frequency points; the excitation amplitude is preferably 0.1V to 1V to reduce the impact on the ion skin state.

[0027] The complex impedance response of each injection-measurement combination is collected and defined as Z_ij(f_k) = Re(Z) + jIm(Z), where i and j are electrode indices and f_k is a frequency index. All measurement results constitute a complex impedance spectrum matrix. This matrix can be represented by a combination of amplitude |Z| and phase ∠Z, or it can be further represented by a parameter vector through equivalent circuit fitting, serving as the "health fingerprint" of the ion skin.

[0028] In this embodiment, the specific content of spectrum fitting and feature decoupling is as follows: To avoid misjudging environmental temperature drift as material aging and causing compensation overshoot, this invention introduces feature decoupling logic. For example... Figure 2 As shown, an equivalent circuit fitting is performed on the complex impedance spectrum matrix. The preferred approach is to use an extended Randles equivalent circuit model that includes bulk resistance Rs, charge transfer resistance Rct, interface capacitance Cdl, and diffusion term σ. The currently fitted parameter vector is compared with a pre-stored healthy baseline. The offset is decomposed into two components using spectral fitting or physical constraint mapping: 1. Reversible environmental disturbance component: parameter drift related to environmental variables such as temperature and humidity, which can be estimated using characteristic quantities in the mid-frequency range or independent temperature sensing channels; 2. Irreversible structural aging component: a long-term monotonic drift vector of parameters, such as a rightward shift of Rs, a decrease in Cdl, and an increase in Rct, reflecting the irreversible degradation of the material and interface. The decoupled output is the aging drift vector ΔA(t), used for subsequent tactile inversion model updates and remaining lifetime prediction.

[0029] In this embodiment, the online computation of the haptic inversion model involves mapping the decoupled aging drift vector ΔA(t) to the weight matrix, Jacobian matrix, or nonlinear mapping operator in the haptic inversion model for online updating. Stability constraints are introduced during the update process, including at least one of a step size upper limit, hysteresis threshold, and confidence threshold, to prevent overcorrection caused by short-term disturbances or noise. When the amplitude of the aging drift vector falls below a preset threshold, weight updates are paused to ensure system stability.

[0030] In this embodiment, the specific content of the virtual electrode reconstruction is as follows: when the contact impedance Rc of any electrode exceeds the preset safety threshold, the electrode is determined to be partially failed, and the corresponding electrode channel is cut off in the tactile inversion.

[0031] By leveraging the redundancy of the electric field distribution in a continuous medium, the sensitivity matrix J'(t) is recalculated based on the remaining effective electrode combinations. Virtual electrode mapping is reconstructed by spatial interpolation or regularized reconstruction based on a physical forward model of the failed region, thereby maintaining the tactile sensing availability of that region and avoiding the defect of traditional matrix sensors where "a single defective electrode results in a complete blind spot."

[0032] In this embodiment, the specific content of Remaining Life Prediction (RUL Prediction) is as follows: based on the historical change trajectory of the aging drift vector accumulated from multiple executions of step S3, the evolution trend of key parameters (such as Rs, Cdl, Rct, Rc) is analyzed to predict the time when they reach the failure threshold. Figure 3 As shown, the Arrhenius model can be introduced into the prediction process, and the acceleration factor can be calculated by combining historical temperature data, thereby correcting the failure node prediction results under different operating conditions and providing a quantitative basis for the system's preventive maintenance and degradation strategies.

[0033] In this embodiment, the health fingerprint matrix is ​​defined and its dimensions are as follows: The health fingerprint is preferably defined as a set of tensors or matrices formed by the complex impedance responses of N edge electrodes at M injection-measurement combinations and K frequency points. It can be represented in complex form as Z_ij(f_k), or by the aggregation of parameter vectors obtained from equivalent circuit fitting on each electrode pair.

