A Smart Control Method Based on the Internet of Things

Abstract

This invention relates to the field of soft robotics and intelligent control technology, and particularly to an intelligent control method based on the Internet of Things (IoT). The method includes the following steps: multimodal data acquisition and preprocessing of the contact process between the gripper and the object to obtain a time-synchronized dataset; constructing a tactile feature tensor containing pressure, friction, and strain based on the time-synchronized dataset; calculating and mapping the stress gradient based on the tactile feature tensor to obtain a stress gradient field and a region threshold mapping table; adaptively adjusting the stress gradient threshold based on the region threshold mapping table and the stress gradient field to obtain an adaptive threshold distribution map; and predicting stress risk based on the adaptive threshold distribution map to obtain a stress risk map. This invention significantly improves the adaptability, robustness, and success rate of the gripping process by sensing the gripper's contact state and stress concentration to predict potential damage risks, and is particularly suitable for gripping fragile, irregular, and other difficult-to-predict objects.

Description

Technical Field

[0001] This invention relates to the field of soft robotics and intelligent control technology, and in particular to an intelligent control method based on the Internet of Things. Background Technology

[0002] Existing IoT-based intelligent control methods for soft robots suffer from severe deficiencies in force feedback accuracy and safety when handling fragile or irregular objects. Traditional soft grasping devices typically employ a single pressure sensor or a simple tactile array, failing to accurately perceive local stress distribution. This leads to excessive compression during grasping, causing irreversible damage to fragile objects. Furthermore, there is an irreconcilable contradiction between global stability and local safety. Traditional methods either overemphasize grasping stability while neglecting local stress safety, or overemphasize safety, resulting in grasping instability. The lack of an effective balancing mechanism fails to simultaneously meet these two critical requirements. Moreover, current soft grasping control generally lacks real-time adaptive adjustment capabilities. Mainstream technologies often rely on preset grasping modes or fixed pressure threshold control strategies, failing to dynamically adjust to complex and changing environments and object characteristics, resulting in high grasping failure rates and poor adaptability.

[0003] In summary, existing technologies suffer from problems such as a lack of precise perception in grasping fragile / irregular objects, difficulty in balancing stability and safety, and insufficient real-time adaptive capabilities, which urgently need to be addressed. Summary of the Invention

[0004] Therefore, it is necessary to provide an IoT-based intelligent control method to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, an intelligent control method based on the Internet of Things (IoT) includes the following steps:

[0006] Step S1: Perform multimodal data acquisition and preprocessing on the contact process between the soft gripper and the object to obtain a time-synchronized dataset; construct a tactile feature tensor containing pressure, friction, and strain based on the time-synchronized dataset;

[0007] Step S2: Calculate and map the stress gradient based on the tactile feature tensor to obtain the stress gradient field and the region threshold mapping table; adaptively adjust the stress gradient threshold based on the region threshold mapping table and the stress gradient field to obtain the adaptive threshold distribution map; predict the stress risk based on the adaptive threshold distribution map to obtain the stress risk map.

[0008] Step S3: Analyze the contact points based on the stress risk map to obtain a contact point feature table; calculate the anti-slip force based on the contact point feature table to obtain an anti-slip force requirement map; fuse local and global constraints based on the stress risk map and the anti-slip force requirement map to obtain an initial force distribution scheme; perform local and global force balance iterative optimization on the initial force distribution scheme to obtain a multi-constraint driving instruction set.

[0009] Step S4: Perform progressive drive execution on the multi-constraint drive instruction set to obtain drive execution status data; evaluate the deformation effect data of the gripper in real time; perform contact slip risk monitoring based on the tactile feature tensor to obtain contact status monitoring results; generate fine-tuning compensation instructions based on the deformation effect data and contact status monitoring results; drive the feedback loop in real time based on the drive execution status data and fine-tuning compensation instructions to achieve intelligent control of the gripper.

[0010] This invention utilizes high-density, multimodal tactile sensors to simultaneously acquire pressure, friction, and strain data, obtaining far more refined and comprehensive information on the internal and external mechanical states of the gripper than traditional grasping devices. Precise filtering, noise reduction, and time synchronization processing of these raw signals ensure data accuracy and usability. The resulting tactile feature tensor integrates the dispersed sensor data into a unified, high-dimensional representation, providing a rich and pre-processed reliable data foundation for subsequent stress analysis, risk assessment, and control decisions. This enables the system to perceive the complex and subtle interactions between objects and the gripper, a crucial prerequisite for the safe grasping of fragile and irregular objects.

[0011] By calculating stress gradients and identifying stress concentration areas, potential risk points of excessive stress on the gripper or object can be accurately located. Introducing historical data correlation and regionally differentiated threshold calculations means that safety threshold settings are no longer fixed, universal values, but are personalized based on the characteristics of different areas of the gripper and experience in gripping similar objects, improving the rationality and relevance of the settings. Furthermore, adaptive threshold adjustment based on real-time stress change rate allows the system to dynamically respond to uncertainties during the gripping process, increasing vigilance for rapidly changing risk areas and avoiding overly conservative approaches in stable areas, thereby maximizing gripping efficiency while ensuring safety. The final generated stress risk map intuitively and comprehensively displays the current stress risk level, stress change direction, and future trend predictions for each area of ​​the gripper, providing crucial safety guidance information for subsequent force planning and effectively preventing object damage during the gripping process.

[0012] When planning the gripping force, this approach breaks through the limitations of traditional methods that only consider global stability. By precisely calculating the anti-slip force requirements, the overall robustness of the gripping process is ensured. More importantly, it creatively integrates local safety constraints (preventing object damage) in the stress risk map with global stability constraints (preventing slip instability) in the anti-slip force requirement map into a unified optimization framework. Employing a local-global force balance iterative optimization method, by alternately optimizing the force distribution in different regions, it effectively balances the often contradictory goals of global gripping stability and local stress safety, finding an optimal force distribution scheme that can stably grip without damaging the object. The final multi-constraint drive instruction set directly reflects this optimization result, providing a scientific and safe mechanical foundation for subsequent precise control.

