Adaptive flexible electronic skin dexterous hand multi-source energy supply tactile perception integrated structure system

Abstract

The application discloses a flexible electronic skin adaptive dexterous hand multi-source energy supply tactile perception integrated structure system, which comprises an integrated flexible carrier, a tactile perception array, a multi-source energy pickup unit, an auxiliary power supply conditioning unit, a deformation feedback unit, a cooperative power supply conditioning unit and a signal decoupling transmission unit. The tactile perception array generates a tactile data frame, the multi-source energy pickup unit generates energy input state data, the auxiliary power supply conditioning unit generates an energy state frame and a power supply disturbance frame, and the deformation feedback unit generates a deformation state frame. The cooperative power supply conditioning unit is connected with the region structure identifier, the tactile data frame, the energy state frame, the power supply disturbance frame and the deformation state frame under the same time reference. The system can realize the linkage operation of the structure conformation, auxiliary energy supply, tactile sampling and signal decoupling of the flexible electronic skin on the curved surface of the dexterous hand.

Description

Technical Field

[0001] This invention relates to the field of robot flexible sensing and self-powered electronic skin technology, specifically to a dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin. Background Technology

[0002] As the end effector of robots, dexterous hands typically require flexible electronic skin to acquire tactile information such as contact pressure, slippage, and force changes to achieve precise grasping, compliant manipulation, and contact status feedback. Existing flexible electronic skins often employ FPC carriers, flexible electrode arrays, piezoresistive sensing units, capacitive sensing units, or piezoelectric sensing units attached to the palm and finger surfaces of the robotic hand. Localized tactile detection is achieved through the intersection of horizontal and vertical electrodes or arrayed sensing units. While this approach can improve the tactile perception capabilities of dexterous hands to some extent, its structural layout is usually centered on tactile perception, lacking system-level collaborative design between power supply units, energy storage units, signal acquisition units, and deformation compensation units.

[0003] When flexible electronic skin is applied to the curved structure of a dexterous hand, the fingertips, sides, backs, and joint transition areas have different radii of curvature, contact probabilities, and bending stress distributions. Existing solutions typically deploy them using fixed electrode patterns or fixed sensor arrays, failing to correlate the parameters of the curved structure and contact conditions with subsequent data processing. When the dexterous hand performs grasping, pinching, slip suppression, or joint bending actions, the flexible substrate undergoes bending, torsion, and local stretching, leading to baseline drift in the sensing channel, changes in wire impedance, and mechanical coupling between adjacent functional areas, thus affecting the stability and consistency of tactile detection results. Simultaneously, to improve the endurance and energy utilization efficiency of flexible electronic skin, some solutions attempt to introduce energy pickup structures such as triboelectric, piezoelectric, photovoltaic, or thermoelectric elements into the flexible substrate. However, within the limited curved space of a dexterous hand, the tactile sensing array, multi-source energy pickup units, and flexible interconnects typically need to be arranged close together. Output ripple from multi-source energy pickup, power supply path switching disturbances, energy storage charging and discharging transients, and common reference potential fluctuations can easily couple to the tactile signal acquisition channel. Traditional static filtering or single-calibration methods are ill-suited to adapting to time-varying interference generated by flexible substrates under different bending postures, contact loads, and long-term aging conditions. Existing power supply methods often employ fixed output or passive compensation, failing to dynamically adjust the power supply path, sampling frequency, and time slot allocation based on tactile workload, regional deformation state, and energy availability. This can easily lead to insufficient power supply in high-load contact areas, energy waste in non-contact areas, and mutual interference between power supply disturbances and tactile sampling. Summary of the Invention

[0004] The purpose of this invention is to provide a dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin includes: An integrated flexible carrier is used to conformally fit the curved surface of a dexterous hand, and generates regional structure identifiers based on the curved structure parameters and contact condition parameters of different regions of the dexterous hand, so that the tactile sensing array and the multi-source energy pickup unit are integrated in the same flexible substrate according to the regional structure identifiers; The tactile sensing array is used to collect raw tactile signals from each tactile channel and generate tactile data frames carrying the regional structure identifier. The multi-source energy pickup unit is used to pick up electrical energy generated by at least two types of energy sources and generate energy input status data carrying the regional structure identifier; An auxiliary power supply conditioning unit is used to condition the electrical energy output by the multi-source energy pickup unit into auxiliary compensation power supply, and to generate an energy status frame and a power supply disturbance frame. The deformation feedback unit is used to acquire the deformation signal of the dexterous hand surface and generate a deformation state frame carrying the structural identifier of the region. The collaborative power supply control unit is used to match the tactile data frame, energy status frame, power supply disturbance frame and deformation status frame according to the regional structure identifier based on the same time reference, generate a regional load, energy and deformation association table, and generate a power supply control strategy and a tactile sampling control strategy according to the regional load, energy and deformation association table, so that the auxiliary compensation power supply parameters and tactile sampling parameters are adjusted in conjunction with each other according to the tactile workload and regional deformation status. The signal decoupling transmission unit is used to compensate and dynamically correct the power supply interference and deformation interference in the tactile data frame according to the power supply disturbance frame and the deformation state frame, and to feed back the corrected tactile data frame to the cooperative power supply control unit to update the power supply control strategy and the tactile sampling control strategy.

[0006] Specifically, the coordinated power supply control unit generates auxiliary power supply level, power supply path, tactile channel scanning sequence, tactile sampling frequency, and the allocation ratio of energy pickup time slot to tactile sampling time slot based on the regional load, energy, and deformation correlation table. When the predicted tactile load of any finger area increases, the power supply stability level and sampling priority of the tactile channel located in that finger area are increased before other finger areas. When the available energy value in the finger area is lower than the preset available energy threshold, the tactile channel in the contact area used to obtain contact intensity and contact area is continuously sampled, and the scanning frequency of the tactile channel in the non-contact area is reduced, or the energy pickup time slot is delayed.

[0007] Specifically, the regional structure identifier includes regional number, curvature level, contact probability level, bending stress level, and functional zoning type; The functional partition type includes at least a tactile main sensing area, an energy pickup area, a deformation reference area, and an isolation transition area, and the tactile data frame, energy status frame, power supply disturbance frame, and deformation status frame all carry a regional structure identifier for representing the data acquisition area.

[0008] Specifically, the main tactile sensing area is set on the fingertip, finger pad, or an area where the contact probability level reaches a preset high contact probability level; The energy pickup area is located on the side of the finger, the back of the finger, the outer bend of the joint, or in an area where the contact probability level is lower than the preset low contact probability level and the degree of contact wear is lower than the preset wear threshold. The deformation reference area is set in the joint transition area or in the area where the bending stress level reaches the preset stress level. The isolation transition zone is located between the main tactile sensing area and the energy pickup area to reduce tactile signal interference caused by power supply disturbances and mechanical deformations generated in the energy pickup area being transmitted to the main tactile sensing area.

[0009] Specifically, when the tactile sensing array generates tactile data frames, the tactile channels are grouped according to the region structure identifier, the initial baseline of each tactile channel in the non-contact state is read, tactile channel data with noise amplitude exceeding the preset noise range is removed, and the tactile channel data that has not been removed is normalized to extract contact intensity, contact area, number of activated channels, force change rate, force diffusion direction and sliding trigger frequency.

[0010] Specifically, when the auxiliary power supply conditioning unit generates the energy status frame and the power supply disturbance frame, it assigns a source identifier to each energy source to uniquely identify the energy source, collects the input voltage, input current and instantaneous power of each energy source, calculates the power fluctuation rate and input stability of each energy source, and generates the energy status frame and the power supply disturbance frame by combining the energy storage voltage, charging and discharging state, equivalent internal resistance, output voltage deviation, output ripple amplitude and power supply path switching state.

[0011] Specifically, when the deformation feedback unit generates a deformation state frame, it collects the bending angle, torsion angle, local stretching amount, and deformation rate of the region indicated by the regional structure identifier, and maps the collected bending angle, torsion angle, local stretching amount, and deformation rate of the region to the region indicated by the regional structure identifier. The deformation feedback unit is also used to extract the deformation baseline under the condition of no external contact, and to perform differential processing on the current deformation signal and the deformation baseline to obtain the deformation disturbance amount of the area indicated by the regional structure identifier.

[0012] Specifically, the signal decoupling transmission unit uses the output ripple, power supply path switching time, and energy storage charging and discharging state in the power supply disturbance frame as power supply interference features, and the bending angle, torsion angle, local stretching amount, and deformation change rate in the deformation state frame as deformation interference features. It then calculates the power supply interference weight of the power supply interference features on the tactile data frame and the deformation interference weight of the deformation interference features on the tactile data frame, respectively, to generate power supply noise compensation terms and deformation noise compensation terms.

[0013] Specifically, the collaborative power supply control unit divides the same control cycle into energy pickup time slots, tactile sampling time slots, and disturbance assessment time slots, and updates the allocation ratio of each time slot according to the tactile load prediction value, energy availability value, power supply interference weight, and deformation interference weight. When the power supply interference weight increases to above the preset power supply interference weight threshold, the overlap ratio between the energy pickup time slot and the tactile sampling time slot is reduced, or the tactile sampling time slot is adjusted so that the tactile sampling time avoids the power supply path switching time. When the deformation interference weight increases to above the preset deformation interference weight threshold, the data acquisition frequency of the deformation reference area is increased, and the tactile baseline of the area indicated by the regional structure identifier carried by the deformation state frame is corrected according to the deformation state frame.

[0014] Specifically, when generating the power supply control strategy and the tactile sampling control strategy, the collaborative power supply control unit performs the following processing steps: S1. Bind the tactile data frame, energy status frame, power supply disturbance frame and deformation status frame to the region according to the region structure identifier; S2. Based on the synchronization time tag, perform time alignment on various data frames with the same regional structure identifier; S3. Calculate the predicted tactile load value based on the tactile data frame, calculate the available energy value based on the energy state frame, and calculate the deformation interference value based on the deformation state frame. S4. Generate a regional load, energy, and deformation correlation table based on the tactile load prediction value, energy availability value, and deformation interference value; S5. Generate a power supply control strategy and a tactile sampling control strategy based on the aforementioned regional load, energy, and deformation correlation table. S6. Based on the source calculation results of power supply interference characteristics and deformation interference characteristics, perform closed-loop updates on the power supply control strategy and tactile sampling control strategy.

[0015] Specifically, the coordinated power supply control unit is also used to perform full lifecycle self-calibration processing; When the predicted tactile load is lower than the preset idle threshold and the available energy is higher than the preset calibration threshold, the deformation reference area and the tactile channel in a non-contact state are selected as calibration reference objects. Zero-load tactile data frames, no-load power supply disturbance frames and natural deformation state frames are collected, and the tactile baseline drift, power supply disturbance transmission coefficient, deformation to tactile coupling coefficient and energy storage equivalent internal resistance change are calculated. Based on the tactile baseline drift, power supply disturbance transmission coefficient, deformation-to-tactile coupling coefficient, and energy storage equivalent internal resistance change, update the generation parameters of the noise compensation term, the weighting coefficients in the regional load, energy, and deformation association table, and the allocation boundary between the energy pickup time slot and the tactile sampling time slot. This ensures that the power supply parameters output by the power supply control strategy, the sampling parameters output by the tactile sampling control strategy, and the compensation parameters used for dynamic correction of tactile data are updated synchronously after the flexible substrate undergoes creep, the electrodes age, or the energy storage performance degrades.

[0016] Compared with existing technologies, the advantages of this invention are as follows: This invention integrates the tactile sensing array and multi-source energy pickup unit into the same flexible substrate according to regional structural identifiers through an integrated flexible carrier, enabling tactile deployment, energy pickup, deformation reference, and isolation transition to match the curved surface structure and contact conditions of the dexterous hand; by associating tactile data frames, energy status frames, power supply disturbance frames, and deformation status frames under the same time reference, a correspondence between regional load, energy, and deformation is established, allowing the auxiliary compensation power supply and tactile sampling strategy to be dynamically adjusted according to the tactile workload and regional deformation state; and by decoupling the signal transmission... The transmission unit performs source compensation and dynamic correction for power supply interference and deformation interference, reducing the impact of power supply disturbances, mechanical deformation, and baseline drift on tactile data. By updating the power supply control strategy, tactile sampling control strategy, and time slot allocation ratio through interference source feedback, the system can prioritize continuous sampling and power supply stability of the core contact area under limited energy storage conditions. It also maintains the matching relationship between tactile detection, auxiliary power supply, and dynamic correction after flexible substrate creep, electrode aging, or energy storage performance degradation, thereby improving the surface adaptability, signal stability, energy utilization efficiency, and long-term operational reliability of the dexterous hand flexible electronic skin. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the overall structure of the dexterous hand multi-source power supply tactile sensing integrated structure system of the present invention.

[0019] Figure 2 This is a schematic diagram of the functional partitions of the integrated flexible carrier of the present invention.

[0020] Figure 3 This is a schematic diagram of the closed-loop control process of power supply regulation, tactile sampling and signal decoupling in this invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figures 1-3 As shown, the present invention provides a technical solution: a dexterous hand multi-source powered tactile sensing integrated structural system adapted to flexible electronic skin, comprising: An integrated flexible carrier is used to conformally fit the curved surface of a dexterous hand, and generates regional structure identifiers based on the curved structure parameters and contact condition parameters of different regions of the dexterous hand, so that the tactile sensing array and the multi-source energy pickup unit are integrated in the same flexible substrate according to the regional structure identifiers; The tactile sensing array is used to collect raw tactile signals from each tactile channel and generate tactile data frames carrying the regional structure identifier. The multi-source energy pickup unit is used to pick up electrical energy generated by at least two types of energy sources and generate energy input status data carrying the regional structure identifier; An auxiliary power supply conditioning unit is used to condition the electrical energy output by the multi-source energy pickup unit into auxiliary compensation power supply, and to generate an energy status frame and a power supply disturbance frame. The deformation feedback unit is used to acquire the deformation signal of the dexterous hand surface and generate a deformation state frame carrying the structural identifier of the region. The collaborative power supply control unit is used to match the tactile data frame, energy status frame, power supply disturbance frame and deformation status frame according to the regional structure identifier based on the same time reference, generate a regional load, energy and deformation association table, and generate a power supply control strategy and a tactile sampling control strategy according to the regional load, energy and deformation association table, so that the auxiliary compensation power supply parameters and tactile sampling parameters are adjusted in conjunction with each other according to the tactile workload and regional deformation status. The signal decoupling transmission unit is used to compensate and dynamically correct the power supply interference and deformation interference in the tactile data frame according to the power supply disturbance frame and the deformation state frame, and to feed back the corrected tactile data frame to the cooperative power supply control unit to update the power supply control strategy and the tactile sampling control strategy.

[0023] In the above implementation, the integrated flexible carrier is first physically divided into the fingertips, finger pads, knuckle sides, knuckle bending areas, and palmar contact areas of the dexterous hand. Each physical division corresponds to a region structure identifier. When generating the region structure identifier, the curvature radius, bending angle, surface arc length, attachment width, allowable tensile strain, upper limit of rated contact pressure, number of contacts per unit time, and whether there is a slip contact state are first collected for that division. The curvature level is determined by the curvature radius; specifically, the smaller the curvature radius, the higher the degree of bending in the region, and the higher the corresponding curvature level. The contact strength level is jointly determined by the upper limit of rated contact pressure, the number of contacts per unit time, and the slip contact state. Specifically, the upper limit of rated contact pressure is first compared with the preset rated pressure of the dexterous hand to obtain the pressure ratio, then the number of contacts per unit time is compared with the rated contact frequency to obtain the contact frequency ratio, and then the slip contact state is converted into a contact additional item. The pressure ratio, contact frequency ratio, and contact additional item are multiplied by their corresponding weights and then added together to obtain the contact strength value of the region. The aforementioned weights are determined during the assembly and calibration phase. During calibration, three sets of actions—static pressing, periodic grasping, and sliding touch—are executed sequentially. The contribution ratios of pressure changes, contact frequency changes, and sliding state changes to the tactile channel output changes are calculated, and then these three contribution ratios are converted into weights that sum to 1. The area structure identifier is formed by sequentially combining hand position number, finger number, joint number, area sequence number, curvature level, contact intensity level, and channel group number, ensuring that each area structure identifier corresponds to a unique attachment position, surface shape, contact condition, and tactile channel group.

[0024] The tactile sensing array is integrated within the same flexible substrate according to the regional structure identifier.

[0025] Several tactile channels are set up within each region, and each tactile channel is bound according to the structural identifier of its region and the channel sequence number. After the tactile channel outputs the original tactile signal within the sampling period, the tactile sensing array encapsulates the original tactile signal, sampling time, region structural identifier, channel sequence number, and current scanning sequence into a tactile data frame. To ensure that the tactile data frame reflects the actual contact state, baseline subtraction processing is performed before the tactile data frame is generated. The specific process of baseline subtraction processing is as follows: when the dexterous hand is not in contact with an external object and remains in a naturally extended state, the output values ​​of the tactile channel are continuously collected for 60 seconds, and the average value of all collected values ​​of the channel is taken as the non-contact baseline of the channel; during actual sampling, the current tactile channel output value is subtracted from the non-contact baseline to obtain the baseline-subtracted tactile signal.

[0026] The contact determination threshold is obtained from the non-contact calibration phase. Specifically, the average value of the channel's output value during the non-contact calibration phase is first calculated, then the dispersion of the channel's output value relative to the average value is calculated, and the sum of the average value and three times the dispersion is used as the contact determination threshold for that channel. When the tactile signal after baseline subtraction continuously reaches or exceeds the contact determination threshold, the tactile channel is determined to be a contact channel; tactile channels that do not reach the contact determination threshold are determined to be non-contact channels. Through the above processing, the tactile data frame not only contains the original tactile signal, but also its location, channel position, and the source of the contact state.

[0027] The multi-source energy pickup units are arranged according to the regional structure identifiers and the corresponding tactile sensing arrays, with each region configured with at least two types of energy pickup structures.