[0034] In this embodiment, the decoupling algorithm is implemented as follows: Scheme A uses a physical fitting method, performing equivalent circuit fitting of the spectrum of each electrode pair using Randles or extended Randles to obtain the parameter vector, and then using a temperature model to separate the reversible environmental disturbance component; Scheme B uses a constraint decomposition method, decomposing the change in the frequency domain into a diffusion / interface component dominated by the low-frequency band and a bulk component dominated by the mid-to-high-frequency band, and setting reversible constraints on the environmental disturbance. Both schemes require the output of an irreversible aging drift vector for online updating, and provide confidence levels to control the update step size.

[0035] In this embodiment, the principle of physical field redundancy in virtual electrode mapping reconstruction is as follows: the electric field distribution of a continuous medium has inherent redundancy. In an edge-sparse electrode architecture, as long as the number of remaining electrodes meets the minimum observable condition (e.g., at least 3 or 4 effective injection paths), the original failure region can be covered by changing the current injection path. When the system detects that the contact impedance Rc of an electrode exceeds a predetermined threshold, it automatically removes the failed electrode and recalculates the sensitivity matrix, thereby achieving self-healing sensing.

[0036] In this embodiment, the remaining lifetime prediction and maintenance strategy is as follows: By monitoring the long-term evolution trajectory of EIS parameters and electrode contact impedance Rc, and combining it with a threshold model, the failure time is predicted, providing a quantitative basis for system maintenance, including strategies such as module replacement, repackaging, and degraded operation. Optionally, an Arrhenius model is introduced to perform equivalent conversion of lifetime under different temperature conditions to correct the prediction results.

[0037] Technical principles of this invention: Under the long-term influence of electric fields, temperature, humidity, and mechanical loads, ion-conducting media undergo slow and partially irreversible physicochemical evolution. These evolutions manifest as changes in bulk resistivity (reflecting material aging, changes in moisture content, and fluctuations in ion concentration), drift in interfacial capacitance and charge transfer resistance (reflecting electrode contact state and interfacial polarization), and alterations in diffusion impedance (reflecting ion activity). These changes cause the tactile mapping matrix at the factory to no longer correspond to the actual tactile input, resulting in a systematic deviation in tactile inversion. This is the fundamental problem that this invention aims to solve.

[0038] Complex impedance spectra contain rich physical information: bulk resistance reflects material aging, water content, and ion concentration; interfacial capacitance and charge transfer resistance reflect electrode contact state and interfacial polarization characteristics; and diffusion-related features reflect ion activity. Therefore, constructing a high-dimensional complex impedance spectrum matrix from complex impedance responses acquired under multi-electrode, multi-injection, and multi-frequency conditions can serve as an intrinsic state vector characterizing the state of ion-derived skin materials and interfaces, i.e., a "health fingerprint."

[0039] Based on the above principles, this invention constructs a complete logical closed loop: by performing self-excited scanning during the robot's non-interactive "idle" period, a healthy fingerprint at the current moment is obtained; then, using spectrum fitting or feature decomposition algorithms, the fingerprint offset is decomposed into reversible environmental disturbance components (such as conductivity changes caused by temperature) and irreversible structural aging components (such as a decrease in effective ion concentration and electrode contact degradation); on this basis, the weights of the tactile inversion model are updated online without introducing compensation overshoot, realizing a leap from "static fixed parameters" to "real-time state-driven", significantly extending the effective service life of the ion skin.