[0013] This system achieves refined and dynamic control of the soft gripper. The gripper is divided into multiple independent control regions and driven progressively, avoiding local overload or underload caused by traditional unidirectional drive. Through coordination of inter-region variation constraints and risk-sequential execution, the smoothness and stability of the drive process are ensured. Real-time monitoring of gripper deformation and contact slip risk monitoring using multimodal tactile data, particularly sensitive detection of micro-slippage and pressure anomalies, provides critical status feedback during the gripping process. Based on these real-time monitoring results, the system can promptly detect deviations between the actual state and the expected state, as well as potential instability or damage risks, and quickly calculate and generate fine-tuning compensation commands. The final real-time drive feedback loop enables the system to dynamically adjust drive commands according to the real-time gripping state, forming a closed-loop control system. This significantly improves the adaptability, robustness, and success rate of the gripping process, making it particularly suitable for gripping fragile, irregular, and other unpredictable objects.

[0014] Therefore, this invention provides an intelligent control method based on the Internet of Things, which achieves precise perception of the contact state between the object and the gripper through multimodal tactile sensing, and predicts potential damage risks by combining stress concentration early warning analysis; it also adopts local-global force balance iterative optimization to simultaneously consider gripping stability and local stress safety during the force planning stage; and finally achieves dynamic adaptive adjustment of the gripping process through zoned differentiated drive control and real-time feedback loop, thereby effectively solving the above-mentioned drawbacks of existing technologies in gripping fragile and irregular objects. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the steps of an IoT-based intelligent control method.

[0016] Figure 2 This is a detailed flowchart illustrating the implementation steps of step S1 in this invention.

[0017] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0019] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0020] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] In this embodiment of the invention, reference 2 is shown as a flowchart illustrating the steps of the IoT-based intelligent control method of the present invention. In this example, the IoT-based intelligent control method includes the following steps:

[0022] Step S1: Perform multimodal data acquisition and preprocessing on the contact process between the soft gripper and the object to obtain a time-synchronized dataset; construct a tactile feature tensor containing pressure, friction, and strain based on the time-synchronized dataset;

[0023] In this embodiment of the invention, an 8×8 flexible pressure sensor grid and a matching friction sensor are embedded on the inner surface of the soft gripper, and 12 stretchable strain sensors are distributed at key internal locations. During object grasping, these sensors synchronously acquire resistance or voltage signals at a frequency of 200Hz, forming a raw sensor signal dataset. A low-pass filter with a 20Hz cutoff frequency is applied to the pressure signal, a median filter is used to remove spikes from the friction signal, and a Kalman filter is used to eliminate drift in the strain signal, resulting in a filtered signal dataset. Based on sensor timestamps and response delay compensation (pressure 5ms, friction 8ms, strain 12ms), the filtered data is resampled to a uniform 5ms time series, forming a time-synchronized dataset. Using the time-synchronized dataset, a 64×64 contact pressure distribution map is constructed by bilinear interpolation of the 8×8 pressure data, a two-dimensional friction vector field is constructed from the friction data, and the local friction coefficient is calculated. The three-dimensional strain field inside the soft gripper is reconstructed using radial basis function interpolation based on the data from the 12 strain nodes. The pressure distribution (two-dimensional), friction vector field (two-dimensional), and strain field (three-dimensional) are organized into a three-dimensional structure according to their spatial correspondence. Then, the time series is stacked to finally construct a four-dimensional tactile feature tensor [space_x×space_y×space_z×time]. Each spatial point contains pressure, friction vector, and strain tensor components, which fully characterize the interaction between the gripper and the contact interface of the object and the deformation state inside the gripper.

[0024] Step S2: Calculate and map the stress gradient based on the tactile feature tensor to obtain the stress gradient field and the region threshold mapping table; adaptively adjust the stress gradient threshold based on the region threshold mapping table and the stress gradient field to obtain the adaptive threshold distribution map; predict the stress risk based on the adaptive threshold distribution map to obtain the stress risk map.

[0025] In this embodiment of the invention, pressure, friction, and strain tensors are extracted from the tactile feature tensor slice at the current moment. Equivalent stress is calculated to form a tactile feature mapping table, identifying regions with stress exceeding 0.3 MPa. Three-dimensional central difference is applied to the tactile feature mapping table to calculate the partial derivatives of stress in the x, y, and z directions, synthesizing a stress gradient vector field. A 3×3×3 adaptive convolution is applied to the stress gradient field to detect gradient abrupt changes, marking regions with gradients exceeding 0.5 MPa / mm as stress concentration areas, generating a stress concentration region marking map. Based on the stress concentration region marking map, similar objects and grasping records are searched in the historical database, calculating historical case similarity, extracting stress concentration and damage association rules, and obtaining a historical association feature set. Combining the historical association feature set, gripper region sensitivity classification (high, medium, and low sensitivity areas), and material properties, personalized stress thresholds are calculated for each region of the gripper, forming a region threshold mapping table. Stress or stress gradient change rate is monitored in real time (e.g., based on data from the last 5 time points), identifying rapidly changing regions with change rates exceeding 0.05 MPa / ms. Based on the regional threshold mapping table, the threshold for rapidly changing regions is dynamically reduced according to the formula "new threshold = baseline threshold × (1 - α × rate of change - normalization)," where α is determined based on regional sensitivity. The adjusted threshold is then subjected to three-dimensional Gaussian smoothing to ensure spatial continuity, generating an adaptive threshold distribution map. Based on the adaptive threshold distribution map, the ratio of each stress value to the adaptive threshold is calculated as the stress exceedance risk index (0-1). Based on the tactile feature tensor time series, a linear prediction algorithm is used to predict the stress change rate and peak time for the next 100 ms, generating a stress evolution prediction map. The stress exceedance risk index distribution, stress gradient field, and stress evolution prediction map are integrated to construct a three-dimensional stress risk map. Each point contains the risk index, gradient direction, and predicted rate of change, and is divided into green, yellow, and red regions according to the risk index.

[0026] Step S3: Analyze the contact points based on the stress risk map to obtain a contact point feature table; calculate the anti-slip force based on the contact point feature table to obtain an anti-slip force requirement map; fuse local and global constraints based on the stress risk map and the anti-slip force requirement map to obtain an initial force distribution scheme; perform local and global force balance iterative optimization on the initial force distribution scheme to obtain a multi-constraint driving instruction set.