[0028] In this embodiment, the multi-source energy pickup unit picks up mechanical deformation energy generated by bending or pressing with a dexterous hand, triboelectric energy generated by contact friction, and can also pick up ambient light energy or local temperature difference energy when a corresponding functional layer is provided. After each type of energy source outputs, its output voltage, output current, duration, and energy conversion efficiency are first detected. Then, the effective energy that the energy source can provide in the current control cycle is calculated based on the product relationship of voltage, current, duration, and energy conversion efficiency. For areas with energy storage capacitors, the energy storage terminal voltage is also detected, and the releasable stored energy is calculated based on the energy storage capacitor capacity, the current energy storage terminal voltage, and the minimum allowable operating voltage. During the calculation, the energy storage level corresponding to the current energy storage terminal voltage and the minimum allowable operating voltage are first obtained, and then the difference between the two is taken as the base value of the releasable stored energy. The usable energy value of this area is determined by the releasable stored energy, the effective energy that various energy sources can provide in the current control cycle, and the reserved energy. Specifically, the releasable stored energy is added to various types of effective energy, and then the reserved energy required to ensure the continuous operation of the control circuit and communication link is subtracted. The reserved energy is determined during the factory calibration phase. During calibration, the core control circuit and communication link are continuously run for 10 control cycles, and the actual energy consumed in each cycle is recorded. The maximum energy consumed is taken as the reserved energy. The multi-source energy pickup unit generates energy input status data by generating the output status, effective energy, energy source type, acquisition time, and regional structure identifier for each type of energy source.

[0029] After receiving electrical energy from the multi-source energy pickup unit, the auxiliary power supply conditioning unit sequentially performs rectification, current limiting, energy storage, voltage boosting or bucking, and voltage stabilization to bring the output voltage up to the reference voltage required by the tactile sensing array. Within each control cycle, the auxiliary power supply conditioning unit detects the deviation between the output voltage and the reference voltage, and also detects the fluctuation amplitude of the output voltage within that control cycle. The deviation between the output voltage and the reference voltage characterizes the degree of power supply offset, and the ratio of the output voltage fluctuation amplitude to the reference voltage characterizes the degree of power supply ripple. The auxiliary power supply conditioning unit generates an energy status frame based on the available energy value, output voltage, output current, current power supply level, and current power supply path, and generates a power supply disturbance frame based on the degree of power supply offset, the degree of power supply ripple, the power supply path switching time, and transient current changes. The power supply stability level is determined jointly based on the degree of power supply deviation and the degree of power supply ripple: when the degree of power supply deviation is no greater than 5% and the degree of power supply ripple is no greater than 3%, it is determined to be a low-stability power supply level; when the degree of power supply deviation is no greater than 3% and the degree of power supply ripple is no greater than 1.5%, it is determined to be a medium-stability power supply level; and when the degree of power supply deviation is no greater than 1% and the degree of power supply ripple is no greater than 0.8%, it is determined to be a high-stability power supply level. Areas with increased tactile load are prioritized for adjustment to a high-stability power supply level, while non-contact areas maintain a low-stability power supply level, thereby concentrating auxiliary compensation power supply on serving areas where there is a current demand for tactile information.

[0030] The deformation feedback unit is located in the bending concentration area and the dense tactile channel area of ​​the flexible integrated carrier to collect the bending, stretching, compression, and torsion states of the flexible integrated carrier generated by the movements of the dexterous hand. During the assembly and calibration phase, the deformation feedback unit first obtains the deformation baseline and deformation sensitivity. Specifically, the flexible integrated carrier is sequentially placed in a state without deformation, a state with half the rated maximum allowable deformation, and a state with the rated maximum allowable deformation, and the output value of the deformation feedback unit is recorded for each state. The deformation sensitivity is determined by the correspondence between the change in deformation and the change in output value, and the output value in the state without deformation is used as the deformation baseline.

[0031] In actual operation, the current deformation feedback unit output value is compared with the deformation-free baseline, and converted into the current strain value based on the deformation sensitivity. Then, the current strain value is compared with the strain value of the previous control cycle, and the deformation rate is obtained by combining the control cycle duration. The deformation state frame includes the current strain value, deformation rate, deformation direction, sampling time, and region structure identifier. The deformation interference threshold is determined through no-load bending calibration. Specifically, the dexterous hand performs multiple bending actions without contacting an external object, collects the deformation feedback value of that region, calculates its average level and dispersion, and uses the sum of the average level and three times the dispersion as the deformation interference threshold for that region. When the current strain value reaches or exceeds this deformation interference threshold, the tactile data for that region enters the deformation compensation processing flow.

[0032] The coordinated power supply control unit receives area structure identifiers, tactile data frames, energy status frames, power supply disturbance frames, and deformation status frames using a unified time reference, and generates an area load, energy, and deformation correlation table within each control cycle. This correlation table uses the area structure identifier as an index, with each row recording the number of contact channels, the number of core tactile channels, average contact pressure, contact pressure change rate, available energy, power supply offset, power supply ripple, current strain value, and deformation change rate for each area. The number of contact channels is determined by the number of tactile channels that reach the contact judgment threshold; the core tactile channels are determined by the channels with higher tactile signal amplitudes among the contact channels. Specifically, the contact channels are first sorted from largest to smallest according to the tactile signal after baseline subtraction, and then the top 30% are selected as core tactile channels; when the number of contact channels is less than 3, all contact channels are considered core tactile channels.

[0033] It should be noted that the average contact pressure is calculated by subtracting the baseline of all contact channels in the area from the tactile signal. The rate of change of contact pressure is determined by the difference between the average contact pressure of the current control cycle and the average contact pressure of the previous control cycle, combined with the duration of the control cycle.

[0034] The predicted tactile load is obtained by combining the proportion of contact channels, the proportion of average contact pressure, the proportion of contact pressure change rate, and the proportion of deformation change rate.

[0035] In the specific calculation, the number of contact channels is first compared with the total number of tactile channels in the area to obtain the contact channel ratio; then, the average contact pressure is compared with the rated contact pressure to obtain the average contact pressure ratio; next, the contact pressure change rate is compared with the rated pressure change rate to obtain the pressure change rate ratio; and finally, the deformation change rate is compared with the rated deformation change rate to obtain the deformation change rate ratio. These four ratios are multiplied by their corresponding load weights and then summed to obtain the predicted tactile load value for the current control cycle in the area. The load weights are determined during the calibration phase using four actions: standard gripping, sliding recognition, light touch positioning, and no-load bending. Specifically, the contribution of the contact channel ratio, average contact pressure ratio, pressure change rate ratio, and deformation change rate ratio to the increase in tactile recognition error is calculated for each of the four actions, and each contribution is converted into a weight with a sum of 1. The tactile load increase threshold is determined by the rated grasping action. Specifically, the dexterous hand runs continuously for 100 control cycles according to the rated grasping action. The predicted tactile load value for each control cycle is calculated, and then the average level and dispersion of the 100 predicted tactile load values ​​are obtained. The sum of the average level and twice the dispersion is used as the tactile load increase threshold. When the predicted tactile load value of a certain area reaches or exceeds the tactile load increase threshold for three consecutive control cycles, the coordinated power supply control unit determines that the predicted tactile load value of that finger area has increased. The determination condition of three consecutive control cycles is used to filter out single spike disturbances and avoid triggering power supply level and scanning sequence switching due to occasional noise.

[0036] The determination of insufficient energy availability is based on the energy required to maintain continuous sampling of the core haptic channel within the preceding control window. The collaborative power supply control unit first determines the number of core haptic channels in the region, then reads the average power consumption of a single core haptic channel at the target sampling frequency, and then multiplies the average power consumption of a single core haptic channel, the number of core haptic channels, and the control window duration to obtain the energy required for core haptic channel sampling. The energy required for the control circuit and transmission circuit in the same control window in this region is then added to this, yielding the energy required to maintain continuous sampling of the core haptic channel. The average power consumption of a single core haptic channel, the power consumption of the control circuit, and the power consumption of the transmission circuit are all determined during factory testing through current integration at rated voltage. The energy insufficiency threshold is set at 1.2 times the above-mentioned required energy, with the additional 20% serving as a margin for power supply path switching losses, transient voltage drops at the energy storage end, and energy pickup fluctuations. When the energy availability in the region is lower than the energy insufficiency threshold, the collaborative power supply control unit determines that the energy availability in the corresponding region is insufficient.

[0037] When generating power supply control strategies, the collaborative power supply control unit first arranges the regions from high to low according to the predicted tactile load value, and then increases the power supply stability level for regions where the predicted tactile load value reaches or exceeds the tactile load increase threshold. For high-load regions where the available energy value is not lower than the insufficient energy threshold, the auxiliary power supply level is switched to the high-stability sampling level, and the power supply path is switched to the power supply path jointly output by the energy storage end and the real-time energy pickup end; for high-load regions where the available energy value is lower than the insufficient energy threshold, the power supply of the branch where the core tactile channel is located is retained, the auxiliary power supply branch of the non-contact tactile channel is shut down, and the power supply level of the non-contact region is reduced to the low-power maintenance level. Before the power supply path switching, the auxiliary power supply conditioning unit detects the difference between the target power supply path voltage and the current output voltage; when the ratio of this difference to the reference voltage is not greater than 3%, the path switching is performed; when the ratio is greater than 3%, the target power supply path is first current-limited charged to make the target power supply path voltage close to the current output voltage before the switching is performed, thereby reducing the electrical disturbance to the tactile data frame caused by the power supply path switching instant.

[0038] When the collaborative power supply control unit generates a tactile sampling control strategy, it determines the tactile channel scanning order according to the regional sampling priority. The regional sampling priority is jointly determined by the predicted tactile load, the proportion of core tactile channels, the degree of power supply offset, and the degree of power supply ripple. During calculation, the predicted tactile load and the proportion of core tactile channels are used as factors to increase the sampling priority, while the degree of power supply offset and the degree of power supply ripple are used as factors to decrease the sampling priority. Each factor is multiplied by its corresponding priority weight, and then added or subtracted to obtain the regional sampling priority. The priority weights are obtained through tactile recognition accuracy calibration under a rated grasping task. Specifically, during calibration, the predicted tactile load, the proportion of core tactile channels, the degree of power supply offset, and the degree of power supply ripple are changed respectively, the decrease in tactile recognition accuracy is recorded, and each decrease is converted into a weight with a sum of 1. The scanning order is arranged from high to low according to the regional sampling priority. The tactile sampling frequency is increased or decreased based on the predicted tactile load; the higher the predicted tactile load, the closer the tactile sampling frequency is to the preset maximum sampling frequency. The tactile sampling frequency is also limited by the power supply stability level. Under a low power supply stability level, the tactile sampling frequency does not exceed 50% of the preset maximum sampling frequency; under a medium power supply stability level, the tactile sampling frequency does not exceed 80% of the preset maximum sampling frequency; and under a high power supply stability level, the tactile sampling frequency reaches the preset maximum sampling frequency. For the core tactile channel in the contact area, the tactile sampling frequency is always no lower than the minimum continuous sampling frequency of the core tactile channel. This minimum frequency is determined by the dexterous hand tactile closed-loop control cycle, which is determined by obtaining at least two core tactile sampling results within one tactile closed-loop control cycle.

[0039] Energy pickup time slots and tactile sampling time slots are allocated according to a control window. Each control window is divided into several time slots of equal length. The coordinated power supply control unit first sets the rated sampling ratio, then increases the sampling time slot ratio based on the predicted tactile load value, and decreases the sampling time slot ratio based on the degree of energy insufficiency. The degree of energy insufficiency is determined by the difference between the energy insufficiency threshold and the available energy value; the larger the difference, the higher the degree of energy insufficiency. The sampling time slot ratio is limited between a preset minimum sampling ratio and a preset maximum sampling ratio. The preset minimum sampling ratio is used to ensure continuous sampling of the core tactile channel, and the preset maximum sampling ratio is used to reserve necessary energy pickup time. The energy pickup time slot ratio is the remaining time slot ratio of the control window. When the predicted tactile load of any finger area reaches or exceeds the tactile load increase threshold, and the available energy value of that area is lower than the energy shortage threshold, the coordinated power supply control unit first locks the sampling time slot of the core tactile channel so that the time interval between two adjacent core tactile samples does not exceed the maximum interval corresponding to the lowest continuous sampling frequency of the core tactile channel; then the scanning frequency of the non-contact area tactile channel is reduced to 30% to 50% of the original scanning frequency; for energy pickup actions that may introduce power supply disturbances, they are arranged to be executed outside the core tactile sampling interval to avoid the transient disturbances generated by energy pickup being superimposed on the core tactile data frame.

[0040] When the signal decoupling transmission unit performs source compensation and dynamic correction on the tactile data frame, it first reads the power supply offset degree, power supply ripple degree, power supply path switching time, and transient current change in the power supply disturbance frame, and then reads the current strain value, deformation rate, and deformation direction in the deformation state frame. For each tactile channel, the signal decoupling transmission unit first calculates the tactile signal deviation caused by the power supply offset based on the power supply offset degree, then calculates the tactile signal deviation caused by the ripple based on the power supply ripple degree, then calculates the tactile signal deviation caused by the static deformation based on the current strain value, and finally calculates the tactile signal deviation caused by the dynamic deformation based on the deformation rate.

[0041] The four types of deviations mentioned above are determined using calibration coefficients. These coefficients are obtained during the calibration phase, specifically by sequentially performing voltage step tests, ripple injection tests, static bending tests, and dynamic bending tests. Each test changes only one type of interference factor, recording the proportional relationship between the change in the tactile channel output and the corresponding change in the interference factor. This proportional relationship is then used as the corresponding calibration coefficient. The signal decoupling transmission unit sequentially subtracts the power supply offset deviation, ripple deviation, static deformation deviation, and dynamic deformation deviation from the original tactile signal to obtain the corrected tactile data. After correction, the signal decoupling transmission unit uses the median output of adjacent similar tactile channels within the same area as a reference value to calculate the residual between the corrected tactile data and this reference value. The residual threshold is determined through compensation calibration under rated contact conditions. Specifically, after compensation calibration, the area is placed in rated contact condition, multiple sets of corrected residuals are continuously collected, the average residual level and residual dispersion are calculated, and the sum of the average residual level and three times the residual dispersion is used as the residual threshold. When the correction residual of a certain tactile channel reaches or exceeds the residual threshold for five consecutive control cycles, the signal decoupling transmission unit feeds back the correction deviation information of that channel to the collaborative power supply control unit. After receiving the correction deviation information, the collaborative power supply control unit reduces the sampling priority of non-core tactile channels in that area, and re-determines the auxiliary power supply level, power supply path, tactile channel scanning sequence, tactile sampling frequency, and the allocation ratio of energy pickup time slots to tactile sampling time slots, so that subsequent tactile sampling and auxiliary compensation power supply are updated according to the current disturbance state.

[0042] In one possible implementation, the regional structure identifier includes a region number, curvature level, contact probability level, bending stress level, and functional partition type; the functional partition type includes at least a tactile main sensing area, an energy pickup area, a deformation reference area, and an isolation transition area, and the tactile data frame, energy status frame, power supply disturbance frame, and deformation status frame all carry the corresponding regional structure identifier.

[0043] In the above embodiments, the regional structure identifier is generated before the integrated flexible carrier is attached to the dexterous hand, and is written into the collaborative power supply control unit through a calibration program after assembly.

[0044] Region numbering is used to indicate the actual attachment position of the flexible integrated carrier on the dexterous hand. The generation process includes: first, determining the primary position according to the thumb, index finger, middle finger, ring finger, little finger, and palmar surface; then, determining the secondary position according to the fingertip, fingertip, proximal phalanx, middle phalanx, distal phalanx, joint bending area, and lateral contact edge; finally, determining the tertiary number according to the arrangement order of the region in the flexible substrate unfolding plane. Through the combination of these three levels of numbering, each region obtains a unique region number, ensuring that no two regions have the same number on the same dexterous hand.

[0045] The curvature level is determined based on the degree of surface curvature in the corresponding area. To determine the curvature level, the radius of curvature along the length and width of the finger is first obtained using 3D contour scanning, structured light measurement, or assembly fixture measurement. The direction with the greater curvature is then taken as the curvature evaluation direction for that area. Next, the radius of curvature of this area is compared with the curvature grading threshold in the dexterous hand electronic skin installation reference library. The smaller the radius of curvature, the more severe the surface bending, and the higher the curvature level. In this embodiment, the curvature level is divided into 5 levels. The curvature grading threshold is determined during the assembly calibration stage. Specifically, the radius of curvature of all attachment areas of the dexterous hand is collected, sorted from largest to smallest, and the sorted results are divided into 5 continuous intervals. The boundary values ​​of adjacent intervals are used as the curvature grading thresholds. For the fingertip and joint bending areas, since local bending is more pronounced during attachment, the smaller value between the measured radius of curvature and the radius of curvature under the maximum working bending posture is used as the criterion for determining the curvature level.

[0046] The contact probability level indicates the frequency of effective contact in a corresponding area during typical grasping tasks. To determine the contact probability level, the dexterous hand is first subjected to the following tasks sequentially: grasping cylindrical objects, grasping spherical objects, pressing flat surfaces, edge gripping, and sliding detection. Each task is repeated at least 20 times. In each task, when the signal after baseline subtraction of at least one tactile channel within the area reaches the contact determination threshold and the duration exceeds one tactile closed-loop control cycle, it is recorded as one effective contact in that area. After completing all calibration tasks, the number of effective contacts in that area is compared with the total number of calibration tasks participated in by that area to obtain the contact probability of that area. The contact probability level is divided into 5 levels: areas with a contact probability of 80% or higher are classified as Level 5; areas with a contact probability of 60% or higher but lower than 80% are classified as Level 4; areas with a contact probability of 40% or higher but lower than 60% are classified as Level 3; areas with a contact probability of 20% or higher but lower than 40% are classified as Level 2; and areas with a contact probability lower than 20% are classified as Level 1. The above levels are used in subsequent operations to determine the default scanning priority of the tactile main sensory area and the frequency reduction order of non-contact areas when energy is insufficient.