[0040] This invention presents a low-cost, embeddable online health fingerprint and compensation closed-loop solution, upgrading the tactile inversion of ionic skin from traditional "static mapping" to "real-time state-driven dynamic mapping," thereby solving the problem of tactile perception drift caused by the evolution of materials and interfaces over time. This invention decouples the system through spectral features, comparing the complex impedance spectrum matrix obtained by idle-time self-excitation scanning with a pre-stored health baseline. Using spectral fitting or physical constraint mapping, the offset is separated into reversible environmental disturbance components and irreversible structural aging components, thus avoiding miscompensation caused by the mutual confusion between environmental factors such as temperature and humidity and material aging, ensuring the accuracy and stability of the tactile inversion model update. This invention reconstructs the system through virtual electrode mapping, performing real-time monitoring based on the decoupled output electrode contact impedance. When the contact impedance of any electrode exceeds a predetermined threshold, the redundancy of the electric field distribution in the continuous medium is utilized to trim the failed electrode channel and perform spatial interpolation or regularization reconstruction of the failed area based on the remaining electrodes, achieving "self-healing perception." This avoids the defect of traditional matrix sensors where "one defect means a blind spot," improving the system's fault tolerance and maintainability. Furthermore, this invention provides a set of reproducible and verifiable procedures and threshold definitions, including idle state multi-criteria judgment, scanning frequency band and frequency point setting, parameter fitting and aging drift vector extraction, contact impedance over-limit triggering reconstruction mechanism, and remaining lifetime prediction model, providing quantitative basis for the engineering application of the system. This invention can achieve online health assessment and adaptive compensation without introducing expensive impedance analyzers, improving the long-term stability and maintainability of ion skin.

[0041] Experimental Example Experiment 1: To verify the effectiveness of idle-time scanning and decoupling compensation, the number of electrodes N=12 and the frequency K=12 (logarithmic interval from 10Hz to 100kHz) were set, with each idle-time window scan lasting less than 2 seconds. Extended Randle fitting was used to extract the parameter vector {Rs,Rct,Cdl,σ}, and the reversible component was estimated using the temperature channel, outputting the irreversible aging drift vector ΔA(t). The weights of the tactile inversion model were updated online based on ΔA(t), with a maximum step size and confidence threshold set. Experiments demonstrate that this method effectively suppresses tactile accuracy deviations caused by material aging.

[0042] Experiment 2: To verify the feasibility of electrode failure and virtual reconstruction, single-electrode contact degradation was artificially introduced, causing the electrode contact impedance Rc to exceed the safety threshold. The system automatically pruned the failed electrode channel and recalculated the sensitivity matrix J'(t) based on the remaining electrodes, performing spatial interpolation regularization reconstruction on the failed region. Tests showed that the tactile mapping in the reconstructed region remained usable, and the system's fault tolerance was significantly improved.

[0043] Control group: Without decoupling and online compensation, the traditional fixed parameter plus one calibration method was used, and tactile inversion was performed only using the factory calibration matrix. Under temperature changes and material aging conditions, the tactile mapping error accumulated over time, showing significant drift, and lacked a virtual reconstruction mechanism. Electrode failure led to a fixed sensing blind zone.

[0044] To verify the effectiveness of this invention, the following testing protocol and evaluation indicators were used: Idle state determination threshold definition: A unified definition of proximity signal threshold, tactile change rate threshold, motor load threshold, and corresponding time window is used to verify the accuracy of idle state determination.

[0045] Scanning protocol settings: clearly define the frequency point set, excitation amplitude, injection / measurement combination method, sampling time, and filtering strategy to ensure the reliability and repeatability of the complex impedance spectrum matrix construction.

[0046] Decoupling effectiveness verification: The reproducibility of reversible environmental disturbance components is verified under controllable temperature and humidity conditions; the monotonic drift characteristics of irreversible structural aging components are verified under accelerated aging conditions to prove the correctness of the feature decoupling algorithm.

[0047] Evaluation of compensation effect: The improvement effect of perception accuracy after dynamic evolution of the tactile inversion model is evaluated by using the monotonicity, repeatability, long-term drift rate and tactile positioning error of pressure-impedance mapping as indicators.

[0048] Failure reconstruction capability test: Record the electrode contact impedance Rc over-limit event, the execution of the trimming strategy, the change of tactile mapping error after reconstruction, and the available sensing area to verify the fault tolerance capability of virtual electrode mapping reconstruction.