[0027] In this embodiment of the invention, based on the stress risk map and tactile feature tensor, the effective contact area is identified, the contact area and stress state are calculated, and the spatial locations of key contact points such as the pressure center are determined, forming a contact point feature table. Material and weight characteristics of the object are obtained. Combining the contact point feature table and object characteristics, a velocity-dependent friction model is used. Calculate the velocity-corrected friction coefficient. Considering surface roughness Ra, use μ_effective= Calculate the effective friction coefficient. Based on the effective friction coefficient, local pressure, and tangential force, divide the contact area into an adhesion zone and a potential slip zone, forming contact state partitioning data. Based on the contact state partitioning data, calculate the minimum total normal force required to satisfy the overall anti-slip condition, and optimize its distribution to each region according to the contact area, local curvature, and contact state, forming a normal force distribution scheme. This scheme, together with the contact state partitioning data, constructs an anti-slip force demand map. Based on the anti-slip force demand map and the contact point characteristic table, establish the overall force balance and moment balance equations and friction cone constraints, constructing global stability constraints. Extract high-risk regions with a risk index > 0.8 from the stress risk map, set the maximum allowable stress σ_max_i according to its adaptive threshold, and transform it into constraints on the normal force P_i and tangential force F_friction_i of that region (equivalent stress ≤ σ_max_i), constructing local safety constraints. Merge the global stability constraints and local safety constraints, and solve using a quadratic programming method to obtain an initial force distribution scheme that satisfies all constraints. The gripper area is divided into core gripping blocks, safety-sensitive blocks, and transition support blocks based on stress risk maps and contact patterns, forming a regional block table. A composite objective function J(f) = λG(f) + (1-λ)L(f) is constructed, where G(f) minimizes the deviation from the anti-slip requirement map, L(f) penalizes high-stress areas, and the weight w_i is determined based on the regional sensitivity. The initial force distribution scheme is used as the starting point for iteration, with a maximum number of iterations of 50 and a convergence threshold of 0.01. During iteration, the forces of other blocks are alternately fixed, and the gradient descent method is used to optimize the force of the current block (first the core block, then the safety-sensitive block, and finally the transition support block), while λ is dynamically adjusted to balance G and L. After iteration convergence, the final force distribution scheme is converted into target air pressure / voltage values ​​for each drive unit. Combined with the execution sequence, timing, monitoring threshold, and emergency plan, a multi-constraint drive instruction set is generated.

[0028] Step S4: Perform progressive drive execution on the multi-constraint drive instruction set to obtain drive execution status data; evaluate the deformation effect data of the gripper in real time; perform contact slip risk monitoring based on the tactile feature tensor to obtain contact status monitoring results; generate fine-tuning compensation instructions based on the deformation effect data and contact status monitoring results; drive the feedback loop in real time based on the drive execution status data and fine-tuning compensation instructions to achieve intelligent control of the gripper.

[0029] In this embodiment of the invention, the soft gripper is divided into 8 independent control regions based on its physical structure and drive units. These regions are numbered and a topological relationship graph (adjacency matrix) is constructed to form a region division scheme. A multi-constraint drive instruction set is parsed, and the instructions of each drive unit are distributed and mapped to the corresponding region, generating a drive execution parameter table containing target air pressure / voltage values. Based on the target value change amplitude ΔP_i in the drive execution parameter table and the stress risk map (regions with a risk index > 0.7 have an increased number of segments), the number of drive segments in each region is dynamically determined. A regional segmentation table is generated. For each region, nonlinear parameter increments are designed according to the segmentation number (small increments in the initial and final stages, large increments in the middle stage), forming a segmented parameter table. Based on the regional topology and the segmented parameter table, the parameter variation differences between adjacent regions are constrained to not exceed a threshold (e.g., first-level adjacent difference ≤ C_1), and a "lead-follow" mode is introduced to obtain a coordinated segmented parameter table. Regional risks are ranked according to the stress risk map, generating an execution order list. Combining the coordinated segmented parameter table and the execution order list, a sequence of target parameter values ​​for each region at each time step (e.g., 50ms interval) is generated, forming a phased execution plan. The underlying drive control module sends control signals to the drive units of each region according to the plan, while simultaneously monitoring the actual air pressure / voltage values ​​and the actual deformation (e.g., bending angle) calculated from strain sensor data, recording this as drive execution status data. Using the tactile feature tensor (updated in real time), the current pressure distribution and friction vector field are extracted to construct contact feature data. Contact dynamic characteristics such as pressure center, contact area, and key point pressure friction are calculated based on the contact feature data. Based on contact dynamic characteristics, the ratio of local friction coefficient to effective friction coefficient is calculated, or changes in friction force direction are monitored to detect micro-slippage signs and generate a slippage risk assessment map. Areas where pressure values ​​deviate from expectations are identified, forming a pressure anomaly area map. Combining the pressure anomaly area map and the slippage risk assessment map, the local stability risk of each contact area is assessed, generating a local stability assessment map. Based on the slippage risk assessment map and contact characteristic data, global stability indicators such as gripping torque margin and pressure center fluctuation are calculated. Integrating the local stability assessment map and global stability indicators, contact state monitoring results (overall stable state, unstable areas, potential slippage direction) are generated. By comparing the actual deformation with the expected deformation deviation in the drive execution state data, and combining the local unstable areas and slippage risks in the contact state monitoring results, the required drive parameter fine-tuning amount is calculated (e.g., increasing air pressure in a certain area to improve gripping force, or adjusting torque to correct attitude), generating fine-tuning compensation commands. By integrating drive execution status data, fine-tuning compensation instructions, deformation effect data, and contact status monitoring results into a real-time data stream, a closed-loop feedback mechanism is constructed. The execution status is monitored at a frequency of 20ms, deformation and contact are evaluated at 50ms, and fine-tuning instructions are generated and executed at 100ms, thereby achieving dynamic adaptive intelligent control of the gripper's grasping process.

[0030] As an example of the present invention, reference is made to Figure 2 As shown, in this example, step S1 includes:

[0031] Step S11: The sensor array installed on the soft gripper collects signals during the contact process with the object to obtain raw sensor signals, which include pressure signals, friction signals and strain signals.