[0047] Bending stress level is used to indicate the magnitude of bending stress experienced by a corresponding region during dexterity hand movement. To determine the bending stress level, a deformation feedback unit is first installed on the flexible integrated carrier, or a temporary strain gauge is attached to the same region during the calibration phase. Then, the dexterity hand is controlled to move from its natural extended posture to its maximum permissible bending posture, and the maximum strain value of that region is recorded during the movement. For flexible substrates with a known elastic modulus, the bending stress evaluation value of the region is obtained based on the correspondence between the material's elastic modulus and the measured strain. For multi-layered composite flexible substrates, equivalent elastic parameters are first determined according to the thickness and elastic modulus of each layer, and then the bending stress evaluation value is obtained by combining this with the measured strain. The bending stress level is divided into 5 levels, with the grading threshold jointly determined by the material's permissible stress and the measured stress distribution. Specifically, the permissible stress of the flexible substrate material is first set as the highest level boundary. Then, the bending stress evaluation values ​​measured in each region during the assembly calibration process are sorted from smallest to largest and divided into 5 continuous intervals based on the highest level boundary. Regions in the highest interval are determined as the 5th bending stress level, and regions in the lowest interval are determined as the 1st bending stress level. The higher the bending stress level, the more priority should be given to arranging deformation reference areas or isolation transition areas in this region to avoid high-stress locations directly bearing high-density tactile channels and fragile interconnect lines.

[0048] Specifically, the functional zoning type is determined based on the area number, curvature level, contact probability level, and bending stress level. Areas with a high contact probability level and a bending stress level not exceeding level 3 are designated as the primary tactile sensing area. High-density tactile channels are prioritized in this area, and a higher default scanning priority is configured in the sampling strategy. Areas with a low contact probability level, a curvature level not exceeding level 3, and stable deformation or friction input conditions are designated as energy pickup areas. Piezoelectric, triboelectric, photovoltaic, or thermoelectric energy pickup structures are prioritized in this area, and their outputs are connected to the auxiliary power supply conditioning unit. Areas with stable bending stress levels, low contact probability levels, and minimal external contact interference are designated as deformation reference areas. These areas provide reference signals for bending, stretching, or torsional states and participate in deformation compensation of tactile data. The area located between the main tactile sensing area and the energy pickup area, or between the high bending stress area and the high-density tactile channel, is defined as the isolation transition area. This area is equipped with a flexible buffer layer, shielded wiring, low-stiffness transition structure, or blank buffer interval to reduce the impact of energy pickup disturbance, deformation concentration, and interconnection stress on the main tactile sensing area.

[0049] The tactile data frame, energy status frame, power supply disturbance frame, and deformation status frame all carry corresponding area structure identifiers. When generating a tactile data frame, the original tactile signal, the tactile signal after baseline subtraction, the sampling time, and the channel number are written into the frame header field along with the area number, curvature level, contact probability level, bending stress level, and functional zone type of the area. This allows the collaborative power supply control unit to determine whether the tactile data originates from the main tactile sensing area, the isolation transition area, or other functional areas while reading the tactile data. When generating an energy status frame, the energy source type, current available energy value, energy storage terminal status, power supply level, and power supply path are written into the frame header field along with the area structure identifier corresponding to the energy pickup area. This allows the energy status output by the auxiliary power supply conditioning unit to be traced back to the specific energy pickup area. When generating a power supply disturbance frame, the power supply offset degree, power supply ripple degree, path switching time, and transient current change are bound to the area structure identifier that generates or is affected by the disturbance. This allows the signal decoupling transmission unit to distinguish whether the power supply disturbance originates from the energy pickup area, the power supply path switching area, or the isolation transition area. When the deformation state frame is generated, the current strain value, deformation rate, and deformation direction are bound to the deformation reference area or the regional structure identifier of the strain area, so that the deformation compensation can be matched according to the actual surface position.

[0050] Specifically, by carrying the same format of regional structure identifiers in various data frames, the collaborative power supply control unit can map tactile data, energy status, power supply disturbances, and deformation status to the same region or adjacent functional regions under the same time reference. For the tactile main sensing area, when its contact probability level is high and the tactile load prediction value increases, the collaborative power supply control unit increases the sampling priority and power supply stability level of this area. For the energy pickup area, when the ripple level displayed in its power supply disturbance frame reaches the power supply disturbance threshold, the signal decoupling transmission unit applies this disturbance information to the tactile data correction of the adjacent tactile main sensing area. For the deformation reference area, when the current strain value displayed in its deformation status frame reaches the deformation interference threshold, the collaborative power supply control unit reduces the scanning frequency of non-core channels in high bending stress areas and prioritizes the sampling of core channels in the tactile main sensing area. The power supply disturbance threshold is determined during the calibration stage of the auxiliary power supply conditioning unit. Specifically, under the condition that the tactile channel has no external contact, the energy pickup output fluctuation is increased step by step, the ripple level corresponding to the measurable offset of the tactile channel is recorded, and 80% of this ripple level is used as the power supply disturbance threshold. The deformation interference threshold is determined by no-load bending calibration. Specifically, the dexterous hand performs multiple bending actions without external contact, records the average level and dispersion of the output of the deformation reference area, and uses the sum of the average level and three times the dispersion as the deformation interference threshold.

[0051] After adopting the above processing method, the regional structure identifier is no longer used only as a regional number, but also carries information on surface morphology, contact probability, bending stress, and functional zoning. With the same regional structure identifier, tactile data frames, energy status frames, power supply disturbance frames, and deformation status frames can correspond to specific regions, avoiding confusion in regional attribution between data from different sources. Since the tactile main sensing area, energy pickup area, deformation reference area, and isolation transition area are clearly distinguished at both the structural and data layers, the tactile sensing array can be preferentially arranged in areas with high contact probability and low to moderate bending stress. The energy pickup structure can avoid high-precision tactile main sensing positions, the deformation reference signal can be obtained from areas with less contact interference, and the isolation transition area can reduce the impact of high-stress areas and energy pickup disturbances on tactile data. This improves the adhesion reliability of the flexible electronic skin on the complex curved surfaces of the dexterous hand, the traceability of tactile data, and the accuracy of multi-source power supply control.

[0052] In another embodiment, the curvature level, contact probability level, and bending stress level in the area structure identifier are divided into 3, 4, or 6 levels. The number of levels is determined based on the density of the dexterous hand's tactile channels, the controller's computing power, and the size of the flexible substrate. Increasing the number of levels improves the accuracy of area control, but also increases the amount of calibration data. Decreasing the number of levels simplifies the calibration process and is suitable for dexterous hands with fewer tactile channels. Functional partition types also include communication wiring areas, encapsulation reinforcement areas, or maintenance connection areas. Communication wiring areas are used to arrange flexible interconnect lines, encapsulation reinforcement areas are used to cover easily worn areas, and maintenance connection areas are used to set up detachable interfaces or test terminals. The addition of the above functional partition types does not change the basic way in which the area structure identifier carries the area number, curvature level, contact probability level, and bending stress level, nor does it change the processing procedure in which tactile data frames, energy status frames, power supply disturbance frames, and deformation status frames all carry the corresponding area structure identifier.

[0053] In one possible implementation, the main tactile sensing area is located on the fingertip, fingertip, or a high-contact-probability area; the energy pickup area is located on the finger side, finger back, joint outer bending area, or low-contact-wear area; the deformation reference area is located in the joint transition area or bending stress concentration area; and the isolation transition area is located between the main tactile sensing area and the energy pickup area to reduce the coupling effect of power supply disturbances and mechanical deformation on the tactile signal.

[0054] In the above implementation, when the contact probability is not less than 60%, the area is identified as a high contact probability area and is preferentially included in the tactile main sensing area. For the fingertips and finger pads, even if their local contact probability is slightly less than 60%, as long as they are located at the end tactile feedback position required for dexterous hand closed-loop grasping, they are still treated as the tactile main sensing area to ensure that key tactile signals can be continuously obtained during grasping, pressing, clamping, and sliding recognition.

[0055] A high-density tactile channel is set within the primary tactile sensing area, with the sensitive surface of the tactile channel facing the external contact surface of the dexterous hand. The boundary of the primary tactile sensing area does not extend directly to the maximum bending line of the joint; an isolation distance is maintained between them. This isolation distance is jointly determined by the maximum allowable strain of the flexible substrate and the distortion threshold of the tactile channel. To determine this isolation distance, the distance between the tactile channel and the joint bending line is gradually reduced under bending calibration conditions, and the output drift of the tactile channel in the absence of external contact is recorded. When the output drift reaches 30% of the contact determination threshold, this location is determined as an unsuitable location for placing the primary tactile sensing channel, and it is moved outward at least one tactile channel spacing away from the bending line, serving as the boundary of the primary tactile sensing area. This ensures that the primary tactile sensing area maintains contact sensitivity while avoiding output drift in the absence of contact due to bending concentration.

[0056] The finger sides and backs of the fingers do not directly bear the main contact pressure in most grasping tasks, making them suitable for triboelectric, piezoelectric, photovoltaic, or thermoelectric energy pickup structures. The external bending area of ​​the joint produces significant periodic deformation during finger bending, making it suitable for piezoelectric or bending deformation energy pickup structures. Low-contact-wear areas have fewer effective contact cycles and lower surface wear risk during calibration tasks, making them suitable for energy pickup electrodes, flexible energy storage connections, and energy input detection nodes. The criteria for determining low-contact-wear areas include contact probability and wear calibration results. During wear calibration, the dexterous hand repeatedly performs at least 1000 grasping and releasing actions with rated grasping force, detecting impedance changes and surface morphology changes in the flexible surface functional layer. Areas where the impedance change does not exceed 10% of the initial impedance and where no continuous cracks appear in the surface functional layer are identified as low-contact-wear areas. Separating the energy pickup area from the main tactile sensing area reduces the impact of voltage fluctuations, triboelectric charge accumulation, and local interlayer stiffness changes of the energy pickup structure on the output signal of the main tactile sensing area.

[0057] The joint transition region is the connection area between the rigid support segment and the flexible bending segment of the joint. This region reflects the continuous deformation changes caused by the joint movement when the finger bends. The bending stress concentration region is the area with higher strain under the maximum allowable bending posture. This region can provide the reference signal required for deformation compensation. The process of determining the deformation reference region is as follows: first, control the dexterous hand to move from the natural extension state to the maximum allowable bending state, and then return to the natural extension state, repeating this at least 10 times; during the above process, collect the deformation feedback signals of each candidate region, and screen out the region with stable deformation feedback amplitude, repeatability deviation not exceeding 5%, and minimal influence from external contact pressure as the deformation reference region. The repeatability deviation is determined by the ratio between the maximum deviation and the average value of the deformation feedback signal under the same posture in multiple bending actions. The deformation reference region is not used as the main tactile contact area. Its output is primarily used to determine the bending amplitude, bending direction, and deformation rate of the flexible integrated carrier, and serves as the basis for deformation compensation by the signal decoupling transmission unit.

[0058] In this embodiment, an isolation transition zone is located between the main tactile sensing area and the energy pickup area. The isolation transition zone is continuously arranged along the boundary between the two, and its width is determined based on the power supply disturbance propagation distance and the mechanical deformation influence distance. The power supply disturbance propagation distance is obtained through pulse output calibration of the energy pickup area. Specifically, with no external contact in the main tactile sensing area, the energy pickup area operates at rated output and peak output respectively, and the output offset of adjacent tactile channels is recorded. When the output offset of a tactile channel drops below 20% of the contact determination threshold, the distance from the energy pickup area to that channel is taken as the power supply disturbance propagation distance. The mechanical deformation influence distance is obtained through bending calibration. Specifically, the maximum permissible bending action is performed in the joint's outer bending area, and the contactless drift of adjacent tactile channels is recorded. When the contactless drift drops below 20% of the contact determination threshold, the distance from the bending concentration point to that channel is taken as the mechanical deformation influence distance. The width of the isolation transition zone is not less than the larger of the power supply disturbance propagation distance and the mechanical deformation influence distance.

[0059] Specifically, the isolation transition zone incorporates at least one of the following structures: a thickened flexible insulating layer, a low-modulus buffer layer, a grounding shielding trace, a floating isolation electrode, or a blank buffer interval. The thickened flexible insulating layer increases the insulation distance between the output of the energy pickup area and the signal line of the main tactile sensing area, reducing charge coupling. The low-modulus buffer layer disperses the tensile strain transmitted from the joint's external bending area to the main tactile sensing area, reducing baseline drift caused by mechanical deformation. The grounding shielding trace is positioned between the output line of the energy pickup area and the signal line of the main tactile sensing area, providing a discharge path for power supply disturbances. The floating isolation electrode is positioned in the middle layer of the flexible substrate to weaken the impact of triboelectric charge accumulation on the tactile sensitive layer. The blank buffer interval does not house tactile sensitive units or energy pickup electrodes, retaining only the necessary flexible support layer and encapsulation layer to reduce abrupt changes in structural stiffness. Through these structures within the isolation transition zone, power supply disturbances are attenuated first in the isolation path, and bending deformation is released first in the buffer layer, ensuring that the disturbances do not reach the level of tactile signal misjudgment when they reach the main tactile sensing area.

[0060] During tactile data processing, the main tactile sensing area, energy pickup area, deformation reference area, and isolation transition area all carry corresponding regional structure identifiers. The tactile data frames output from the main tactile sensing area serve as the primary contact information input to the collaborative power supply control unit. The energy status frames and power supply disturbance frames output from the energy pickup area are used to determine the auxiliary compensation power supply capability and its impact on adjacent areas. The deformation status frames output from the deformation reference area are used to determine whether the main tactile sensing area needs to perform deformation compensation. The regional structure identifier of the isolation transition area is used to determine the adjacency relationship between the main tactile sensing area and the energy pickup area. When the power supply disturbance frame of the energy pickup area shows that the power supply ripple or transient current change reaches the power supply disturbance threshold, the collaborative power supply control unit determines the affected main tactile sensing area based on the regional structure identifier corresponding to the isolation transition area and notifies the signal decoupling transmission unit to perform power supply disturbance compensation on the tactile data of that main tactile sensing area. When the deformation state frame of the deformation reference area shows that the current strain has reached the deformation interference threshold, the signal decoupling transmission unit performs deformation correction on the tactile data of the tactile main sensing area according to the positional relationship between the deformation reference area and the tactile main sensing area.

[0061] In this embodiment, with the above arrangement, the main tactile sensing area is located at the position where contact information is most needed, ensuring the continuity of tactile data for the fingertips, fingertips, and high-probability contact areas; the energy pickup area is located in a position where main contact wear is less likely to occur and suitable for acquiring deformation or environmental energy, reducing the occupancy of the energy pickup structure on the tactile sensitive surface; the deformation reference area is located in a position that can stably characterize the joint bending state, providing a deformation basis for tactile data correction; the isolation transition area is located between the main tactile sensing area and the energy pickup area, simultaneously reducing electrical disturbances and mechanical deformation transmission. Therefore, the flexible electronic skin can achieve a partitioned arrangement of the tactile sampling area, energy pickup area, and deformation feedback area on the dexterous hand surface, reducing the coupling effect of power supply disturbances and mechanical deformation on the tactile signal, and improving the stability and correctability of the tactile signal.

[0062] In another embodiment, the main tactile sensing area is located not only on the fingertips and pads, but also on the edge of the palm, the web area, or the lateral contact area for holding thin sheets; the energy pickup area is located not only on the sides, backs, and outward bends of the joints, but also on the flexible extension section on the back of the hand or the wrist connection transition section; the deformation reference area is located not only in the joint transition area and the bending stress concentration area, but also in the flexible connection area between adjacent phalanges; the structure of the isolation transition area can be selected according to the source of disturbance. When the main disturbance is power supply ripple, grounding shielding traces and thickened insulation layers are preferred; when the main disturbance is mechanical stretching, low-modulus buffer layers and blank buffer intervals are preferred; when power supply disturbance and mechanical deformation coexist, a composite structure of shielding traces, buffer layers, and blank intervals is used. The above deformation methods do not change the basic process of zoning the main tactile sensing area, energy pickup area, deformation reference area, and isolation transition area according to contact probability, wear degree, and bending stress.

[0063] In one possible implementation, when the tactile sensing array generates tactile data frames, it groups the tactile channels according to the regional structure identifier, reads the initial baseline of each tactile channel in the non-contact state, removes abnormal channel data that exceeds the preset noise range, and normalizes the retained tactile channel data to extract contact intensity, contact area, number of activated channels, force change rate, force diffusion direction and sliding trigger frequency.

[0064] Before generating tactile data frames, the tactile sensing array first groups the tactile channels based on the region structure identifier. The region structure identifier includes the region number, curvature level, contact probability level, bending stress level, and functional partition type. After the power supply control unit reads the region structure identifier, tactile channels belonging to the same region number and with a functional partition type of tactile main sensing area are grouped into the same tactile channel group. For tactile channels located at the edge of the isolation transition zone and undertaking auxiliary contact detection functions, they are grouped into the extended channel group of the adjacent tactile main sensing area, but given a lower weight in subsequent contact intensity and contact area calculations. Each tactile channel group stores the channel sequence number, channel coordinates, effective sensing area of ​​a single channel, channel sensitivity coefficient, and the region structure identifier to which it belongs. When generating tactile data frames, the tactile sensing array reads the channel outputs sequentially according to the tactile channel groups, ensuring that the tactile signals in the same tactile data frame all have a clear region source and channel location.

[0065] Specifically, during assembly calibration, the dexterous hand is positioned in a naturally extended state without contact with external objects. Channel output is continuously collected for at least 60 seconds. After removing instantaneous spikes caused by power input, communication activation, or mechanical vibration during the collection period, the average level of the remaining output values ​​is taken as the initial baseline for the tactile channel, and this initial baseline is written to the storage unit. During power-on calibration, if the collaborative power supply control unit confirms that the dexterous hand is in a non-contact state based on the contact determination results of all areas, another segment of non-contact output values ​​is collected. The average level of this segment of output values ​​is compared with the initial baseline obtained from assembly calibration. When the difference between the two does not exceed the baseline's allowable drift range, the current baseline is updated with the weighted result of the two. When the difference exceeds the baseline's allowable drift range, the initial baseline obtained from assembly calibration is maintained, and the tactile channel is marked as a channel to be reviewed. The baseline's allowable drift range is determined by the output dispersion during the non-contact calibration phase. Specifically, the dispersion of the non-contact output values ​​relative to the average level is first calculated, and then three times this dispersion is taken as the allowable drift boundary.