[0049] Example 2 like Figure 4 As shown, this embodiment of the invention provides an online compensation method for aging of ion-conducting continuous dielectric skin. It constructs a healthy fingerprint by performing multi-frequency self-excited complex impedance scanning during idle time and decomposes environmental disturbances and irreversible aging, thereby realizing the online evolution of the tactile inversion model and the virtual reconstruction of electrode failure.

[0050] The online aging compensation method for ion-conducting continuous dielectric skin according to an embodiment of the present invention includes the following steps: S1. Synchronously monitor proximity signals, tactile change rate, and load characteristics of the drive unit to determine whether the system has entered a non-interactive idle state; S2. In response to the determination of entering the idle state, multi-frequency excitation is applied to the sparse edge electrode array and complex impedance response is acquired to construct a complex impedance spectrum matrix containing amplitude and phase information. S3. Compare the complex impedance spectrum matrix with the pre-stored healthy baseline, and use the feature decoupling algorithm to separate the offset into reversible environmental disturbance components and irreversible structural aging components, and output the aging drift vector including electrode contact impedance. S4. Based on the aging drift vector, update the mapping weight or sensitivity matrix in the tactile inversion model in real time to compensate for the tactile perception deviation caused by material aging. S5. Based on the output electrode contact impedance, the contact impedance of any electrode exceeds a predetermined threshold. The failed electrode channel is cut off, and the failed area is reconstructed by virtual electrode mapping based on the electric field distribution formed by the remaining electrodes, so as to maintain the integrity of tactile perception.

[0051] Further, step S1 includes: determining that the robot is in a non-interactive idle state by using the back electromotive force or current ripple of the drive unit.

[0052] Furthermore, in step S2, the set of frequency points for the multi-frequency excitation covers logarithmically spaced frequency points from 10 Hz to 100 kHz, and each scan includes no less than 8 frequency points.

[0053] Further, in step S3, the feature decoupling algorithm includes performing equivalent circuit fitting on the complex impedance spectrum matrix to extract a parameter vector including bulk resistance, charge transfer resistance, interface capacitance, and electrode contact impedance.

[0054] Furthermore, in step S3, the reversible environmental disturbance component is driven by temperature or humidity variables and estimated through characteristic quantities in the mid-frequency range or an independent temperature sensing channel.

[0055] Furthermore, in step S4, the real-time update includes stability constraints, which include at least one of step size upper limit, hysteresis threshold or confidence threshold. When the amplitude of the aging drift vector is lower than a preset threshold, the weight update is paused to avoid short-term noise interference.

[0056] Furthermore, in step S5, the predetermined threshold is set based on the drift rate or absolute value of the electrode contact impedance extracted in the state decoupling step.

[0057] Furthermore, in step S5, the virtual electrode mapping reconstruction includes recalculating or updating the sensitivity matrix, and performing spatial interpolation or regularized reconstruction based on the physical forward model on the failure region in tactile inversion.

[0058] Furthermore, it also includes step S6: based on the historical change trajectory of the aging drift vector accumulated by executing step S3 multiple times, predict the time when it reaches the failure threshold, and introduce the Arrhenius model to calculate the acceleration factor based on the temperature history to correct the prediction result.

[0059] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A method for online compensation of aging in ion-conducting continuous dielectric skin, characterized in that, Includes the following steps: S1. Synchronously monitor proximity signals, tactile change rate, and load characteristics of the drive unit to determine whether the system has entered a non-interactive idle state; S2. In response to the determination of entering the idle state, multi-frequency excitation is applied to the sparse edge electrode array and complex impedance response is acquired to construct a complex impedance spectrum matrix containing amplitude and phase information. S3. Compare the complex impedance spectrum matrix with the pre-stored healthy baseline, and use the feature decoupling algorithm to separate the offset into reversible environmental disturbance components and irreversible structural aging components, and output the aging drift vector including electrode contact impedance. S4. Based on the aging drift vector, update the mapping weight or sensitivity matrix in the tactile inversion model in real time to compensate for the tactile perception deviation caused by material aging. S5. Based on the output electrode contact impedance, the contact impedance of any electrode exceeds a predetermined threshold. The failed electrode channel is cut off, and the failed area is reconstructed by virtual electrode mapping based on the electric field distribution formed by the remaining electrodes, so as to maintain the integrity of tactile perception.