[0032] In this embodiment of the invention, the high-density flexible pressure sensor array adopts an 8×8 grid arrangement. Each sensor unit converts the normal contact pressure into a resistance change based on the piezoresistive effect, and the resistance change is acquired and digitized by an integrated circuit. A micro-friction force sensor is installed in conjunction with each pressure sensor location, employing multi-axis force sensor technology. By measuring the minute deformation of the cantilever beam or elastic body, the tangential contact force is decomposed into two orthogonal components (e.g., along the x and y directions of the gripper surface) and converted into a voltage signal for digitization. A stretchable strain sensor network is distributed with 12 nodes along the key deformation areas of the soft gripper (such as joint bending areas and fingertip stretching areas). It uses stretchable conductive materials based on carbon nanotubes or liquid metals, whose resistance changes with the deformation caused by material stretching or compression. The resistance change is detected by a Wheatstone bridge circuit and converted into a voltage signal for digitization. All sensor signals are synchronously acquired by a high-speed data acquisition module with a sampling frequency set to 200Hz, generating a dataset of raw pressure, friction, and strain signals including timestamps.

[0033] Step S12: Perform signal filtering and noise cancellation on the original sensor signal to obtain the filtered signal dataset;

[0034] In this embodiment of the invention, the acquired raw pressure signal contains high-frequency noise caused by mechanical vibration or power fluctuations. A digital low-pass filter (e.g., implemented using a Butterworth filter algorithm) is used to process the pressure signal with a cutoff frequency of 20Hz to retain the effective contact force signal and remove high-frequency interference. The raw friction force signal contains instantaneous spikes or glitches. A median filtering algorithm is used to process the friction force signal. This algorithm smooths the signal by taking the median of the data within a sliding window, effectively eliminating abnormal spike interference. The raw strain signal is affected by temperature drift or long-term creep. A Kalman filter algorithm is used to process the strain signal. The Kalman filter performs optimal estimation of the signal by combining sensor measurements and a system dynamic model, effectively reducing random noise and compensating for signal drift. These filtering algorithms are implemented on an embedded processing unit to process the raw sensor signals in real-time or near real-time.

[0035] Step S13: Perform sensor data time synchronization on the filtered signal dataset to obtain a time-synchronized dataset;

[0036] In this embodiment of the invention, each set of sensor data acquired in step S11 is accompanied by a precise timestamp. Although the sensor acquisition is synchronized, different sensor types have inherent response delays. A 5ms time compensation is applied to the filtered pressure signal data (shifting the data points forward 5ms on the time axis), an 8ms time compensation is applied to the friction signal data, and a 12ms time compensation is applied to the strain signal data to calibrate their response delays. Although the compensated data streams have the same initial sampling frequency, the time points are not perfectly aligned due to compensation and subtle acquisition differences. Therefore, all data are resampled into a unified time series, for example, one data point every 5ms. A linear interpolation method is used to calculate the data values ​​at non-original sampling time points, ensuring that all sensor data are perfectly aligned in the time dimension, forming a time-synchronized dataset.

[0037] Step S14: Construct the contact pressure distribution map, contact friction vector field, and internal strain field of the soft body based on the time synchronization dataset;

[0038] In this embodiment of the invention, pressure sensor data from a time-synchronized dataset is used to map discrete pressure values ​​from an 8×8 grid onto a two-dimensional planar coordinate system representing the contact surface of the soft gripper. To obtain more refined distribution information, bilinear interpolation is used to perform interpolation calculations between sensor points, generating a 64×64 resolution continuous pressure distribution heatmap that visually represents the normal pressure distribution at the contact interface. Using friction sensor data from the time-synchronized dataset, the tangential force components in the x and y directions collected at each sensor location are used to construct the friction vector at that location, forming a sparse friction vector field. Simultaneously, at each sensor location, the friction coefficient μ_i = |F_friction_i| / P_i is calculated, where |F_friction_i| is the magnitude of the friction vector at that location, and P_i is the corresponding pressure value. Using data from 12 strain sensor nodes in the time-synchronized dataset, the radial basis function (RBF) interpolation method is used to reconstruct the three-dimensional continuous strain field inside the soft gripper based on these 12 nodes. RBF interpolation approximates strain values ​​by constructing a linear combination of distance-based kernel functions. Its coefficients are determined by fitting known sensor data using the least squares method, thereby predicting the strain state at any point inside the gripper and further calculating the principal strain direction and strain energy density distribution.

[0039] Step S15: Organize the spatial positional relationship of the contact pressure distribution map, the contact friction vector field, and the internal strain field of the soft body to obtain the tactile feature tensor.

[0040] In this embodiment of the invention, the contact pressure distribution map (two-dimensional data), contact friction vector field (two-dimensional vector data, which can be regarded as two two-dimensional component maps) and internal strain field of the soft body (three-dimensional data, containing multiple components of the strain tensor) generated in step S14 are spatially aligned and integrated. On a predefined three-dimensional grid representing the space of the soft gripper, for each grid cell or key spatial point, the pressure value, friction two-dimensional vector information, and strain three-dimensional tensor component values ​​extracted from the reconstructed strain field at that point are organized together. For example, the data of a spatial point includes: pressure value, friction x component, friction y component, strain tensor xx component, yy component, zz component, xy component, yz component, and zx component. These spatial point data are collected to form a three-dimensional data structure, representing the complete mechanical state of the soft gripper at a certain moment. By stacking these three-dimensional data structures in a time series, a four-dimensional data structure, namely the tactile feature tensor, is obtained, whose dimensions are [space_x×space_y×space_z×time], containing complete mechanical state information of the contact interface between the object and the gripper and the interior of the gripper as it changes over time.

[0041] Preferably, step S2, which involves calculating and mapping the stress gradient based on the tactile feature tensor, includes:

[0042] Extract the tactile feature mapping table from the tactile feature tensor;

[0043] Calculate the stress gradient field based on the tactile feature mapping table;

[0044] Stress concentration regions are identified in the stress gradient field to obtain a stress concentration region marking map;

[0045] Historical data correlation analysis was performed on the stress concentration area marker map to obtain a historical correlation feature set;

[0046] Based on the historical correlation feature set and stress concentration area labeling map, the regional differential threshold is calculated to obtain the regional threshold mapping table.