[0066] The preset noise range is determined independently for each tactile channel, rather than using a fixed range shared by all channels. When determining the preset noise range, the output value of each channel is collected during the non-contact calibration phase. The average level and dispersion of the non-contact output for that channel are calculated, and the average level is extended upwards and downwards by a factor of three to form the initial noise range. For tactile channels located in areas with high curvature or high bending stress, their non-contact output drift is recorded during no-load bending calibration, and this drift is included in the noise range to ensure the preset noise range covers baseline changes caused by normal bending. For areas with significant power supply disturbances, the non-contact output offset caused by power path switching is also recorded during auxiliary power supply switching calibration, and this offset is included in the noise range. The resulting preset noise range simultaneously considers static electrical noise, flexible substrate bending drift, and power supply switching disturbances.

[0067] Specifically, the removal of abnormal channel data is completed during the generation of the tactile data frame. After the tactile sensing array reads the current channel output, it first compares the current channel output with the current baseline to obtain the tactile signal after baseline subtraction; then it determines whether the tactile signal exceeds the preset noise range corresponding to the channel. If the tactile signal of a certain channel exceeds the preset noise range alone when there is no contact response in adjacent channels, and the duration is less than one tactile closed-loop control cycle, then the current data of that channel is determined to be transient abnormal data and removed from the feature extraction process of the current tactile data frame. If a certain channel outputs saturation values, zero values, or abrupt values ​​with the opposite output direction to adjacent similar channels for multiple consecutive control cycles, then the channel is marked as an abnormal channel. The data of abnormal channels does not participate in the extraction of contact intensity, contact area, and force diffusion direction, but its channel number, abnormal type, and removal mark are still written into the tactile data frame for subsequent maintenance and compensation. If a certain channel exceeds the preset noise range and forms a continuous spatial distribution with adjacent channels, and the duration reaches more than one tactile closed-loop control cycle, it will not be removed as abnormal data, but will be retained as valid contact data to avoid the accidental deletion of real pressing or sliding contact.

[0068] Specifically, the normalization process is based on the channel's initial baseline, contact determination threshold, and saturation output value. For each retained channel, the current output is first subtracted from the channel's initial baseline to obtain the effective output after baseline subtraction. When the effective output is lower than the contact determination threshold, the normalization result of that channel is set to 0. When the effective output reaches or exceeds the saturation output value, the normalization result of that channel is set to 1. When the effective output is between the contact determination threshold and the saturation output value, the position of the effective output relative to that interval is converted to a value between 0 and 1. The contact determination threshold is determined by the average level of non-contact output and 3 times the dispersion, and the saturation output value is determined by the rated maximum contact pressure calibration. During calibration, standard pressure is applied to the tactile channels step by step, and the output value when the channel output no longer changes significantly with increasing pressure is recorded and used as the saturation output value. For tactile channels with different sensitivities, compensation is performed using the channel sensitivity coefficient before normalization to obtain similar normalization results when the same contact pressure is applied to different channels.

[0069] Contact intensity is determined by the normalized results of the retained and activated tactile channels within the same region. A tactile channel is activated when its normalized result reaches or exceeds an activation threshold, and the duration exceeds one sampling period. The activation threshold is determined during the touch calibration phase. Specifically, it involves applying the minimum effective contact force that can be stably recognized by the dexterous hand control system to the main tactile sensory area, recording the average level of the normalized results for each tactile channel, and taking 80% of this average level as the activation threshold. When extracting contact intensity, all activated channels are first screened out, and then a weighted average is calculated based on the normalized results and channel sensitivity compensation results of each activated channel to obtain the contact intensity of that region. For the fingertip region, tactile channels located at the center of the fingertip have a higher weight than those at the edges; for the fingertip region, tactile channels located within the main contact band have a higher weight than those near the isolation transition zone.

[0070] Furthermore, the contact area is determined by the number of activated channels and the effective sensing area corresponding to each activated channel. When extracting the contact area, first, the number of tactile channels reaching the activation threshold within the same tactile channel group is counted to obtain the number of activated channels; then, the effective sensing area of ​​each activated channel is read, and the effective sensing areas of all activated channels are added together to obtain the basic contact area. For cases where there are continuous intervals between adjacent activated channels, if the interval is less than half the center-to-center distance between adjacent tactile channels, the interval is included in the contact area; if the interval is greater than or equal to half the center-to-center distance between adjacent tactile channels, it is not included in the contact area. For curved surface areas with inconsistent channel spacing, the contact area is corrected according to the mapping relationship between each channel on the flexible substrate's unfolded plane and the attached curved surface, making the contact areas of the fingertip arc surface and the smooth fingertip area comparable.

[0071] The force change rate is obtained from the change in contact intensity between consecutive tactile data frames. After extracting the contact intensity in the current tactile data frame, the tactile sensing array compares it with the contact intensity in the previous tactile data frame to obtain the contact intensity increment; then, this increment is divided by the time interval between two frames to obtain the force change rate. If the current contact area significantly increases while the contact intensity increment is small, the force change rate is used to characterize the slow bonding process; if the contact area changes little while the contact intensity increment is large, the force change rate is used to characterize the rapid increase in local pressing pressure. The force change rate threshold is determined through rated gripping and impact contact calibration. Specifically, the force change rate distribution under stable gripping, rapid pressing, and undesired impact states is recorded respectively. The boundary value between the upper limit of stable gripping and the lower limit of rapid pressing is used as the rapid force change threshold, and the boundary value between the upper limit of rapid pressing and the lower limit of undesired impact is used as the impact warning threshold.

[0072] Furthermore, the force diffusion direction is determined by the change in the spatial distribution of the activated channels across continuous tactile data frames. The tactile sensing array first determines the force center position of the current frame based on the channel coordinates and normalization results of each activated channel. The determination process involves weighting the positions of each activated channel according to its normalization result; channels with higher normalization results contribute more to the force center position. Then, the force center position of the current frame is compared with that of the previous frame to obtain the force center movement direction. If the force center movement distance is less than the center drift threshold, it is determined that there is no effective force diffusion direction in that frame; if the force center movement distance reaches or exceeds the center drift threshold, the direction from the previous position to the current position is taken as the force diffusion direction. The center drift threshold is determined by static pressure calibration. Specifically, the fluctuation of the force center position is collected while the standard flat pressure head remains stationary, and three times this fluctuation range is taken as the center drift threshold. For sliding contact scenarios, the force diffusion direction is also determined by the activation order of the activated channels. The direction from the first activated channel to the later activated channel is used as the candidate sliding direction, and consistency is checked with the direction of movement of the force center.

[0073] The frequency of slip triggering is determined by the number of slip triggering events within a preset time window. A slip triggering event is determined by simultaneously meeting three conditions: 1. The order of activation of channels within the same tactile channel group changes continuously in one direction; 2. The distance the force center moves in that direction reaches or exceeds the center drift threshold; 3. The contact intensity or force change rate exhibits periodic fluctuations different from static pressure. The static pressure fluctuation range is determined during static holding calibration. Specifically, this involves having a dexterous hand hold a fixed object with a rated grip force, recording the natural fluctuation range of contact intensity and force change rate, and using three times this natural fluctuation range as the slip fluctuation determination threshold.

[0074] Furthermore, when the above three conditions are simultaneously met within the same preset time window, it is recorded as one slide trigger event. The process of extracting the slide trigger frequency is as follows: count the number of slide trigger events within the preset time window, and then convert the number of events into the number of triggers per unit time according to the length of the time window. The preset time window is determined by the dexterous hand closed-loop control cycle, and its length is not less than 3 tactile closed-loop control cycles to ensure that the complete process of slide occurrence, diffusion, and confirmation can be covered.

[0075] The final tactile data frame contains the region structure identifier, channel grouping information, initial baseline, abnormal channel removal flag, normalized tactile data, contact intensity, contact area, number of active channels, force change rate, force diffusion direction, and slip trigger frequency. For channel groups after removing abnormal data, if the number of remaining effective channels is less than 70% of the total number of tactile channels in the region, a low confidence flag is written into the tactile data frame, and the coordinated power supply control unit is notified to increase the scanning priority of adjacent channel groups in the region; if the number of remaining effective channels is not less than 70% of the total number of tactile channels in the region, the tactile data frame participates in the update of the region load, energy, and deformation correlation table according to normal confidence. The effective channel number threshold is determined by channel failure tolerance calibration, specifically by gradually blocking different numbers of channels in the main tactile sensing area, recording the tactile recognition accuracy, and when the recognition accuracy drops beyond the preset allowable error, the corresponding effective channel ratio is taken as the minimum effective channel ratio, which is 70% in this embodiment.

[0076] Through the above processing, the tactile sensing array completes region grouping, baseline subtraction, abnormal data removal, and normalization during the tactile data frame generation stage, reducing the impact of non-contact drift, local dead pixels, power supply transients, and mechanical bending on feature extraction. Contact intensity reflects the magnitude of local force, contact area reflects the tactile contact range, the number of activated channels reflects the size of the contact area, the rate of change of force reflects the pressing or releasing speed, the direction of force diffusion reflects the spatial migration of the contact area, and the frequency of sliding triggers reflects the relative sliding trend between the object and the dexterous hand. After these features are output along with the tactile data frame, the collaborative power supply control unit can determine the tactile load of the region and adjust the power supply level, scanning sequence, and sampling frequency accordingly. The signal decoupling transmission unit can also perform subsequent source compensation based on abnormal channel markings and normalized tactile data.

[0077] In another implementation, abnormal channel elimination is not limited to judgment based on a preset noise range, but also combines the consistency of adjacent channels, temporal continuity, and power supply disturbance frames for verification. When a power supply disturbance frame shows a power supply path switch or increased ripple in the current cycle, the tactile sensing array temporarily suspends marking the single frame of out-of-limit data as an abnormal channel, and waits for verification in the next control cycle; when the same type of abnormality occurs in two consecutive control cycles, and adjacent channels do not show a corresponding contact distribution, then abnormal elimination is performed. Normalization processing can also adopt a piecewise linear normalization method, using higher resolution normalization for the light touch range and compressed normalization for the near-saturated heavy pressure range to improve the light touch recognition sensitivity. The above processing method does not change the basic process of grouping according to regional structure identifiers, reading the non-contact initial baseline, eliminating abnormal channel data, and extracting contact intensity, contact area, number of activated channels, force change rate, force diffusion direction, and slip trigger frequency.

[0078] In one possible implementation, when the auxiliary power supply conditioning unit generates energy status frames and power supply disturbance frames, it identifies the source identifiers corresponding to different energy sources, collects the input voltage, input current and instantaneous power of each energy source, calculates the power fluctuation rate and input stability, and generates corresponding data frames by combining the energy storage voltage, charging and discharging status, equivalent internal resistance, output voltage deviation, output ripple amplitude and power supply path switching status.

[0079] In the above implementation, the auxiliary power supply conditioning unit first identifies the source identifier of the multi-source energy pickup unit within each control cycle. The source identifier consists of the energy source type, the area structure identifier, the energy pickup structure number, and the access port number. The energy source type includes at least two of the following: mechanical deformation energy, contact friction energy, ambient light energy, and thermoelectric energy; the area structure identifier indicates the area number, curvature level, contact probability level, bending stress level, and functional zone type corresponding to the energy source; the energy pickup structure number distinguishes different energy pickup units within the same area; and the access port number distinguishes different input channels on the auxiliary power supply conditioning unit. During the power-on initialization phase, the auxiliary power supply conditioning unit reads the access port number of each input channel and matches it with the pre-written area structure identifier and energy pickup structure number, thus ensuring that each input power source has a unique source identifier.

[0080] The auxiliary power supply conditioning unit collects input voltage, input current, and instantaneous power for each input channel corresponding to a source identifier. When collecting input voltage, the voltage sampling sequence within the current control cycle is obtained through an input voltage detection circuit. When collecting input current, the current sampling sequence within the current control cycle is obtained through a series sampling resistor, a Hall current detection structure, or a current mirror detection structure. Instantaneous power is obtained by multiplying the input voltage and input current at the same sampling moment. To reduce single-point sampling errors, the auxiliary power supply conditioning unit acquires multiple sets of voltage, current, and instantaneous power data within one control cycle and records the maximum input voltage, minimum input voltage, average input voltage, maximum input current, minimum input current, average input current, maximum instantaneous power, minimum instantaneous power, and average instantaneous power within that control cycle. This data, along with the source identifier, is written into the input-side field of the energy status frame.

[0081] Furthermore, the power fluctuation rate is determined by the instantaneous power change amplitude of the same source identifier within the current control cycle. Specifically, the auxiliary power supply conditioning unit first reads the maximum and minimum instantaneous power within the control cycle, calculates the difference between them to obtain the instantaneous power swing; then, it compares this instantaneous power swing with the average instantaneous power to obtain the power fluctuation rate. When the average instantaneous power is lower than the effective input power lower limit of the input channel, the power fluctuation rate is not directly calculated. Instead, the input state corresponding to the source identifier is marked as a low-energy input state to avoid misjudgment caused by proportional amplification under weak input conditions. The effective input power lower limit is determined during the factory calibration stage. During calibration, the corresponding energy pickup structure is continuously outputting under the lowest available operating state, and the minimum average power that can maintain stable identification by the auxiliary power supply conditioning unit is recorded. 1.2 times this minimum average power is taken as the effective input power lower limit.

[0082] Input stability is determined by the stability of the input voltage, the stability of the input current, and the degree of power fluctuation.

[0083] In specific processing, the auxiliary power supply conditioning unit first uses the average input voltage of the current control cycle as a benchmark to determine the degree of vertical deviation of the input voltage within that control cycle; then, it uses the average input current as a benchmark to determine the degree of vertical deviation of the input current within that control cycle; subsequently, it reads the power fluctuation rate. Input stability is evaluated using a percentage system, with a maximum score of 100 points. When the degree of input voltage deviation, the degree of input current deviation, and the power fluctuation rate are all within their respective calibrated allowable ranges, the input stability remains in the high stability range; when any item exceeds the corresponding calibrated allowable range, the corresponding score is deducted from 100 points according to the degree of deviation; when two or more items exceed the corresponding calibrated allowable ranges simultaneously, the deducted scores are accumulated. The deduction weights corresponding to input voltage deviation, input current deviation, and power fluctuation rate are determined during the calibration stage. During calibration, voltage disturbances, current disturbances, and power pulsations are input to the auxiliary power supply conditioning unit respectively, and the output voltage deviation, output ripple amplitude, and tactile channel noise increment are recorded. The normalized proportions of the impact of the three types of disturbances on the output side stability are used as the deduction weights. Input stability of 80 points or higher is considered stable; input stability of 50 points or higher but lower than 80 points is considered adjustable; input stability of less than 50 points is considered unstable. Unstable inputs are not directly connected to the core haptic channel power supply branch, but are only allowed to enter the energy storage branch or be connected later.

[0084] The energy storage voltage and charge / discharge state are obtained from the energy storage detection branch of the auxiliary power supply conditioning unit. The energy storage voltage is the current voltage across the energy storage capacitor, micro-energy storage unit, or flexible energy storage device. The charge / discharge state is determined by the direction of the energy storage branch current and the trend of energy storage voltage change. When the energy storage branch current flows into the energy storage device and the energy storage voltage increases over time, it is determined to be in a charging state. When the energy storage branch current flows out of the energy storage device and the energy storage voltage decreases over time, it is determined to be in a discharging state. When the energy storage branch current is lower than the resting current threshold and the energy storage voltage change does not exceed the resting voltage drift range, it is determined to be in a holding state. The resting current threshold and resting voltage drift range are determined during the no-load calibration phase of the energy storage device. During calibration, the load and input source are disconnected, and the natural changes in the energy storage branch current and energy storage voltage are continuously detected. Three times the natural change range is taken as the corresponding threshold boundary.

[0085] The equivalent internal resistance is used to characterize the voltage drop characteristics of the energy storage branch and its connecting paths during power supply. The auxiliary power supply conditioning unit performs short-term load disturbance detection within a time slot that does not affect the continuous sampling of the core haptic channel. It first records the energy storage voltage and output current of the energy storage branch before the disturbance, then connects a known additional load and records the energy storage voltage and output current after the disturbance. Subsequently, it determines the equivalent internal resistance based on the correspondence between the decrease in energy storage voltage and the increase in output current before and after the disturbance. When the increase in output current is lower than the minimum current change required for internal resistance detection, the current equivalent internal resistance detection result is discarded, and the previous valid detection result is used. The minimum current change is jointly determined by the energy storage voltage detection resolution and the current detection resolution. Specifically, the voltage detection result should be able to distinguish voltage changes of at least 3 sampling resolution units, and the current detection result should be able to distinguish current changes of at least 3 sampling resolution units. When the equivalent internal resistance exceeds the internal resistance warning threshold, the auxiliary power supply conditioning unit reduces the power supply priority of that energy storage branch to the core haptic channel. The internal resistance warning threshold is determined during the factory testing phase. Specifically, the series impedance of the energy storage branch is increased step by step, and the equivalent internal resistance when the output voltage deviation reaches 3% of the reference voltage is recorded. This equivalent internal resistance is then used as the internal resistance warning threshold.

[0086] The output voltage deviation is determined by the difference between the output voltage of the auxiliary power supply conditioning unit and the reference voltage of the tactile sensing array.