2. The online aging compensation method for ion-conducting continuous dielectric skin according to claim 1, characterized in that, Step S1 includes: determining that the robot is in a non-interactive idle state by using the back electromotive force or current ripple of the drive unit.

3. The online aging compensation method for ion-conducting continuous dielectric skin according to claim 1, characterized in that, In step S2, the set of frequency points for the multi-frequency excitation covers logarithmically spaced frequency points from 10 Hz to 100 kHz, and each scan includes no less than 8 frequency points.

4. The online aging compensation method for ion-conducting continuous dielectric skin according to claim 1, characterized in that, In step S3, the feature decoupling algorithm includes performing equivalent circuit fitting on the complex impedance spectrum matrix to extract a parameter vector including bulk resistance, charge transfer resistance, interface capacitance, and electrode contact impedance.

5. The online aging compensation method for ion-conducting continuous dielectric skin according to claim 4, characterized in that, In step S3, the reversible environmental disturbance component is driven by temperature or humidity variables and estimated through characteristic quantities in the mid-frequency range or an independent temperature sensing channel.

6. The online aging compensation method for ion-conducting continuous dielectric skin according to claim 1, characterized in that, In step S4, the real-time update includes stability constraints, which include at least one of step size upper limit, hysteresis threshold or confidence threshold. When the amplitude of the aging drift vector is lower than the preset threshold, the weight update is paused to avoid short-term noise interference.

7. The online aging compensation method for ion-conducting continuous dielectric skin according to claim 1, characterized in that, In step S5, the predetermined threshold is set based on the drift rate or absolute value of the electrode contact impedance extracted in the state decoupling step.

8. The online aging compensation method for ion-conducting continuous dielectric skin according to claim 1, characterized in that, In step S5, the virtual electrode mapping reconstruction includes recalculating or updating the sensitivity matrix, and performing spatial interpolation or regularized reconstruction based on the physical forward model on the failure region in tactile inversion.

9. The online aging compensation method for ion-conducting continuous dielectric skin according to claim 1, characterized in that, It also includes step S6: based on the historical change trajectory of the aging drift vector accumulated by executing step S3 multiple times, predict the time when it reaches the failure threshold, and introduce the Arrhenius model to calculate the acceleration factor based on the temperature history to correct the prediction result.

10. An online aging compensation system for ion-conducting continuous dielectric skin, characterized in that, include: The idle state determination module is used to synchronously monitor proximity signals, tactile change rate, and load characteristics of the drive unit to determine whether the system has entered a non-interactive idle state. The self-excited scanning module is used to apply multi-frequency excitation to the sparse edge electrode array and collect complex impedance response in response to the determination of entering the idle state, and construct a complex impedance spectrum matrix containing amplitude and phase information. The fingerprint construction and decoupling module is used to compare the complex impedance spectrum matrix with the pre-stored health baseline, and to separate the offset into reversible environmental disturbance components and irreversible structural aging components through a feature decoupling algorithm, and output the aging drift vector including electrode contact impedance. An online compensation module is used to update the mapping weights or sensitivity matrix in the tactile inversion model in real time based on the aging drift vector, so as to compensate for the tactile perception deviation caused by material aging. The failure reconstruction module is used to monitor the contact impedance of the output electrodes. When the contact impedance of any electrode exceeds a predetermined threshold, the failure electrode channel is cut off, and the failure area is reconstructed by virtual electrode mapping based on the electric field distribution formed by the remaining electrodes, so as to maintain the integrity of tactile perception.

Images