[0047] In this embodiment of the invention, a slice of the tactile feature tensor at the current moment is selected. This is a three-dimensional data structure [space_x×space_y×space_z], where each spatial point stores the pressure value, friction vector, and strain tensor. For subsequent analysis, key mechanical parameters are extracted from this three-dimensional data to construct a tactile feature mapping table. For the gripper surface area, the contact pressure value P and the friction vector F_friction are mainly extracted, and the resultant force F_total is calculated. For the internal region of the gripper, the components of the strain tensor ε are mainly extracted, and the equivalent stress σ_e is calculated, for example using the VonMises equivalent stress criterion:

[0048] σ_e= ;

[0049] Where σ is the stress tensor, and its components σ_ij and strain tensor ε_ij are calculated using the constitutive relations of the material (e.g., Hooke's law for linear elastic materials: σ_ij = C_ijklε_kl, where C is the tensor of the material's elastic constants). These calculated equivalent stress values ​​or other key parameters (such as the maximum principal stress) are stored in a three-dimensional mesh data corresponding to the spatial structure of the soft gripper, forming a tactile feature mapping table. Simultaneously, regions in the tactile feature mapping table where the stress amplitude exceeds a preset reference value (e.g., for soft silicone materials, the reference value is set to 0.3 MPa) are identified as initial regions of interest.

[0050] The stress gradient represents the rate of change of stress in space and is a key indicator for identifying stress concentration areas. For each micro-region (grid point) in the tactile feature mapping table (a three-dimensional stress value distribution) obtained in the previous step, the change in its stress value relative to its neighboring micro-regions is calculated. The partial derivatives of the stress in the three orthogonal directions (x, y, z) are calculated using the three-dimensional central difference method. Similar calculations and Δx, Δy, and Δz are the grid step sizes. The partial derivatives in these three directions are used as components to construct the stress gradient vector for each micro-region. These vector sets form the stress gradient field, representing the direction and magnitude of the steepest stress change, and the gradient magnitude. .

[0051] Stress concentration regions are characterized by significantly higher stress gradient amplitudes than surrounding areas. A 3D adaptive convolution algorithm is applied to detect abrupt changes in the stress gradient field. A 3×3×3 3D convolution kernel slides across the stress gradient field to calculate the local average gradient or gradient variance. Adaptability is reflected in the dynamic adjustment of the kernel weights or size based on local gradient characteristics; for example, a smaller kernel can be used to focus details in regions of drastic gradient changes. An initial gradient threshold (e.g., 0.5 MPa / mm) is set to control the stress gradient amplitude. Regions exceeding this threshold are marked as potential stress concentration areas. These marked regions form a three-dimensional binary map, namely the stress concentration area marking map, where the value of the marked regions is 1 and the value of the unmarked regions is 0.

[0052] Establish a historical grasping database, storing tactile feature tensors, stress risk maps, drive command sets, and final grasping results (success, failure, object damage type, and location) from previous grasping processes of various objects (especially fragile and irregular objects). For the currently identified stress concentration area marker map, search the historical database for historical records with similar grasping object types, similar initial contact postures, or similar stress distribution patterns. Calculate the similarity score between the current stress distribution and historical successful and failed cases, for example, using methods based on image features (such as Hu moments, Zernike moments) or signal correlation (such as Pearson correlation coefficient). Extract the historical cases that are most strongly associated with the current stress concentration area in terms of spatial location and stress characteristics. Analyze the stress evolution process of the corresponding area in these historical cases, the set safety thresholds, and the final grasping results (whether damage occurred, damage type, and location). These highly correlated historical case information, similarity scores, and correlation rules between stress concentration areas extracted from historical cases and the capture results (e.g., stress gradients higher than X MPa / mm in this area caused damage to objects in Y% of historical failure cases) are integrated into a historical correlation feature set.

[0053] Based on the potential risk areas identified by the stress concentration region marker map and combined with information extracted from historical correlation feature sets, personalized stress thresholds are calculated for different regions of the soft gripper. For regions exhibiting high damage risk in historical data (e.g., stress concentration accompanied by object damage in multiple gripping failure cases), a lower safe stress threshold is set for that region. For relatively safe regions in historical data, a more lenient threshold can be set. The threshold calculation also considers the properties of the soft gripper material itself (e.g., yield strength, fracture strain) and the geometry of different regions of the gripper (e.g., weak points, stress concentration design points). For example, for a certain region in historical data, its historical safe stress upper limit is P_hist. Combining the safety factor SF of the gripper material (e.g., SF=0.8), the baseline threshold for that region is set as P_hist×SF. These region-specific thresholds constitute a three-dimensional data structure corresponding to the spatial structure of the soft gripper, namely a region threshold mapping table, where each micro-region has a specific stress threshold.

[0054] Preferably, step S2, which involves adaptively adjusting the stress gradient threshold based on the region threshold mapping table and the stress gradient field, includes:

[0055] The software grippers are classified by region sensitivity, resulting in a region sensitivity classification table;

[0056] Obtain and calculate the similarity of the currently crawled object based on historical crawling data to obtain a set of similar cases;

[0057] By assigning time weights to similar case sets, a weighted case set is obtained.

[0058] The baseline threshold is calculated for the weighted case set to obtain the regional baseline threshold table;

[0059] Stress change rate is monitored based on stress gradient field and tactile feature tensor to obtain stress change rate map;

[0060] The stress change rate map is used to identify the rate-sensitive region to obtain a rapid change region marking map;

[0061] Based on the regional threshold mapping table, the threshold of the rapidly changing area marker map is dynamically adjusted to obtain a preliminary adjusted threshold table;

[0062] Spatial smoothing is performed on the preliminary threshold table to obtain a smoothed threshold table;

[0063] An adaptive threshold distribution map is generated based on the smooth threshold table.

[0064] In this embodiment of the invention, the surface and internal space of the soft gripper are divided into regions of different sensitivities based on the material properties, geometric structure, and design functions of the gripper. For example, regions with thinner material thickness, the highest density of embedded sensors, or those designed to withstand greater strain are classified as high-sensitivity regions; regions with moderate material thickness, average sensor density, or those primarily serving a supporting function are classified as medium-sensitivity regions; and regions with thicker material thickness, lower sensor density, or those primarily serving a connecting function are classified as low-sensitivity regions. This classification is determined based on prior knowledge of the gripper and structural analysis. Each spatial micro-region is labeled with its corresponding sensitivity category (e.g., using integers 1, 2, and 3 to represent low, medium, and high sensitivity, respectively), forming a three-dimensional data structure corresponding to the spatial structure of the gripper, i.e., a region sensitivity classification table.