[0087] In specific processing, the auxiliary power supply conditioning unit collects the output voltage sequence within one control cycle, calculates its average output voltage, and then compares the average output voltage with the reference voltage to obtain the output voltage deviation. The output ripple amplitude is determined by the difference between the highest and lowest values ​​of the output voltage within the same control cycle. For control cycles in which power supply path switching occurs, the auxiliary power supply conditioning unit also records the output voltage deviation and output ripple amplitude for the stable segment before switching, the transient segment after switching, and the stable segment after switching, respectively, so that the power supply disturbance frame can distinguish between continuous ripple and path switching transients. When the output voltage deviation reaches 3% of the reference voltage, the power supply disturbance frame is marked as a medium-level disturbance; when the output voltage deviation reaches 5% of the reference voltage, the power supply disturbance frame is marked as a high-level disturbance. When the output ripple amplitude reaches 1.5% of the reference voltage, the power supply disturbance frame is marked as a medium-level ripple; when the output ripple amplitude reaches 3% of the reference voltage, the power supply disturbance frame is marked as a high-level ripple. The aforementioned proportional threshold is obtained through tactile channel anti-interference calibration. During calibration, voltage deviations and ripples of different amplitudes are injected into the power supply branch under no external contact conditions, and the output offset of the tactile channel is recorded. When the output offset of the tactile channel reaches 20% of the contact judgment threshold, the corresponding power supply disturbance level is determined as the medium-level boundary; when the output offset of the tactile channel reaches 40% of the contact judgment threshold, the corresponding power supply disturbance level is determined as the high-level boundary.

[0088] The power supply path switching status is recorded by the power path management circuit of the auxiliary power supply conditioning unit. The power supply path includes at least a real-time energy pickup path, an energy storage release path, a real-time energy pickup and energy storage parallel path, and a main power supply compensation path. Whenever a power supply path switch occurs, the auxiliary power supply conditioning unit records the path before switching, the target path, the switching trigger reason, the switching start time, the switching completion time, the maximum output voltage deviation during switching, the maximum output ripple amplitude during switching, and whether the switching was successful. Switching trigger reasons include insufficient available energy, increased tactile load prediction, energy storage voltage reaching the discharge allowable range, input stability below the stable input range, or equivalent internal resistance exceeding the internal resistance warning threshold. If the difference between the target path voltage and the current output voltage exceeds 3% of the reference voltage, the auxiliary power supply conditioning unit first performs current-limited pre-charging or delayed switching of the target path; when the target path voltage approaches the current output voltage, the power supply path switching is then performed. If the switching conditions cannot be met within the preset switching waiting time, the original power supply path is maintained, and a switching prevention flag is written into the power supply disturbance frame. The preset switching waiting time is determined by the tactile closed-loop control cycle, and its length does not exceed one tactile closed-loop control cycle, so as to avoid the power supply switching waiting process affecting the real-time performance of tactile control.

[0089] The energy status frame describes the available energy supply status of each energy source and energy storage branch. This frame includes at least the frame number, generation time, area structure identifier, source identifier, input voltage, input current, instantaneous power, average instantaneous power, power fluctuation rate, input stability, energy storage voltage, charging / discharging status, equivalent internal resistance, current power supply level, and current power supply path. In cases where multiple energy sources exist within the same area, the auxiliary power supply conditioning unit writes the input-side data corresponding to each source identifier into the same energy status frame and writes the overall available energy status of the area at the end of the frame. The overall available energy status is determined jointly based on the number of stable inputs, the number of conditionable inputs, whether the energy storage voltage reaches the discharge allowable range, and whether the equivalent internal resistance exceeds the internal resistance warning threshold. When there are a large number of stable inputs, the energy storage voltage is within the discharge allowable range, and the equivalent internal resistance does not exceed the internal resistance warning threshold, this area is marked as capable of supporting high-stability sampling power supply. When there are insufficient stable inputs but conditionable inputs or available energy storage voltage, this area is marked as capable of supporting conventional sampling power supply. When stable inputs, conditionable inputs, and energy storage outputs cannot meet the core haptic channel requirements, this area is marked as in an energy-deficient state.

[0090] The power supply disturbance frame describes the possible sources and intensity of interference to the tactile signal during auxiliary power supply. This frame includes at least the frame number, generation time, affected area structure identifier, disturbance source identifier, output voltage deviation, output ripple amplitude, transient current change, power fluctuation rate, input stability, power supply path switching status, disturbance level, and compensation suggestion flag. The disturbance source identifier indicates whether the disturbance originates from a specific energy source, energy storage branch, or power supply path switching action. The disturbance level is determined jointly based on the output voltage deviation, output ripple amplitude, transient current change, and path switching status. A low-level disturbance is indicated when both the output voltage deviation and output ripple amplitude are below the medium-level boundary and no power supply path switching has occurred. A medium-level disturbance is indicated when any one of the output voltage deviation, output ripple amplitude, or transient current change reaches the medium-level boundary. A high-level disturbance is indicated when any one of the output voltage deviation, output ripple amplitude, or transient current change reaches the high-level boundary, or when the power supply path switching fails. The compensation suggestion flag is generated based on the disturbance level. Low-level disturbances are only written to the record field, medium-level disturbances trigger the signal decoupling transmission unit to perform power supply offset compensation, and high-level disturbances trigger sampling time slot avoidance and power supply path re-evaluation at the same time.

[0091] Through the above processing, the auxiliary power supply conditioning unit can correlate the source identifier, input voltage, input current, instantaneous power, power fluctuation rate, and input stability of the multi-source energy input side with the energy storage voltage, charging / discharging state, and equivalent internal resistance of the energy storage side, as well as the output voltage deviation, output ripple amplitude, and power supply path switching state of the output side. The energy status frame is used to provide the cooperative power supply control unit with the available energy and power supply capacity of each area, while the power supply disturbance frame is used to provide the disturbance source and disturbance intensity to the signal decoupling transmission unit. Since each input, energy storage, and output state is bound to the source identifier and area structure identifier, the cooperative power supply control unit can determine which energy sources are suitable for accessing the core haptic channel for power supply, which energy sources are only suitable for energy storage, and which power supply paths need to avoid haptic sampling time slots; the signal decoupling transmission unit can perform source-specific compensation on the haptic data frame according to the power supply disturbance frame, reducing the errors caused by input power fluctuations, increased energy storage internal resistance, and power supply path switching to the haptic signal.

[0092] In one possible implementation, when the deformation feedback unit generates a deformation state frame, it collects the bending angle, torsion angle, local stretching amount and deformation rate of the corresponding region, and performs region mapping according to the region structure identifier; the deformation feedback unit also extracts the deformation baseline under the state without external contact, and performs differential processing between the current deformation signal and the deformation baseline to obtain the regional deformation disturbance amount.

[0093] In the above embodiment, deformation feedback units are arranged according to regional structure identifiers in the adjacent boundary areas of the deformation reference area, joint transition area, bending stress concentration area, and tactile main sensing area of ​​the integrated flexible carrier. Each deformation feedback unit is bound to a corresponding regional structure identifier during assembly. The regional structure identifier includes a region number, curvature level, contact probability level, bending stress level, and functional partition type. When generating a deformation state frame, the deformation feedback unit first reads the regional structure identifier of its own location, then collects the bending angle, torsion angle, local stretching amount, and deformation rate of that region within the current control cycle, and maps the above data to the corresponding region number, so that the deformation state frame can clearly characterize the surface deformation state of a specific region.

[0094] In this embodiment, the flexible bending detection structure is arranged along the bending direction of the finger, and its output varies with the bending degree of the flexible integrated carrier. During assembly and calibration, the dexterous hand is controlled to sequentially assume a natural extended posture, a moderately bent posture, and a maximum permissible bending posture. The output values ​​of the flexible bending detection structure are recorded for each posture, and a correspondence is established between these output values ​​and the actual joint angles in the corresponding postures. During actual operation, the deformation feedback unit reads the current output value and converts it into the current bending angle based on the correspondence obtained from the calibration. For cases where multiple bending detection structures are arranged in the same area, data with abrupt output changes and inconsistencies with adjacent detection structures are first removed. Then, the bending angles of the remaining detection structures are averaged to obtain the current bending angle for that area.

[0095] Specifically, at least two flexible deformation detection branches with different directions are set in the joint transition area or finger side area, one of which is arranged along the length of the finger, and the other is arranged at an angle relative to the length of the finger. When the dexterous hand undergoes simple bending, the output changes of the detection branches in different directions correspond to a preset relationship; when the dexterous hand undergoes torsion, the output differences of the detection branches in different directions increase. During assembly calibration, the output changes of each detection branch under pure bending and known torsion angle conditions are recorded respectively to establish the correspondence between output differences and torsion angles. During actual operation, the deformation feedback unit reads the current output of the detection branches in different directions, subtracts the output component corresponding to pure bending, and determines the current torsion angle based on the remaining output differences. Thus, normal output changes caused by bending are not mistaken for torsional deformation.

[0096] In this embodiment, the local tensile amount is obtained through a flexible strain detection structure located on the surface or middle layer of the flexible integrated carrier. The flexible strain detection structure is positioned along the direction most prone to stretching and is bound to the regional structural identifiers of the tactile main sensing area, the isolation transition area, or the deformation reference area. During assembly calibration, the flexible integrated carrier is loaded with different stretch lengths under conditions without external contact, and the output value of the flexible strain detection structure at each stretch length is recorded. Based on this, the correspondence between the output change and the actual tensile amount is determined. During actual operation, the deformation feedback unit compares the current output value with the output value under no-stretch conditions and obtains the current local tensile amount based on the aforementioned correspondence. For areas with high curvature levels, the local tensile amount is also corrected in conjunction with the attachment arc length of that area, enabling the geometric elongation caused by bending attachment to be distinguished from the abnormal stretching caused by external contact.

[0097] The deformation rate of change is used to represent how quickly the bending angle, torsional angle, or localized stretching changes over time. After obtaining the bending angle, torsional angle, and localized stretching in the current control cycle, the deformation feedback unit reads the bending angle, torsional angle, and localized stretching corresponding to the previous control cycle, calculates the difference between the current value and the value of the previous control cycle, and combines this with the time interval between two control cycles to obtain the bending rate of change, torsional rate of change, and stretching rate of change. If any one of the three rates of change reaches the deformation abrupt change threshold, a deformation abrupt change flag is written into the deformation state frame. The deformation abrupt change threshold is determined in the no-load rapid bending calibration and rated gripping calibration. Specifically, it involves recording the deformation rate of change distribution under normal rapid bending, stable gripping, and undesired impact, respectively. The upper limit of normal rapid bending is used as the normal deformation boundary, and the lower limit of undesired impact is used as the deformation abrupt change boundary. When the current deformation rate of change reaches the deformation abrupt change boundary, the signal decoupling transmission unit performs dynamic deformation compensation on the tactile data of the same area or adjacent tactile main sensing areas.

[0098] Region mapping is accomplished through region structure identifiers and an adjacency table. The adjacency table is established during the design phase of the integrated flexible carrier and corrected after assembly calibration. This table records the region number corresponding to each deformation feedback unit, the adjacent tactile main sensing area, the adjacent energy pickup area, the width of the isolation transition area, and the direction of deformation influence. When a deformation feedback unit generates a deformation state frame, it writes the current bending angle, torsion angle, local stretching amount, and deformation rate of change into its own region; for the detection results of the deformation reference area, it also maps them to the adjacent tactile main sensing area according to the adjacency table.

[0099] During mapping, the tactile primary sensory area that is closer to the deformation reference area and located on the same bending transmission path receives a higher deformation influence weight; the tactile primary sensory area that is farther away or separated by an isolation transition area receives a lower deformation influence weight.

[0100] The aforementioned deformation influence weights are determined in the bending calibration. Specifically, the joint bending posture is gradually changed under no external contact, the output changes of the deformation reference area and the no-contact drift changes of the adjacent tactile main sensory areas are recorded, and the degree of correspondence between the two is used as the deformation influence weight.

[0101] In this embodiment, the deformation baseline under non-external contact conditions is jointly extracted by the deformation feedback unit during the calibration phase and the idle operation phase. During the calibration phase, the dexterous hand sequentially performs natural extension, low-speed bending, rated bending, and reset actions without contacting external objects. The deformation feedback unit collects the bending angle, torsion angle, and local stretching amount of each region under the above postures, and uses the average level of multiple collection results under the same posture as the deformation baseline of the corresponding posture.

[0102] During idle operation, the coordinated power supply control unit allows the deformation feedback unit to update the deformation baseline if all tactile main sensing areas have not reached the contact determination threshold based on the tactile data frame. If any tactile main sensing area reaches the contact determination threshold, the deformation baseline update is paused to prevent external contact pressure from being written into the non-contact baseline. When updating the deformation baseline, the calibration stage deformation baseline is not directly overwritten; instead, the newly acquired non-contact deformation data is smoothly updated with the original deformation baseline. When the difference between the newly acquired data and the original deformation baseline reaches the baseline anomaly threshold, no update is performed, and a baseline verification mark is written into the deformation status frame.

[0103] The baseline anomaly threshold is determined based on repeated bending calibration under no-external-contact conditions. Specifically, the dexterous hand is controlled to repeatedly perform the same bending action at least 10 times under no-external-contact conditions, and the deformation detection results of the same area under the same posture are recorded. The average level and dispersion are calculated, and the average level is expanded upward and downward by three times the dispersion to form an allowable range. When the current no-contact deformation data exceeds this allowable range, it indicates that the flexible integrated carrier may have loose attachment, local fatigue, detection branch drift, or unidentified external contact. In this case, the deformation feedback unit does not use it as a new deformation baseline.

[0104] The regional deformation disturbance is obtained through differential processing of the current deformation signal and the deformation baseline. Specifically, the deformation feedback unit first reads the bending angle, torsion angle, and local stretching amount within the current control cycle, and then reads the deformation baseline corresponding to the current regional structural identifier and the current posture. Subsequently, it calculates the difference between the current bending angle and the bending angle baseline, the difference between the current torsion angle and the torsion angle baseline, and the difference between the current local stretching amount and the local stretching baseline. These differences are used as the bending disturbance, torsion disturbance, and stretching disturbance, respectively. If the bending disturbance mainly occurs in the joint transition area, it is preferentially determined to be structural deformation caused by posture change; if the stretching disturbance mainly occurs at the edge of the tactile main sensing area and is accompanied by contact channel activation, it is preferentially determined to be local stretching caused by external contact; if the torsion disturbance increases simultaneously in the finger side area and the finger back area, it is preferentially determined to be deformation disturbance caused by finger twisting action. The above regional deformation disturbances are written into the deformation state frame and provided to the signal decoupling transmission unit to perform tactile signal deformation compensation.

[0105] In this embodiment, the deformation state frame includes at least the frame number, generation time, regional structure identifier, bending angle, torsional angle, local stretching amount, bending change rate, torsional change rate, stretching change rate, deformation baseline version number, bending disturbance amount, torsional disturbance amount, stretching disturbance amount, regional mapping result, and deformation abrupt change marker. For deformation state frames generated in the deformation reference area, the tactile main sensing area number and deformation influence weight affected by the frame are also written; for deformation state frames generated in the bending stress concentration area, the bending stress level and whether the deformation interference threshold has been reached are also written. The deformation interference threshold is determined by no-load bending calibration, specifically by recording the deformation disturbance distribution of each region under normal bending conditions without external contact, using the upper limit of the normal distribution as the conventional boundary, and then using the disturbance amount that obviously causes non-contact drift of the tactile channel as the interference boundary. When the regional deformation disturbance amount reaches the interference boundary, the signal decoupling transmission unit includes the tactile data frame corresponding to that region in the deformation compensation process.

[0106] Through the above processing, the deformation feedback unit can not only collect the bending angle, torsion angle, local stretching, and deformation rate of the current region, but also bind them with the regional structural identifier and map them to the affected tactile main sensing area according to the adjacency relationship. The deformation baseline in the state without external contact is used to represent the basic deformation generated by the movement of the dexterous hand itself and the normal attachment of the flexible integrated carrier. The regional deformation perturbation obtained by differentiating the current deformation signal from the deformation baseline is used to characterize the additional deformation beyond the normal posture change. This regional deformation perturbation can provide a basis for compensating for baseline drift, sensitivity changes, and structural noise in the tactile data frame, thereby reducing the misjudgment of tactile signals caused by bending, torsion, and local stretching, and improving the reliability of tactile data of the flexible electronic skin during the dynamic movement of the dexterous hand.

[0107] In another embodiment, the bending angle, torsion angle, and local tensile amount are obtained by one or more of a flexible resistive strain structure, a flexible capacitive strain structure, a flexible fiber optic sensing structure, or a flexible piezoelectric detection structure. The deformation baseline can be stored segmented according to posture, i.e., storing baseline data corresponding to natural extension, semi-bending, full bending, and reset postures respectively; or it can be stored according to a continuous posture curve, i.e., continuously determining the corresponding deformation baseline based on joint angles. Region mapping can use a fixed adjacency table or be regenerated based on the actual attachment deviation after assembly. The above replacement method does not change the basic process by which the deformation feedback unit collects bending angle, torsion angle, local tensile amount, and deformation rate, performs region mapping based on region structure identifiers, and obtains the region deformation disturbance amount by differential processing between the current deformation signal and the deformation baseline in a state without external contact.

[0108] In one possible implementation, the signal decoupling transmission unit uses the output ripple, power supply path switching time, and energy storage charging and discharging state in the power supply disturbance frame as power supply interference features, and the bending angle, torsion angle, local stretching amount, and deformation rate in the deformation state frame as deformation interference features. The unit then calculates the interference weights of the power supply interference features and deformation interference features on the tactile data frame to generate corresponding noise compensation terms.

[0109] In the above implementation, after receiving the tactile data frame, power supply disturbance frame, and deformation state frame, the signal decoupling transmission unit first performs frame alignment according to the region structure identifier and generation time. During frame alignment, the region number, curvature level, contact probability level, bending stress level, and functional partition type in the tactile data frame are matched with the same fields in the power supply disturbance frame and deformation state frame. When the power supply disturbance frame comes from the energy pickup area or energy storage branch, and the tactile data frame comes from the adjacent tactile main sensing area, the signal decoupling transmission unit determines the influence relationship between the two based on the adjacency relationship recorded in the isolation transition area. When the deformation state frame comes from the deformation reference area, and the tactile data frame comes from the adjacent tactile main sensing area, the signal decoupling transmission unit determines whether the deformation state frame participates in tactile data compensation based on the deformation influence weight. For data frames with inconsistent generation times, the power supply disturbance frame and deformation state frame with a time difference not exceeding one tactile closed-loop control cycle are selected based on the sampling time of the tactile data frame; data frames exceeding this time range do not participate in the compensation calculation of the current tactile data frame.