[0065] Retrieve all stored grasping records from the historical grasping database. For the currently grasping object, acquire its 3D shape information using an external vision sensor (e.g., a depth camera integrated with the grasper). Calculate the similarity between the shape features of the current object and the shape features of objects already grasped in the historical database, for example, by matching using feature descriptors such as shape context and surface normal histograms. Simultaneously, consider information such as grasping task type (e.g., grasping, placing, transporting) and object material properties (e.g., preliminary assessment of hardness and surface roughness using sensors) for auxiliary matching. Select historical grasping records with similarity scores higher than a preset threshold (e.g., shape similarity higher than 0.7, task type matching) to form a similarity case set. The similarity case set contains historical data that is comparable to the current grasping situation.

[0066] The historical records in the similar case set have different occurrence times. To ensure that recent cases, which better reflect the current system state and environmental changes, have a greater impact on the threshold calculation, a time weight is assigned to each historical case in the similar case set. The time weight calculation formula uses an exponential decay model: W_time = , where t is the difference between the time the case occurred and the current time (time interval), and T is the time decay constant (for example, setting T to 100 hours means that the weight of a case 100 hours ago decays to approximately 37%). The shorter the time interval, the higher the weight. Each similar case is associated with its calculated time weight to form a weighted case set.

[0067] Based on historical successful crawl records in the weighted case set, a baseline stress gradient threshold is calculated for each region of the software gripper. For each region, the historical distribution of stress gradients appearing in that region among the weighted successful cases is statistically analyzed. The baseline threshold can be set as a statistic of the stress gradients in that region from the weighted historical data, such as the weighted average plus a safety margin, or the 95th quantile of the weighted historical stress gradient distribution. For example, the baseline threshold for region i... ,in is the stress gradient magnitude of region i in the j-th weighted historical case, and SafetyOffset_i is the safety offset set for region i, which can be determined according to the region sensitivity classification table (the offset is larger for highly sensitive areas). These calculated benchmark thresholds constitute the region benchmark threshold table, providing a basis for subsequent real-time adjustments.

[0068] During the grasping process, the current stress gradient field and tactile feature tensor (containing time-series data) are acquired in real time. The rate of change of each stress value (extracted from the tactile feature tensor) or stress gradient amplitude (extracted from the stress gradient field) over time is calculated. For example, the rate of change of stress is calculated using the finite difference method: ΔP / Δt≈(P(t)-P(t-Δt)) / Δt, where P(t) is the stress value at the current time t, P(t-Δt) is the stress value at the previous time t-Δt, and Δt is the sampling time interval. The rate of change of stress gradient amplitude is calculated similarly. These calculated stress or stress gradient rate of change values ​​are stored in a three-dimensional mesh data corresponding to the grasper's spatial structure, forming a stress rate of change map.

[0069] In the stress rate of change map, regions where the rate of change of stress or stress gradient exceeds a preset threshold (e.g., stress rate of change > 0.05 MPa / ms or stress gradient rate of change > 0.1 MPa / mm / ms) are identified. These regions indicate that the stress state is changing rapidly and require special attention. These rapidly changing regions are marked in three-dimensional space to form a three-dimensional binary map, namely the rapid change region marking map.

[0070] For the marked areas in the rapidly changing zone marker map, the stress gradient threshold corresponding to the area in the zone threshold mapping table is dynamically adjusted based on the zone sensitivity classification and the current stress change rate. The threshold adjustment formula based on the stress change rate is used: New Threshold = Baseline Threshold × (1 - α × Change Rate Normalization). Here, the baseline threshold is obtained from the zone baseline threshold table, the change rate normalization normalizes the stress or stress gradient change rate of the zone to between 0 and 1, and α is the sensitivity coefficient, whose value is determined according to the zone sensitivity classification table (a larger α value for high-sensitivity zones, e.g., 0.3; a smaller α value for low-sensitivity zones, e.g., 0.1). The higher the stress change rate, the greater the decrease in the new threshold relative to the baseline threshold, making the system more sensitive to stress concentration in that zone. For non-rapidly changing zones, the threshold remains unchanged or undergoes minor adjustments. These adjusted thresholds constitute the preliminary adjusted threshold table.

[0071] The initial threshold adjustment table suffers from excessively large threshold differences between adjacent regions due to drastic local changes or sensor noise, leading to control instability. To ensure a smooth spatial transition of thresholds and avoid abrupt changes, the initial threshold adjustment table undergoes spatial smoothing. A three-dimensional Gaussian filter or moving average filter is used to weighted average the threshold of each micro-region with the thresholds of its neighboring micro-regions, ensuring that the threshold difference between adjacent regions does not exceed a preset maximum allowable difference (e.g., the threshold difference between adjacent micro-regions does not exceed 30%). The processed threshold distribution is more gradual, forming a smoothed threshold table.

[0072] The smoothing threshold table contains the stress gradient thresholds for each micro-region of the soft gripper at the current gripping moment. These thresholds have been dynamically adjusted and spatially smoothed based on historical data, regional sensitivity, and real-time stress change rate. The smoothing threshold table is stored as a standard three-dimensional data structure, namely an adaptive threshold distribution map. This map provides a real-time, regionally differentiated safety reference standard for subsequent stress risk assessment.

[0073] Preferably, step S2, which involves predicting stress risk based on an adaptive threshold distribution map, includes:

[0074] Stress risk assessment is performed based on the adaptive threshold distribution map to obtain the stress risk index distribution;

[0075] Stress evolution trend is predicted based on the stress risk index distribution and tactile feature tensor, resulting in a stress evolution prediction map;

[0076] A stress risk map is constructed based on the stress risk index distribution, stress gradient field, and stress evolution prediction map.

[0077] In this embodiment of the invention, for each micro-region in the soft gripper space, its current actual stress value (e.g., equivalent stress value extracted from a tactile feature mapping table) and corresponding adaptive stress threshold (obtained from an adaptive threshold distribution map) are acquired. The stress over-limit risk index R_i for each micro-region is calculated. The risk index is a value between 0 and 1, representing the degree of stress tension of the current stress value relative to its threshold. The calculation formula is: R_i = σ_i / T_adaptive_i, where σ_i is the actual stress value of micro-region i, and T_adaptive_i is the adaptive stress threshold of micro-region i. To limit the index to the range of 0-1 and highlight high risk, a sigmoid function can be used for mapping, for example, R_i = , where k is the gain coefficient used to adjust the steepness of the curve. When σ_i is much smaller than T_adaptive_i, R_i approaches 0; when σ_i approaches or exceeds T_adaptive_i, R_i rapidly approaches 1. The risk index value of each micro-region is stored in a three-dimensional data structure corresponding to the gripper's spatial structure, forming a stress risk index distribution.