[0110] Power supply interference characteristics include output ripple, power supply path switching time, and energy storage charging and discharging status.

[0111] Output ripple is used to represent the voltage fluctuation amplitude at the output of the auxiliary power supply conditioning unit within the current control cycle; the power supply path switching time is used to represent the time position when the real-time energy pickup path, energy storage release path, real-time energy pickup and energy storage parallel path, or main power supply compensation path switches; the energy storage charging and discharging status is used to represent whether the energy storage branch is currently in a charging state, discharging state, or holding state. After reading the above power supply interference characteristics, the signal decoupling transmission unit first compares the output ripple with the ripple interference boundary obtained from the tactile channel anti-interference calibration to determine the level of influence of the output ripple on the tactile data; then it compares the power supply path switching time with the sampling time of the tactile data frame to determine whether the tactile sampling falls within the power supply switching influence time window; subsequently, it determines the contribution direction of the energy storage branch to the output voltage deviation and transient current change based on the energy storage charging and discharging status. During charging, the energy storage branch mainly exhibits changes in the input-side load, which can easily cause the output of the energy pickup area to drop. During discharging, the energy storage branch mainly exhibits changes in the output-side support, which can easily cause the power supply voltage of the tactile channel to shift. During holding, the impact of the energy storage branch on tactile data is lower than that during charging and discharging.

[0112] In this embodiment, deformation-related interference features include bending angle, torsion angle, local stretching amount, and deformation rate.

[0113] The bending angle characterizes the overall bending degree of the flexible integrated carrier along the joint bending direction; the torsion angle characterizes the rotational deformation of the flexible integrated carrier along the finger axis; the local stretching amount characterizes the degree of elongation of the flexible substrate around the tactile channel; and the deformation rate of change characterizes how quickly the above deformations change between adjacent control cycles. After reading the above deformation-related interference characteristics, the signal decoupling transmission unit first determines whether the bending angle exceeds the allowable range of the bending baseline in the region without external contact, then determines whether the torsion angle exceeds the allowable range of the torsion baseline, then determines whether the local stretching amount exceeds the allowable range of the stretching baseline, and finally determines whether the deformation rate of change reaches the deformation abrupt change boundary. Each of the above allowable ranges is determined by the deformation baseline in the state without external contact. Specifically, when the dexterous hand has no external contact, it repeatedly performs natural extension, low-speed bending, rated bending, and reset actions, records the average level and dispersion of the bending angle, torsion angle, and local stretching amount in the corresponding region, and uses the average level expanded upward and downward by 3 times the dispersion as the allowable range. The abrupt boundary of the deformation rate of change is determined by the calibration of unloaded rapid bending and undesired impact. The boundary between the upper limit of the normal rapid bending rate of change and the lower limit of the undesired impact rate of change is taken as the abrupt boundary.

[0114] Furthermore, the power supply interference weights are determined by the contribution of power supply disturbances to the output offset of the tactile channel. During calibration, the main tactile sensing area is kept in a state without external contact, and the flexible integrated carrier is kept from bending or stretching. Subsequently, output ripples of different amplitudes, power supply path switching at different time positions, and different energy storage charging and discharging states are applied. Each calibration changes only one type of power supply interference feature, while the other power supply conditions remain unchanged. The signal decoupling transmission unit records the output offset of the tactile channel relative to the interference-free state under each calibration condition, and compares the offsets caused by output ripple, power supply path switching, and energy storage charging and discharging states with the total power supply offset to obtain the contribution ratio of the three types of power supply interference features. These contribution ratios are normalized and then written into the power supply interference weight table. In actual operation, the signal decoupling transmission unit reads the corresponding weight from the power supply interference weight table based on the output ripple amplitude, power supply path switching time position, and energy storage charging and discharging status in the current power supply disturbance frame. When the current feature value is between two calibration points, the current weight is determined according to the changing trend of the two adjacent calibration points. When the current feature value exceeds the upper limit of the calibration range, it is processed according to the weight corresponding to the highest interference level, and an out-of-range mark is written in the power supply disturbance frame.

[0115] Deformation-type interference weights are determined by the contribution of deformation state to non-contact drift and changes in contact sensitivity of the tactile channel.

[0116] During calibration, the tactile main sensing area is initially placed in a state without external contact. Different bending angles, torsion angles, local stretching amounts, and deformation rates are applied, and the contactless output drift of the tactile channel is recorded. Subsequently, the tactile main sensing area is subjected to standard contact pressure, and the above deformation action is repeated, recording the change in tactile output under the same contact pressure. The signal decoupling transmission unit uses both the contactless output drift and the contact output change as the basis for evaluating deformation interference. It calculates the contribution ratio of bending angle, torsion angle, local stretching amount, and deformation rate to the total deformation interference, and writes this contribution ratio into the deformation interference weight table. For areas with high curvature or high bending stress levels, the deformation interference weight table uses the independent calibration results for that area; for areas with low curvature and low bending stress levels, the general calibration results under the same functional partition type are used. In actual operation, the signal decoupling transmission unit reads or determines the corresponding deformation interference weights based on the bending angle, torsion angle, local stretching amount, and deformation rate in the current deformation state frame.

[0117] Power supply interference weights and deformation interference weights are calculated separately within the same tactile data frame and are not simply superimposed. The signal decoupling transmission unit first determines whether the sampling time of the tactile data frame is within the power supply path switching influence time window. The power supply path switching influence time window is determined by the power supply path switching calibration. Specifically, multiple power supply path switchings are performed under no external contact conditions, and the time interval required for the tactile channel output to shift from a stable state to a stable state is recorded. The maximum value of this time interval is taken as the power supply path switching influence time window. When the sampling time of the tactile data frame is within this time window, the interference weight corresponding to the power supply path switching is increased; when the sampling time of the tactile data frame is outside this time window, the interference weight corresponding to the power supply path switching is decreased. For output ripple, the signal decoupling transmission unit determines the weight range based on the low-level, medium-level, or high-level disturbance boundary where the ripple amplitude is located. For energy storage charging and discharging states, the weight corresponding to the discharging state is greater than that of the charging state, and the weight corresponding to the charging state is greater than that of the holding state.

[0118] The deformation-related interference weights are further adjusted based on the regional deformation disturbance and the location of the tactile channel. If the tactile channel is located near a region of concentrated bending stress, the weights corresponding to the bending angle and deformation rate of change are increased; if the tactile channel is located on the finger side or fingertip edge, the weights corresponding to the torsional angle are increased; if the tactile channel is located near the boundary of the main tactile sensing area or an isolation transition area, the weights corresponding to the local stretching amount are increased. The above positional correction relationships are reflected in the regional structure identifier, and the signal decoupling transmission unit completes the correction based on the region number, bending stress level, and functional partition type. For deformation state frames that are mapped from adjacent deformation reference areas to the main tactile sensing area, the deformation-related interference weights are also adjusted based on the deformation influence weights between the two; regions that are close to each other and on the same bending transmission path have higher deformation-related interference weights; regions that are far apart or separated by isolation transition areas have lower deformation-related interference weights.

[0119] The noise compensation items consist of power supply noise compensation items and deformation noise compensation items. When generating power supply noise compensation items, the signal decoupling transmission unit first determines the ripple compensation amount based on the output ripple amplitude and the corresponding interference weight. Then, it determines the switching compensation amount based on the distance between the tactile data frame sampling time and the power supply path switching time, as well as the interference weight corresponding to the path switching. Subsequently, it determines the energy storage state compensation amount based on the energy storage charging, discharging, or holding state and the corresponding interference weight. The ripple compensation amount, switching compensation amount, and energy storage state compensation amount are synthesized according to their respective directions of offset from the tactile channel output to obtain the power supply noise compensation items. When generating deformation noise compensation items, the signal decoupling transmission unit first determines the bending compensation amount based on the deviation of the bending angle from the bending baseline and the corresponding interference weight. Then, it determines the torsional compensation amount based on the deviation of the torsional angle from the torsional baseline and the corresponding interference weight. Subsequently, it determines the stretching compensation amount based on the deviation of the local stretching amount from the stretching baseline and the corresponding interference weight, and finally determines the dynamic deformation compensation amount based on the deformation rate and the corresponding interference weight. The bending compensation, torsional compensation, tensile compensation, and dynamic deformation compensation are synthesized according to the output offset direction obtained from the calibration to obtain the deformation-type noise compensation term.

[0120] When the signal decoupling transmission unit performs compensation on the tactile data frame, it first reads the original tactile signal and normalized tactile data of the tactile channel, and then reads the power supply noise compensation item and the deformation noise compensation item. If the tactile channel is not identified as a contact channel, the power supply noise compensation item and the deformation noise compensation item are used first for baseline drift subtraction; if the tactile channel has been identified as a contact channel, while subtracting the baseline drift, the compensation amplitude is adjusted according to the range of contact intensity to avoid excessive reduction of the real contact signal. When the contact intensity is low, the compensation amplitude is limited by the light touch protection boundary; when the contact intensity is in the middle range, compensation is performed according to the calibration weight; when the contact intensity is close to saturation, the influence of deformation compensation on the contact intensity is reduced to keep the heavy pressure contact signal stable. The light touch protection boundary is determined by the minimum effective contact force calibration, specifically by applying the minimum contact force that can be stably recognized by the dexterous hand control system to the main tactile sensing area, recording the difference in tactile signals before and after compensation, and using the maximum compensation amount that does not cause the minimum effective contact to be eliminated as the light touch protection boundary.

[0121] After compensation, the signal decoupling transmission unit performs residual verification on the correction results. Residual verification is based on the median output of adjacent similar tactile channels within the same area, the historical continuous frame change trend, and the current contact area change. When the deviation between the compensated tactile data and the median output of adjacent channels exceeds the residual threshold, and this deviation exists for multiple consecutive control cycles, the signal decoupling transmission unit determines that the current interference weights are mismatched and sends a weight update flag back to the coordinated power supply control unit. The residual threshold is determined through rated contact compensation calibration. Specifically, the compensated tactile residuals are collected under standard contact pressure and standard bending posture, the average level and dispersion of the residuals are calculated, and the average level is expanded upwards by three times the dispersion as the residual threshold. After receiving the weight update flag, the coordinated power supply control unit reduces the sampling priority of non-core tactile channels in the affected area and re-evaluates the power supply path, tactile sampling time slot, and energy pickup time slot.

[0122] Through the above processing, the signal decoupling transmission unit categorizes output ripple, power supply path switching time, and energy storage charging / discharging state into power supply-related interference features, and bending angle, torsion angle, local stretching, and deformation rate into deformation-related interference features. It then obtains the interference weight of each interference feature on the tactile data frame through independent calibration and real-time correction. Power supply-related noise compensation terms are used to reduce electrical offsets caused by output ripple, power supply path switching, and energy storage state changes, while deformation-related noise compensation terms are used to reduce structural drift caused by bending, torsion, stretching, and dynamic deformation. Since the two types of noise compensation terms are generated separately and applied to the corresponding tactile channels according to the regional structure identifier, the actual contact component in the tactile data frame can be distinguished from the power supply disturbance component and deformation disturbance component, thereby reducing the coupling effect of multi-source power supply and surface deformation on the tactile perception result.

[0123] In another embodiment, the weights for power supply interference and deformation interference are determined using a hierarchical weight table. Power supply interference features are categorized into low, medium, and high levels, corresponding to different power supply compensation intensities; deformation interference features are categorized into normal deformation, significant deformation, and abrupt deformation, corresponding to different deformation compensation intensities. The hierarchical weight table is written into the signal decoupling transmission unit during the calibration phase, and the compensation items are determined directly by looking up the table based on the power supply disturbance frame and deformation status frame during the operation phase. For dexterous hand controllers with high computational resources, the weights for power supply interference and deformation interference can also be slightly updated through online residual verification, but each update increment does not exceed 10% of the factory-calibrated weights to avoid distortion of compensation items due to short-term abnormal data. The above replacement method does not change the basic processing procedure of calculating interference weights for power supply interference features and deformation interference features separately and generating corresponding noise compensation items.

[0124] In one possible implementation, the coordinated power supply control unit divides the same control cycle into energy pickup time slots, tactile sampling time slots, and disturbance assessment time slots, and updates the allocation ratio of each time slot according to the tactile load prediction value, energy availability value, and disturbance weight; when the power supply disturbance weight increases, the overlap ratio between the energy pickup time slot and the tactile sampling time slot is reduced or the power supply path switching time is avoided; when the deformation disturbance weight increases, the data acquisition frequency of the deformation reference area is increased, and the tactile baseline of the corresponding area is corrected according to the deformation state frame.

[0125] In the above implementation, the collaborative power supply control unit uses the dexterous hand tactile closed-loop control cycle as the basic control cycle, and divides each basic control cycle into energy pickup time slots, tactile sampling time slots, and disturbance evaluation time slots. The energy pickup time slot is used to receive energy inputs corresponding to mechanical deformation energy, triboelectric energy, ambient light energy, or temperature difference energy; the tactile sampling time slot is used to read tactile channel data from the main tactile sensing area and necessary auxiliary contact areas; the disturbance evaluation time slot is used to read the power supply disturbance frame, deformation state frame, and disturbance source analysis results output by the signal decoupling transmission unit. These three types of time slots are arranged according to the time sequence of the control cycle, and can also be staggered without affecting the continuous sampling of the core tactile channel. For the core tactile channel in the contact area, the tactile sampling time slot maintains an interval not exceeding the maximum allowable interval of the core tactile closed-loop control between two adjacent samples; for the tactile channel in the non-contact area, its sampling time slot can be reduced or delayed based on the current available energy value and the predicted tactile load value.

[0126] The initial values ​​of the time slot allocation ratio are determined during the assembly and calibration phase. During calibration, the dexterous hand is placed in non-contact extension, stable grasping, rapid pressing, and sliding detection states, respectively. The minimum sampling time required to maintain continuous sampling of the core tactile channel, the minimum energy pickup time required to complete multi-source energy input detection, and the data processing time required to complete disturbance assessment are recorded under different tactile tasks. The collaborative power supply control unit uses the above three minimum times as the lower boundary of the time slot allocation and determines the adjustable range of each type of time slot based on the total duration of one control cycle. The lower boundary of the tactile sampling time slot is determined by the continuous sampling requirement of the core tactile channel, the lower boundary of the energy pickup time slot is determined by the minimum effective input power identification requirement, and the lower boundary of the disturbance assessment time slot is determined by the time required to complete the reading of power supply disturbance frames and deformation state frames, the update of disturbance weights, and the feedback of compensation results.

[0127] During operation, the collaborative power supply control unit first reads the predicted tactile load, available energy, power supply interference weight, and deformation interference weight for each region, and then updates the time slot allocation ratio based on the above data. When the predicted tactile load of a region increases and the available energy is sufficient, the proportion of tactile sampling time slots in that region increases, and the energy pickup time slots are kept within a range that can compensate for sampling power consumption. When the predicted tactile load of a region increases but the available energy is insufficient, the collaborative power supply control unit prioritizes retaining the sampling time slots of the core tactile channel in the contact area and compresses the sampling time slots of the tactile channel in the non-contact area, while allocating the energy pickup time slots to time positions that will not interfere with the core tactile sampling. When the predicted tactile load is low and the available energy is sufficient, the disturbance assessment time slots and self-calibration related acquisition time slots receive higher priority and are used to update the tactile baseline and interference compensation parameters.

[0128] When the power supply interference weight increases, the coordinated power supply control unit reduces the overlap ratio between the energy pickup time slot and the tactile sampling time slot. The criteria for determining an increase in the power supply interference weight include the output ripple reaching the medium-level disturbance boundary, power supply path switching occurring near tactile sampling, the energy storage branch transitioning from a holding state to a discharging state, or transient current changes reaching the disturbance assessment threshold. The threshold for increasing the power supply interference weight is determined during the power supply immunity calibration phase. Specifically, under conditions of no external contact in the main tactile sensing area, the output ripple is increased step by step, power supply path switching and energy storage charging / discharging states are executed, and the tactile channel output offset is recorded. When the tactile channel output offset reaches 20% of the contact determination threshold, the corresponding interference weight is used as the medium-level interference boundary; when the tactile channel output offset reaches 40% of the contact determination threshold, the corresponding interference weight is used as the high-level interference boundary. During operation, when the power supply interference weight reaches the medium-level interference boundary, the coordinated power supply control unit reduces the simultaneous occurrence time of energy pickup and tactile sampling actions; when the power supply interference weight reaches the high-level interference boundary, the coordinated power supply control unit moves the energy pickup action outside the core tactile sampling time slot and prohibits the start of core tactile channel sampling within the power supply path switching influence time window.

[0129] Before switching power supply paths, the auxiliary power supply conditioning unit sends a switching warning flag to the collaborative power supply control unit. This flag includes the target power supply path, the expected start time of the switching, and the expected completion time of the switching. Based on the impact time window obtained from the power supply path switching calibration, the collaborative power supply control unit combines the reserved time before the expected start time of the switching, the switching execution time, and the recovery time after the switching is completed into a buffer zone. If a tactile sampling time slot falls into this buffer zone, the collaborative power supply control unit delays the sampling of non-core tactile channels. For core tactile channels, if sampling cannot be delayed, the energy pickup branch access strength is reduced, and the signal decoupling transmission unit is required to forcibly write a power supply compensation flag into the sampling frame. The impact time window of the power supply path switching is obtained through multiple switching calibrations. During calibration, the complete time for the output voltage to deviate from a stable state to return to stability is recorded, and the maximum value among the multiple results is taken as the base time of the buffer zone.

[0130] When the weight of deformation-related interference increases, the coordinated power supply control unit increases the data acquisition frequency of the deformation reference area. The criteria for determining an increase in the weight of deformation-related interference include bending angle exceeding the allowable range of the bending baseline, torsion angle exceeding the allowable range of the torsion baseline, local stretching exceeding the allowable range of the stretching baseline, or the deformation rate reaching the deformation abrupt change boundary. When the weight of deformation-related interference reaches the medium-level boundary, the data acquisition frequency of the deformation reference area is increased to twice the normal frequency; when the weight of deformation-related interference reaches the high-level boundary, the data acquisition frequency of the deformation reference area is increased to the same sampling frequency as the core tactile channel. These boundaries are determined in no-load bending, no-load torsion, local stretching, and dynamic deformation calibration. Specifically, this involves recording the non-contact drift and contact sensitivity changes of the tactile channel under each deformation action. When the drift reaches 20% of the contact determination threshold, it is considered a medium-level boundary; when the drift reaches 40% of the contact determination threshold, it is considered a high-level boundary.

[0131] Specifically, the coordinated power supply control unit reads the bending angle, torsion angle, local stretching, deformation rate, and regional deformation disturbance in the deformation state frame, and determines the affected tactile main sensing area based on the regional structure identifier. For tactile channels without external contact, the deviation between the current tactile output and the original tactile baseline is recorded as baseline drift caused by deformation; for tactile channels with external contact, the baseline is not directly updated, but a deformation-type noise compensation term generated by the signal decoupling transmission unit is used for temporary correction. Tactile baseline updates are only performed when the predicted tactile load is lower than the idle threshold, the number of contact channels is 0 or lower than the minimum number of contact channels, and no external impact marker is written in the deformation state frame. The idle threshold is determined through contactless action calibration, specifically by controlling the dexterous hand to perform natural extension, low-speed bending, and reset actions in a contactless state, recording the average level and dispersion of the predicted tactile load, and expanding the average level upward by three times the dispersion as the idle threshold.

[0132] Through the aforementioned time slot division and update methods, the collaborative power supply control unit can ensure that energy pickup, tactile sampling, and disturbance assessment have clear time boundaries within the same control cycle. When power supply-related interference increases, the probability of power supply disturbances entering the tactile data frame is reduced by minimizing overlap between energy pickup and sampling, avoiding power supply path switching moments, and protecting the core tactile channel during sampling. When deformation-related interference increases, the drift caused by bending, torsion, and stretching of the flexible substrate can be promptly reflected in the tactile data correction process by increasing the sampling frequency of the deformation reference area and correcting the tactile baseline. Therefore, the auxiliary power supply process, tactile sampling process, and disturbance assessment process can be dynamically adjusted according to tactile load, energy state, and disturbance state, improving the operational stability of the dexterous hand flexible electronic skin under multi-source power supply and dynamic curved surface deformation conditions.

[0133] In another embodiment, the energy pickup time slot, tactile sampling time slot, and disturbance assessment time slot employ a combination of segmented fixed boundaries and locally adjustable boundaries. For core tactile channels, the tactile sampling time slot uses fixed boundaries; for non-core tactile channels and low-contact probability areas, the tactile sampling time slot uses adjustable boundaries; for disturbance assessment time slots, the time slots are shortened when both power supply interference weights and deformation interference weights are low, and lengthened after either type of interference weight reaches the medium-level boundary. These modifications do not alter the process of updating the time slot ratio based on tactile load prediction, available energy, and interference weights, and instead avoid power supply path switching moments when power supply interference weights increase, and increase the sampling frequency of the deformation reference area and correct the tactile baseline when deformation interference weights increase.

[0134] In one possible implementation, the collaborative power supply control unit performs the following processing steps when generating the power supply control strategy and the haptic sampling control strategy: Step 1: Bind tactile data frames, energy status frames, power supply disturbance frames, and deformation status frames to regions based on the region structure identifier; Step 2: Time-align various data frames in the same region based on the synchronization time tag; Step 3: Calculate the tactile load prediction value based on the tactile data frame, the energy availability value based on the energy status frame, and the deformation interference value based on the deformation status frame; Step 4: Generate a region load, energy, and deformation correlation table based on the tactile load prediction value, energy availability value, and deformation interference value; Step 5: Generate a power supply control strategy and a tactile sampling control strategy based on the region load, energy, and deformation correlation table; Step 6: Perform a closed-loop update of the power supply control strategy and the tactile sampling control strategy based on the interference source analysis results.

[0135] In the above implementation, in step 1, after receiving the tactile data frame, energy status frame, power supply disturbance frame, and deformation status frame, the collaborative power supply control unit first reads the regional structure identifier carried in each frame. The regional structure identifier includes the region number, curvature level, contact probability level, bending stress level, and functional partition type. When binding regions, the region number is used as the primary index, and the functional partition type is used as the secondary index to classify tactile data, energy status, power supply disturbance, and deformation status from the same region into the same region buffer. For energy status frames and power supply disturbance frames from the energy pickup area, if they affect adjacent tactile main sensing areas through the isolation transition area, the frame is bound to the affected tactile main sensing area simultaneously according to the adjacency table. For deformation status frames from the deformation reference area, if their deformation influence weight reaches the mapping threshold, the frame is bound to the corresponding tactile main sensing area. The mapping threshold is determined in the bending calibration. Specifically, the bending posture of the deformation reference area is changed in the state of no external contact, and the non-contact drift of the adjacent tactile main sensing area is recorded. When the drift reaches 20% of the contact judgment threshold, the corresponding deformation influence weight is used as the mapping threshold.

[0136] In step 2, the collaborative power supply control unit performs time alignment on various data frames in the same area based on the synchronization time tag. The synchronization time tag is allocated by a unified clock, and the sampling time of the tactile data frame, the detection time of the energy state frame, the disturbance occurrence time of the power supply disturbance frame, and the deformation acquisition time of the deformation state frame are all recorded using this unified clock. During time alignment, the sampling time of the tactile data frame is used as the center time, and energy state frames, power supply disturbance frames, and deformation state frames within the time allowable range are selected to participate in the current cycle control. The time allowable range is determined by the tactile closed-loop control cycle, and its length is no more than one tactile closed-loop control cycle. For power supply path switching events, if the switching time is within the switching impact time window before and after the tactile sampling time, the power supply disturbance frame participates in the interference evaluation of the current tactile data frame; for deformation abrupt events, if the deformation change rate reaches the deformation abrupt boundary, even if there is a deviation of less than half a control cycle between the deformation state frame and the tactile data frame, it still participates in the compensation of the current tactile data frame to avoid rapid deformation not being corrected in time.

[0137] In step 3, the predicted tactile load is calculated based on the tactile data frame. The collaborative power supply control unit reads the contact intensity, contact area, number of activated channels, force change rate, force diffusion direction, and sliding trigger frequency from the tactile data frame, and uses these factors as load evaluation factors. During calculation, each factor is first compared with its corresponding rated value to obtain its relative level; then, weights are assigned according to the degree of influence of each factor on tactile recognition accuracy and closed-loop control requirements; finally, the relative levels are synthesized according to their weights to obtain the predicted tactile load for the current area. The load weights are determined in the standard grasp, sliding recognition, light touch positioning, and rapid press calibration. Specifically, the contact intensity, contact area, number of activated channels, force change rate, and sliding trigger frequency are changed item by item, the increase in tactile recognition error and closed-loop control delay is recorded, and the contribution ratio corresponding to each increase is used as the weight source. When the predicted tactile load reaches the tactile load increase threshold and continues for 3 control cycles, the area is identified as a high-load area.

[0138] In the same step, the available energy value is calculated based on the energy status frame. The coordinated power supply control unit reads the input voltage, input current, instantaneous power, power fluctuation rate, input stability, energy storage voltage, charge / discharge state, and equivalent internal resistance corresponding to each source identifier. For stable input sources, the available input energy is determined according to the effective power and time it can continuously provide within the current control window; for adjustable input sources, the conversion losses in the rectification, current limiting, energy storage, and voltage regulation processes are deducted before being included in the available input energy; for unstable input sources, the available energy for the core haptic channel is not directly included, only the energy storage candidate energy is allowed to be included. The energy storage release capacity is determined based on the energy storage voltage, the minimum allowable operating voltage, and the equivalent internal resistance; the higher the equivalent internal resistance, the lower the proportion of the release capacity allowed for the core haptic channel. The coordinated power supply control unit adds the available input energy to the energy storage release capacity, and then subtracts the reserved energy for the control circuit and communication link to obtain the available energy value for the current area. The energy reserve is based on the factory calibration results, taking the maximum consumption of the core control and communication links in 10 consecutive control cycles and adding a 20% margin.

[0139] In the same step, the deformation interference value is calculated based on the deformation state frame. The coordinated power supply control unit reads the bending angle, torsion angle, local tensile amount, deformation change rate, and regional deformation disturbance amount, and compares them with the allowable range of the deformation baseline under the condition of no external contact. The degree to which the bending angle exceeds the allowable bending range, the degree to which the torsion angle exceeds the allowable torsion range, the degree to which the local tensile amount exceeds the allowable tensile range, and the degree to which the deformation change rate approaches the deformation abrupt boundary are used as deformation interference evaluation factors. The deformation interference value is obtained by synthesizing each evaluation factor according to the deformation interference weight. The deformation interference weight is determined by non-contact bending drift calibration and standard contact sensitivity calibration. Specifically, bending, torsion, tension, and dynamic deformation conditions are changed respectively, and the output changes of the tactile channel under non-contact drift and the same contact pressure are recorded. The contribution ratio of each deformation factor to the total deformation interference is used as the weight.

[0140] In step 4, the collaborative power supply control unit generates a regional load, energy, and deformation correlation table based on the predicted tactile load, available energy, and deformation interference values. This correlation table uses the regional structure identifier as an index, with one row of data corresponding to each region. It records at least the region number, functional zone type, predicted tactile load, tactile load level, available energy, energy status level, deformation interference value, deformation interference level, power supply interference weight, deformation interference weight, number of core tactile channels, current power supply level, current sampling frequency, and current time slot allocation status. The tactile load level is determined by comparing the predicted tactile load with idle thresholds, normal load thresholds, and high load thresholds; the energy status level is determined by comparing the available energy with the energy requirements of core tactile channels, normal sampling energy requirements, and high-stability sampling energy requirements; and the deformation interference level is determined by comparing the deformation interference value with medium-level interference boundaries and high-level interference boundaries. All the above thresholds are obtained through a calibration process and are slightly updated during operation through self-calibration.

[0141] In step 5, the collaborative power supply control unit generates power supply control strategies and tactile sampling control strategies based on the regional load, energy, and deformation correlation table. When generating the power supply control strategy, areas with high tactile load levels and high contact probability levels are first selected as priority power supply areas. Then, the energy status level and power supply interference weight of that area are read. If the energy status level meets the high-stability sampling requirements and the power supply interference weight is below the medium-level boundary, the area is switched to the high-stability power supply level, allowing the real-time energy pickup path and energy storage release path to participate in power supply together. If the energy status level is insufficient but there is a continuous sampling requirement for the core tactile channel, only the core tactile channel power supply branch is retained, and the auxiliary power supply branch for non-contact areas is closed. If the power supply interference weight reaches the high-level boundary, the access of the energy pickup path is delayed or the power supply path switching time is avoided. When generating the tactile sampling control strategy, the regional sampling priority is first determined according to the predicted tactile load value and contact probability level, and then the sampling frequency is adjusted according to the available energy value and deformation interference value. In regions with high tactile load, sufficient energy, and low deformation interference, increase the sampling frequency; in regions with high tactile load but insufficient energy, maintain continuous sampling of the core tactile channel and reduce the sampling frequency of non-core channels; in regions with high deformation interference, increase the sampling frequency of the deformation reference area and perform deformation correction on the tactile baseline.

[0142] In step 6, the collaborative power supply control unit performs closed-loop updates to the power supply control strategy and the tactile sampling control strategy based on the interference source analysis results. The interference source analysis results are output by the signal decoupling transmission unit and include at least power supply noise compensation items, deformation noise compensation items, correction residuals, abnormal channel markers, and weight update markers. If the correction residual exceeds the residual threshold for multiple consecutive control cycles, and the power supply noise compensation items account for the majority, the collaborative power supply control unit reduces the overlap ratio between the energy pickup time slot and the tactile sampling time slot, delays the power supply path switching, and lowers the priority of the corresponding power supply path. If the correction residual mainly comes from deformation noise compensation items, the collaborative power supply control unit increases the data acquisition frequency of the deformation reference area, corrects the tactile baseline of the affected tactile main sensing area, and reduces the sampling frequency of non-core channels in high bending stress areas. If abnormal channel markers are concentrated in the same area, a low-confidence marker is written into the tactile data frame of that area, and the scanning priority of adjacent tactile channel groups is increased. The updated power supply control strategy and tactile sampling control strategy are executed in the next control cycle, and the execution results are rewritten into the regional load, energy and deformation correlation table.

[0143] After adopting the above steps, the collaborative power supply control unit can first complete area binding and time alignment, and then extract control criteria from the three types of data: tactile, energy, and deformation. Subsequently, it generates a correlation table of area load, energy, and deformation, and outputs the power supply level, power supply path, sampling frequency, scanning sequence, and time slot ratio accordingly. After the interference source analysis results enter the closed-loop update, power supply disturbances and deformation disturbances are no longer just supplementary data after compensation, but directly influence subsequent power supply control strategies and tactile sampling control strategies, so that the tactile data correction results can be reflected in the power supply and sampling control of the next cycle.

[0144] In another embodiment, steps 1 to 6 are executed sequentially within the same processor, or they can be executed separately by the tactile sampling processor, the power management processor, and the deformation compensation processor. When executed separately, the area structure identifier and synchronization time tag maintain a unified format, and the area load, energy, and deformation correlation table is centrally maintained by the collaborative power supply control unit. The above variations do not change the basic process of binding areas based on the area structure identifier, aligning time based on the synchronization time tag, calculating the predicted tactile load value, available energy value, and deformation interference value separately, and performing closed-loop updates based on the interference source analysis results.

[0145] In one possible implementation, the collaborative power supply control unit is also used to perform full life cycle self-calibration processing. When the predicted tactile load value is lower than the preset idle threshold and the available energy value is higher than the preset calibration threshold, the deformation reference area and the tactile channel in the non-contact state are selected as calibration reference areas. Zero-load tactile data frames, idle power supply disturbance frames and natural deformation state frames are collected. The tactile baseline drift, power supply disturbance transmission coefficient, deformation-to-tactile coupling coefficient and energy storage equivalent internal resistance change are calculated. Based on the tactile baseline drift, power supply disturbance transmission coefficient, deformation-to-tactile coupling coefficient and energy storage equivalent internal resistance change, the generation parameters of the noise compensation term, the weight coefficients in the regional load, energy and deformation association table, and the allocation boundary between the energy pickup time slot and the tactile sampling time slot are updated, so that the system maintains the matching relationship between the power supply control strategy, the tactile sampling control strategy and the dynamic correction of tactile data after the flexible substrate creeps, electrode aging or energy storage performance decays.

[0146] In the above implementation, the full lifecycle self-calibration process is performed during the idle phase of the dexterous hand operation, without occupying the critical sampling time of the core tactile channel. The collaborative power supply control unit first determines whether the self-calibration entry conditions are met. The entry conditions include a tactile load prediction value lower than a preset idle threshold, an energy availability value higher than a preset calibration threshold, no effective contact detected in the core tactile channel, no emergency switching event in the current power supply path, and no external impact marker written to the deformation state frame. The preset idle threshold is determined through contactless action calibration, specifically by performing natural extension, low-speed bending, and reset actions in a state without external contact, recording the average level and dispersion of the tactile load prediction value, and then expanding the average level upward by three times the dispersion as the preset idle threshold. The preset calibration threshold is determined based on the controller operating energy, tactile idle sampling energy, deformation sampling energy, and power supply disturbance detection energy required to complete one self-calibration, with an additional 20% energy margin. The collaborative power supply control unit only initiates the self-calibration process when the energy availability value is higher than the energy required for calibration with the aforementioned margin.

[0147] The calibration reference area consists of a deformation reference area and tactile channels in a non-contact state. The deformation reference area is selected from regions with good bending output repeatability, minimal impact from external contact, and stable regional structural markings. The tactile channels in a non-contact state are selected from tactile data frames that have not reached the contact judgment threshold, have not formed a continuous contact distribution with adjacent channels, and have not been marked as abnormal channels. If the number of effective non-contact tactile channels in a certain area is less than 70% of the total number of tactile channels in that area, that area is not used as a calibration reference area; if all areas do not meet this condition, the current self-calibration is paused, and the previous calibration parameters are maintained. The 70% effective channel ratio is determined by channel failure tolerance calibration, specifically by gradually blocking different numbers of tactile channels, recording the tactile recognition accuracy, and when the recognition accuracy drops beyond the allowable error, the corresponding effective channel ratio is taken as the minimum calibration reference ratio.

[0148] Zero-load tactile data frames are acquired without external contact. The coordinated power supply control unit controls the tactile sensing array to read the output of each tactile channel within the calibration reference area and records the current tactile baseline, normalization result, abnormal channel marker, and region structure identifier. The tactile baseline drift is determined by the difference between the current zero-load tactile output and the previous valid baseline. In specific processing, for each calibration reference channel, its current zero-load output average is read and compared with the historical tactile baseline in the storage unit. The difference between the two is taken as the tactile baseline drift of that channel. Then, a consistency check is performed on the drift of multiple calibration reference channels in the same area. If most channels in the same area drift in the same direction, it is determined to be a regional baseline drift; if only a few channels drift significantly and are inconsistent with adjacent channels, it is determined to be a local channel anomaly, and the regional baseline corresponding to those few channels is not updated. When the regional baseline drift reaches the baseline update threshold, the coordinated power supply control unit corrects the regional tactile baseline according to the smooth update method; if the baseline update threshold is not reached, only the drift amount is recorded, and no baseline update is performed. The baseline update threshold is determined by long-term zero-load stability calibration, and is taken as 3 times the natural drift range during long-term contactless operation as the update boundary.