[0078] Using time-series data of the stress risk index distribution or raw stress values ​​at the most recent 5 time points in the tactile feature tensor, the stress change trend of each microregion over a future period is predicted. For each microregion, the time-series data of its stress risk index (or raw stress value) is extracted. This time-series data is fitted using a linear prediction algorithm (e.g., a least squares-based linear regression model) or a more complex time-series prediction model (e.g., an autoregressive moving average model ARMA or a long short-term memory network LSTM; linear prediction is used here as a simple example) to predict the rate of stress change (e.g., MPa / ms) of the microregion within the next 100 ms and the predicted time point when the peak stress may occur. For example, for microregion i, its most recent stress value sequence is σ_i(t-4Δt),...,σ_i(t). This sequence is fitted by linear regression to obtain the linear model σ_i(t). )=a* +b, where This represents a future time relative to t. The predicted stress change rate is a. The predicted stress change rate and predicted peak time for each micro-region are stored in a three-dimensional data structure corresponding to the gripper's spatial structure, forming a stress evolution prediction map.

[0079] By integrating the stress risk index distribution (representing the current risk level), stress gradient field (representing the severity and direction of stress changes), and stress evolution prediction map (representing future risk trends) obtained from the stress exceedance risk assessment, a comprehensive stress risk map is constructed. The stress risk map is a three-dimensional data structure where each spatial micro-region contains the following information: 1) a stress exceedance risk index (a value between 0 and 1), representing the risk level of the current stress relative to the adaptive threshold; 2) a stress gradient direction vector (a three-dimensional vector), representing the direction of the fastest current stress change; and 3) a predicted stress change rate (a signed value), indicating whether the future stress will increase or decrease and the rate of increase / decrease. To visually represent the risk level, the micro-regions can be divided into different levels based on the stress exceedance risk index. For example, a risk index less than 0.4 is a safe zone (marked in green), 0.4 to 0.8 is a warning zone (marked in yellow), and greater than 0.8 is a danger zone (marked in red). The stress risk map provides comprehensive and real-time stress safety information for subsequent mechanical stability constraint optimization and drive control.

[0080] Preferably, the calculation of anti-slip force based on the contact point characteristic table in step S3 includes:

[0081] Obtain the object's characteristic parameters, combine them with the contact point feature table to calculate the velocity-dependent friction coefficient of the contact point, and obtain the velocity-corrected friction coefficient.

[0082] Based on the speed-corrected friction coefficient and the contact point characteristic table, the surface roughness effect analysis is performed to obtain the effective friction coefficient;

[0083] Contact area state analysis is performed based on the effective friction coefficient and object characteristic parameters to obtain contact state partitioning data.

[0084] Calculate the minimum normal force value based on the contact state partition data;

[0085] The normal force distribution is optimized based on the minimum normal force value to obtain a normal force allocation scheme;

[0086] A slip resistance requirement diagram is constructed based on the normal force distribution scheme and contact state partitioning data.

[0087] In this embodiment of the invention, before the grasping task begins or through preliminary sensing by sensors, characteristic parameters of the object to be grasped are obtained, such as the object's material (e.g., glass, plastic, ceramic), surface treatment (e.g., smooth, rough, coated), estimated weight, and center of gravity position. Combining a contact point feature table (containing the spatial location of key contact points, current grasping force, local pressure, and friction information), the minute relative sliding velocity v0 of the gripper surface relative to the object surface at each key contact point is calculated. Based on advanced tribology theory, a velocity-dependent friction model is used to calculate the friction coefficient μ(v) considering the influence of sliding velocity. The model formula is μ(v) = ,in μ∞ is the static friction coefficient, μ∞ is the dynamic friction coefficient during high-speed sliding, and v0 is the characteristic velocity. These parameters ( ,μ∞, The specific combination of gripper material and object material is determined through offline experiments or by looking up tables. For example, if the gripper material is silicone and the object is glass, the result can be obtained by looking up a table. =0.8, μ∞=0.5, =1mm / s. The calculated μ(v) value at each contact point is used as the velocity-corrected friction coefficient at that point.

[0088] The influence of the microstructure of the object surface and the gripper surface on the actual friction force is considered. Micro-roughness parameters of the object and gripper surfaces are obtained, for example, the average roughness Ra is estimated using visual or tactile sensors. Based on the velocity-corrected friction coefficient μ(v), ​​the effective friction coefficient μ_effective, considering the influence of microstructure, is calculated using a surface roughness effect model. The model formula is μ_effective = μ(v) × [1 + k_r × ... The effective friction coefficient is calculated as follows: μ(v) is the velocity-corrected friction coefficient, Ra is the surface roughness parameter, Ra0 is the reference roughness, and k_r and α are fitting parameters determined experimentally. For example, k_r = 0.1 and α = 0.5. This model reflects that the rougher the surface, the higher the effective friction coefficient (within a certain range). The μ_effective value calculated for each contact point is taken as the effective friction coefficient for that point.

[0089] At the interface between the gripper and the object, different regions are in different contact states, such as complete adhesion (no relative slippage), micro-slip (minor slippage in a local area), or macro-slip (significant slippage across the entire contact area). The contact state of this region is analyzed based on the effective friction coefficient μ_effective and the current normal pressure P and tangential force F_friction at each contact point. For example, when |F_friction| < μ_effective × P, the region is in an adhesive state; when |F_friction| is close to or reaches μ_effective × P, the region is in a micro-slip state. Combining object characteristic parameters (such as surface hardness and elastic modulus) and gripper material properties, a contact mechanics model (e.g., a simplified model based on finite element analysis or Hertzian contact theory) is used to perform a more refined analysis of the contact interface, dividing the contact region into adhesive and potential slip zones. Each micro-region is labeled with its corresponding contact state category (e.g., adhesion, micro-slip), forming contact state partitioning data.