[0149] The no-load power supply disturbance frame is acquired when there is no external contact and the tactile channel is in a zero-load state. The auxiliary power supply conditioning unit sequentially maintains the current power supply path, switches to the energy storage release path, switches to the real-time energy pickup path, and switches to the real-time energy pickup and energy storage parallel path, and records the output ripple, output voltage deviation, transient current change, and energy storage charging and discharging status in each path. The power supply disturbance transmission coefficient is determined based on the correspondence between the power supply disturbance change and the tactile zero-load output offset. In specific processing, the collaborative power supply control unit first reads the output ripple and power supply path switching status in the power supply disturbance frame, and then reads the output offset of the zero-load tactile data frame in the same time window; when the output ripple increases and the tactile zero-load output offset occurs synchronously, the correspondence between the tactile offset and the power supply disturbance increment is recorded as the power supply disturbance transmission coefficient for that area. If the power supply disturbance change does not cause the tactile zero-load output offset, the power supply disturbance transmission coefficient for that area remains unchanged. When the power supply disturbance transmission coefficient exceeds 1.5 times the factory calibration value, the collaborative power supply control unit increases the generation weight of power supply noise compensation items in this area and restricts the overlap between energy pickup time slot and tactile sampling time slot.

[0150] Natural deformation state frames are acquired during the dexterous hand's natural movements without external contact. A coordinated power supply control unit controls the dexterous hand to perform low-speed bending, resetting, and small-amplitude torsion movements. The deformation feedback unit records the bending angle, torsion angle, local stretching, and deformation rate of the deformation reference area. The coupling coefficient from deformation to tactile feedback is determined based on the correspondence between natural deformation changes and the zero-load tactile output drift. Specifically, the regional deformation disturbance in the deformation state frame is first read, followed by the zero-load tactile output drift of the corresponding primary tactile sensing area within the same time window. When the deformation disturbance increases and causes the zero-load tactile output to shift in the same or stable direction, the correspondence between the tactile drift and the deformation disturbance is recorded as the coupling coefficient from deformation to tactile feedback. For different deformation types, the coupling coefficients from bending to tactile feedback, torsion to tactile feedback, stretching to tactile feedback, and dynamic deformation to tactile feedback are recorded separately. If a certain coupling coefficient continues to increase compared to the factory calibration value, it indicates that the flexible substrate may be creeping, the attachment layer may be loosening, or the stiffness of the encapsulation layer may be changing. The coordinated power supply control unit increases the generation weight of the corresponding deformation-type noise compensation term and reduces the baseline update step size of the tactile main sensing area under high deformation.

[0151] During the self-calibration phase, the collaborative power supply control unit selects a non-core tactile sampling period and controls the auxiliary power supply conditioning unit to perform short-term load disturbance detection. During detection, the energy storage voltage and branch current of the energy storage branch before the disturbance are first recorded. Then, a known additional load is connected, and the energy storage voltage and branch current after the disturbance are recorded. The current equivalent internal resistance is determined based on the correlation between voltage drop and current increase. The current equivalent internal resistance is compared with the equivalent internal resistance benchmark recorded in the previous healthy state to obtain the change in energy storage equivalent internal resistance. If the change in equivalent internal resistance does not reach the internal resistance aging threshold, the original power supply path priority continues to be used; if the change in equivalent internal resistance reaches the internal resistance aging threshold, the priority of this energy storage branch in high-stability sampling power supply is reduced, and the access priority of the real-time energy pickup path or main power compensation path is increased. The internal resistance aging threshold is determined through energy storage device lifetime testing, and the internal resistance increment corresponding to the output voltage deviation reaching 3% of the reference voltage is taken as the aging boundary.

[0152] The collaborative power supply control unit updates the generation parameters of the noise compensation terms based on the tactile baseline drift, power supply disturbance transmission coefficient, deformation-to-tactile coupling coefficient, and energy storage equivalent internal resistance change. The tactile baseline drift is used to update the zero-load baseline and normalization starting point of the tactile channel; the power supply disturbance transmission coefficient is used to update the power supply-related noise compensation intensity corresponding to output ripple, power supply path switching, and energy storage charging / discharging states; the deformation-to-tactile coupling coefficient is used to update the deformation-related noise compensation intensity corresponding to bending, torsion, local tension, and deformation rate of change; and the energy storage equivalent internal resistance change is used to update the disturbance prediction and energy availability reduction ratio during power supply path switching. Parameter updates are performed using a limiting method, with each self-calibration updating the compensation parameters by no more than 10% of the factory-calibrated parameters. If three consecutive self-calibrations show changes in the same direction, updates are allowed to continue within a cumulative range not exceeding 30%. If this range is exceeded, the collaborative power supply control unit writes a maintenance reminder mark and does not further increase the parameter update range.

[0153] The weighting coefficients in the regional load, energy, and deformation correlation table are also updated based on the self-calibration results. If the tactile baseline drift increases, the impact of baseline drift on the reliability evaluation of tactile load is increased, causing areas with drift to be marked as areas requiring compensation when participating in tactile load prediction. If the power supply disturbance transmission coefficient increases, the influence of power supply-related disturbance weights on time slot allocation is increased, making energy pickup time slots more inclined to avoid tactile sampling time slots. If the coupling coefficient from deformation to tactile feedback increases, the influence of deformation disturbance values ​​on sampling frequency and baseline correction is increased, allowing the deformation reference area to obtain a higher sampling frequency. If the change in the equivalent internal resistance of energy storage increases, the contribution ratio of that energy storage branch to the available energy value is reduced, and the entry conditions for high-stability power supply levels are tightened. All of the above weight updates are performed within a preset limit range to avoid abrupt changes in the control strategy caused by a single abnormal calibration data.

[0154] The experimental group adopted the partitioned integration, dynamic power supply, time slot allocation, deformation feedback, and interference source compensation scheme described in this application; the control group adopted a tactile perception scheme without functional partitions, a fixed power supply path, and a fixed sampling frequency. Both groups used the same flexible substrate size, number of tactile channels, tactile sensor sensitivity, energy storage capacity, initial power supply voltage, grasping action sequence, and external load conditions. During the simulation, the dexterous hand was controlled to sequentially perform static contact, continuous bending, repetitive grasping, object sliding, and dynamic grasping actions, and the tactile zero-load baseline, power supply output ripple, remaining energy storage, tactile recognition results, and regional deformation data were recorded simultaneously. The tactile baseline drift was characterized by the difference in zero-load output before and after continuous operation; the power supply ripple interference was characterized by the amplitude of tactile signal fluctuation caused by the power supply output ripple within the tactile sampling frequency band; the system endurance was characterized by the time it took for the energy storage unit to drop from a fully charged state to a preset minimum operating voltage; and the dynamic grasping tactile recognition error was characterized by the difference between the recognition result and the preset contact state label.

[0155] Simulation results show that the normalized tactile baseline drift of the control group is 1.00, while that of the experimental group is 0.38, a reduction of 62% compared to the control group; the normalized power supply ripple interference amplitude of the control group is 1.00, while that of the experimental group is 0.29, a reduction of 71% compared to the control group; under the same energy storage capacity and motion load conditions, the normalized battery life of the control group is 1.00, while that of the experimental group is 1.45, representing a 45% improvement in system battery life; the normalized dynamic grasping tactile recognition error of the control group is 1.00, while that of the experimental group is 0.42, representing a 58% reduction in dynamic grasping tactile recognition error.

[0156] The results show that by binding the tactile channels, energy pickup units, and deformation feedback data to different regions using regional structure identifiers, the collaborative power supply control unit can adjust the auxiliary power supply level, power supply path, tactile sampling frequency, and the allocation ratio of various time slots according to the tactile load, energy availability, and deformation state of different regions. When the tactile load in the contact area increases, the power supply stability level and sampling priority of the tactile channels in that region are improved. When the energy availability decreases, the tactile channels used to obtain contact intensity and contact area in the contact area are continuously sampled, and the scanning frequency of the tactile channels in the non-contact area is reduced, thereby reducing unnecessary energy consumption.

[0157] Simultaneously, the signal decoupling transmission unit generates power supply noise compensation terms and deformation noise compensation terms based on the power supply path switching time, output ripple, energy storage charging and discharging status, as well as bending angle, torsion angle, local tensile amount, and deformation rate, reducing the coupling influence of power supply fluctuations and flexible substrate deformation on the tactile signal. Therefore, the system described in this application can reduce tactile baseline drift and power supply ripple interference under dynamic grasping and continuous deformation conditions, and extend the system operating time while maintaining the continuity of tactile sampling in the contact area.

[0158] It should be noted that if the power supply disturbance transmission coefficient increases, the collaborative power supply control unit increases the minimum isolation time between the tactile sampling time slot and the energy pickup time slot, and reduces the upper limit of their allowable overlap. If the change in the equivalent internal resistance of the energy storage increases, the collaborative power supply control unit increases the retention ratio of the energy pickup time slot or the main power supply compensation time slot to reduce the instantaneous discharge pressure of the aging energy storage branch during high-load sampling periods. If the tactile baseline drift increases but the power supply disturbance transmission coefficient does not increase, the disturbance assessment time slot and the baseline verification time slot are increased first, without expanding the energy pickup time slot. If the coupling coefficient from deformation to tactile sensation increases, the proportion of the deformation reference area sampling time slot is increased, and the tactile baseline writing during high deformation moments is limited. Through the above boundary adjustments, the time slot allocation can reflect the performance changes of the flexible substrate, power supply branch, and energy storage device after long-term use.

[0159] After the full lifecycle self-calibration is completed, the collaborative power supply control unit writes the updated tactile baseline, power supply disturbance transmission coefficient, deformation-to-tactile coupling coefficient, energy storage equivalent internal resistance reference, compensation parameter version number, and time slot boundary version number into the storage unit. At the start of the next control cycle, the signal decoupling transmission unit generates power supply noise compensation items and deformation noise compensation items according to the new compensation parameters, and the collaborative power supply control unit generates power supply control strategy and tactile sampling control strategy according to the new weighting coefficients and time slot boundaries. If creep of the flexible substrate, electrode aging, or energy storage performance degradation exceeding the maintenance threshold is detected during the self-calibration process, the collaborative power supply control unit maintains the safe power supply level and conservative sampling strategy, and outputs a maintenance reminder mark. The maintenance threshold is determined by the life test, specifically, after repeated bending of the flexible substrate, cyclic loading of the electrodes, and repeated charging and discharging of the energy storage device, the baseline drift, coupling coefficient change, and internal resistance change corresponding to the drop in tactile recognition accuracy to the allowable lower limit are recorded, and this state is used as the maintenance boundary.

[0160] After adopting the above-mentioned full lifecycle self-calibration process, the system can automatically update the noise compensation term generation parameters, regional load, weight coefficients in the energy and deformation correlation table, and the allocation boundary between energy pickup time slots and tactile sampling time slots, even when creep of the flexible substrate leads to deformation and enhanced tactile coupling, electrode aging leads to increased tactile baseline drift, and energy storage performance degradation leads to increased equivalent internal resistance. As a result, the power supply control strategy, tactile sampling control strategy, and dynamic correction of tactile data can maintain a matching relationship during long-term use, reducing tactile misjudgments and power supply instability caused by material aging, changes in adhesion status, and decreased energy storage capacity.

[0161] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A dexterous hand multi-source powered tactile sensing integrated structural system adapted to flexible electronic skin, characterized in that, include: An integrated flexible carrier is used to conformally fit the curved surface of a dexterous hand, and generates regional structure identifiers based on the curved structure parameters and contact condition parameters of different regions of the dexterous hand, so that the tactile sensing array and the multi-source energy pickup unit are integrated in the same flexible substrate according to the regional structure identifiers; The tactile sensing array is used to collect raw tactile signals from each tactile channel and generate tactile data frames carrying the regional structure identifier. The multi-source energy pickup unit is used to pick up electrical energy generated by at least two types of energy sources and generate energy input status data carrying the regional structure identifier; An auxiliary power supply conditioning unit is used to condition the electrical energy output by the multi-source energy pickup unit into auxiliary compensation power supply, and to generate an energy status frame and a power supply disturbance frame. The deformation feedback unit is used to acquire the deformation signal of the dexterous hand surface and generate a deformation state frame carrying the structural identifier of the region. The collaborative power supply control unit is used to match the tactile data frame, energy status frame, power supply disturbance frame and deformation status frame according to the regional structure identifier based on the same time reference, generate a regional load, energy and deformation association table, and generate a power supply control strategy and a tactile sampling control strategy according to the regional load, energy and deformation association table, so that the auxiliary compensation power supply parameters and tactile sampling parameters are adjusted in conjunction with each other according to the tactile workload and regional deformation status. The signal decoupling transmission unit is used to compensate and dynamically correct the power supply interference and deformation interference in the tactile data frame according to the power supply disturbance frame and the deformation state frame, and to feed back the corrected tactile data frame to the cooperative power supply control unit to update the power supply control strategy and the tactile sampling control strategy.

2. The dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin according to claim 1, characterized in that, The coordinated power supply control unit generates auxiliary power supply level, power supply path, tactile channel scanning sequence, tactile sampling frequency, and the allocation ratio of energy pickup time slot to tactile sampling time slot based on the regional load, energy, and deformation correlation table. When the predicted tactile load of any finger area increases, the power supply stability level and sampling priority of the tactile channel located in that finger area are increased before other finger areas. When the available energy value in the finger area is lower than the preset available energy threshold, the tactile channel in the contact area used to obtain contact intensity and contact area is continuously sampled, and the scanning frequency of the tactile channel in the non-contact area is reduced, or the energy pickup time slot is delayed.

3. The dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin according to claim 2, characterized in that, The regional structure identifier includes the region number, curvature level, contact probability level, bending stress level, and functional zoning type. The functional partition type includes at least a tactile main sensing area, an energy pickup area, a deformation reference area, and an isolation transition area, and the tactile data frame, energy status frame, power supply disturbance frame, and deformation status frame all carry a regional structure identifier for representing the data acquisition area.

4. The dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin according to claim 2, characterized in that, The main tactile sensing area is set on the fingertip, finger pad, or an area where the contact probability level reaches a preset high contact probability level. The energy pickup area is located on the side of the finger, the back of the finger, the outer bend of the joint, or in an area where the contact probability level is lower than the preset low contact probability level and the degree of contact wear is lower than the preset wear threshold. The deformation reference area is set in the joint transition area or in the area where the bending stress level reaches the preset stress level. The isolation transition zone is located between the main tactile sensing area and the energy pickup area to reduce tactile signal interference caused by power supply disturbances and mechanical deformations generated in the energy pickup area being transmitted to the main tactile sensing area.

5. The dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin according to claim 1, characterized in that, When the tactile sensing array generates tactile data frames, it groups the tactile channels according to the region structure identifier, reads the initial baseline of each tactile channel in the non-contact state, removes tactile channel data with noise amplitude exceeding the preset noise range, and normalizes the tactile channel data that has not been removed in order to extract contact intensity, contact area, number of activated channels, force change rate, force diffusion direction and sliding trigger frequency.

6. The integrated structure system for dexterous hand multi-source powered tactile sensing adapted to flexible electronic skin according to claim 1, characterized in that, When the auxiliary power supply conditioning unit generates the energy status frame and the power supply disturbance frame, it assigns a source identifier to each energy source to uniquely identify the energy source, collects the input voltage, input current and instantaneous power of each energy source, calculates the power fluctuation rate and input stability of each energy source, and combines the energy storage voltage, charging and discharging state, equivalent internal resistance, output voltage deviation, output ripple amplitude and power supply path switching state to generate the energy status frame and the power supply disturbance frame.

7. The dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin according to claim 1, characterized in that, When the deformation feedback unit generates a deformation state frame, it collects the bending angle, torsion angle, local stretching amount, and deformation rate of the region indicated by the regional structure identifier, and maps the collected bending angle, torsion angle, local stretching amount, and deformation rate of the region to the region indicated by the regional structure identifier. The deformation feedback unit is also used to extract the deformation baseline under the condition of no external contact, and to perform differential processing on the current deformation signal and the deformation baseline to obtain the deformation disturbance amount of the area indicated by the regional structure identifier.

8. The dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin according to claim 1, characterized in that, The signal decoupling transmission unit uses the output ripple, power supply path switching time, and energy storage charging and discharging state in the power supply disturbance frame as power supply interference features, and the bending angle, torsion angle, local stretching amount, and deformation rate in the deformation state frame as deformation interference features. It then calculates the power supply interference weight of the power supply interference features on the tactile data frame and the deformation interference weight of the deformation interference features on the tactile data frame, respectively, to generate power supply noise compensation terms and deformation noise compensation terms.

9. The dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin according to claim 8, characterized in that, The coordinated power supply control unit divides the same control cycle into energy pickup time slots, tactile sampling time slots, and disturbance assessment time slots, and updates the allocation ratio of each time slot according to the tactile load prediction value, energy availability value, power supply disturbance weight, and deformation disturbance weight. When the power supply interference weight increases to above the preset power supply interference weight threshold, the overlap ratio between the energy pickup time slot and the tactile sampling time slot is reduced, or the tactile sampling time slot is adjusted so that the tactile sampling time avoids the power supply path switching time. When the deformation interference weight increases to above the preset deformation interference weight threshold, the data acquisition frequency of the deformation reference area is increased, and the tactile baseline of the area indicated by the regional structure identifier carried by the deformation state frame is corrected according to the deformation state frame.

10. The dexterous hand multi-source powered tactile sensing integrated structure system adapted to flexible electronic skin according to claim 9, characterized in that, When generating the power supply control strategy and the tactile sampling control strategy, the collaborative power supply control unit performs the following processing steps: S1. Bind the tactile data frame, energy status frame, power supply disturbance frame and deformation status frame to the region according to the region structure identifier; S2. Based on the synchronization time tag, perform time alignment on various data frames with the same regional structure identifier; S3. Calculate the predicted tactile load value based on the tactile data frame, calculate the available energy value based on the energy state frame, and calculate the deformation interference value based on the deformation state frame. S4. Generate a regional load, energy, and deformation correlation table based on the tactile load prediction value, energy availability value, and deformation interference value; S5. Generate a power supply control strategy and a tactile sampling control strategy based on the aforementioned regional load, energy, and deformation correlation table. S6. Based on the source calculation results of power supply interference characteristics and deformation interference characteristics, perform closed-loop updates on the power supply control strategy and tactile sampling control strategy.

Images