[0090] To prevent macroscopic slippage or attitude instability of the object during grasping, sufficient normal force is required to generate the frictional force needed to resist slippage. Based on contact state partitioning data, the required normal force contribution is calculated for micro-regions in the adhesion zone and the potential slippage zone. For the adhesion zone, the required normal force is mainly used to maintain the current contact state and resist external disturbance forces (such as gravity and inertial forces). For the potential slippage zone, the minimum normal force required to maintain stable contact or prevent slippage in this region is calculated, which is usually related to the effective friction coefficient of the region and the expected tangential force to be resisted. For example, the required total anti-slip tangential force F_slip_req depends on the object's gravity, expected acceleration, etc. The total normal force F_normal_total needs to satisfy ∑(μ_effective_i×P_i)≥F_slip_req, where P_i is the normal pressure applied in micro-region i. The goal is to find a set of P_i that minimizes the total normal force, satisfies the anti-slip requirements, and considers the characteristics of the contact state partitioning (e.g., the slippage zone requires a higher normal force to generate sufficient friction). Calculate the minimum total normal force value that meets the overall anti-slip requirements.

[0091] Minimum total normal force is a global requirement that needs to be rationally distributed across the contact areas between the gripper and the object. Normal force distribution is an optimization problem, aiming to find the optimal normal pressure distribution scheme P_i while satisfying the minimum total normal force requirement. This scheme balances the gripping torque, stabilizes the object's posture, and minimizes stress concentration. Considerations include the geometric fit between the gripper and the object, the size of the contact area, and contact state partitioning data. The total normal force is distributed proportionally to each contact area, with the proportionality factor positively correlated with the contact area and the local curvature fit. For example, more normal force is allocated to areas with larger contact areas. Simultaneously, based on the contact state partitioning data, areas in potential slip zones are allocated relatively higher normal force densities to increase the friction margin in those areas. Optimization algorithms (e.g., gradient-based optimization methods) can adjust the normal force distribution ratio in each area while satisfying the total force constraint to optimize the gripping torque balance. The target normal force value for each contact area or micro-region constitutes the normal force distribution scheme.

[0092] The normal force distribution scheme obtained in the previous step (i.e., the target normal force value for each contact area or micro-region) is combined with the contact state partitioning data to construct an anti-slip force requirement map. The anti-slip force requirement map is a two-dimensional or three-dimensional data structure corresponding to the contact interface of the soft gripper. Each contact area or micro-region contains the following information: 1) Target normal force value, indicating the normal force required to be applied to this area to meet the overall anti-slip requirement; 2) Local effective friction coefficient, indicating the current friction characteristics of this area; 3) Contact state marker, indicating whether this area is an adhesion zone or a potential slip zone. This map clearly shows the distribution of mechanical loads required for each contact area of ​​the gripper while ensuring non-slip gripping, providing key input for subsequent local-global force balance iterative optimization.

[0093] Preferably, step S3, which involves fusing local and global constraints based on the stress risk map and the anti-slip force demand map, includes:

[0094] Global stability constraints are constructed based on the anti-slip force requirement diagram and the contact point characteristic table.

[0095] High-risk areas are extracted from the stress risk map, and local safety constraints are constructed.

[0096] By fusing local safety constraints and global stability constraints and generating initial solutions, an initial force distribution scheme is obtained.

[0097] In this embodiment of the invention, the global stability constraint aims to ensure that the force applied by the gripper can balance the object's gravity, inertial force, and external disturbance forces, and generate sufficient friction to prevent the object from slipping and rotating. Using the target normal force value and local effective friction coefficient of each contact area in the anti-slip force requirement diagram, as well as the spatial location of key contact points and the object's center of gravity location information in the contact point feature table, the force balance equation and torque balance equation of the object are established. The force balance equation is ∑F_contact_i+F_gravity+F_inertial=0, where F_contact_i is the contact force applied to micro-region i (including normal and tangential components), F_gravity is the object's gravity, and F_inertial is the object's inertial force. The torque balance equation is ∑(r_i×F_contact_i)+M_gravity+M_inertial=0, where r_i is the position vector of micro-region i relative to the object's center of gravity, and M_gravity and M_inertial are the torques generated by gravity and inertial forces, respectively. Simultaneously, considering the friction cone constraint: |F_friction_i|≤μ_effective_i×P_i, it is ensured that the tangential force in each contact micro-region does not exceed its maximum static friction force. These equations and inequalities together constitute the global stability constraints, defining the overall force and torque relationships and local friction force limits required to satisfy stable gripping.

[0098] Local safety constraints aim to prevent the gripper from damaging itself or the object during the gripping process, especially in areas of stress concentration. Using a stress risk map, high-risk areas with a stress exceedance risk index higher than a preset threshold (e.g., risk index > 0.8) are identified. For each high-risk area, the maximum safe stress value or maximum safe load that the area can withstand is determined based on its current stress state, stress gradient, and predicted evolution trend displayed in the stress risk map. For example, the maximum allowable stress value σ_max_i for the area is set based on an adaptive threshold (obtained from an adaptive threshold distribution map) and a certain safety margin. This stress limit is translated into constraints on the normal pressure P_i and tangential force F_friction_i applied to the area. For example, the equivalent stress in the area is required not to exceed σ_max_i, which can be expressed as σ_e(P_i,F_friction_i,shape_i)≤σ_max_i, where σ_e is the equivalent stress calculation function and shape_i is the geometric shape parameter of the area. These maximum allowable forces or stress values ​​set for high-risk areas constitute the local safety constraints.

[0099] The global stability constraints (consisting of a series of equations and inequalities involving the normal and tangential forces in each region of the gripper) established in the previous step are integrated with the local safety constraints (maximum allowable force or stress limits applied to high-risk regions) into a unified optimization problem framework. This optimization problem seeks to find a force distribution scheme at the interface between the gripper and the object (i.e., the normal force P_i and tangential force F_friction_i to be applied to each micro-region) that satisfies both global stability requirements (force balance, moment balance, and anti-slip) and local safety requirements (stress in high-risk regions does not exceed safety limits). This is a constrained optimization problem. Quadratic programming can be used to solve this problem to obtain an initial solution. Quadratic programming is an optimization technique suitable for problems where the objective function is quadratic and the constraints are linear. Here, the objective can be set to minimize the total applied force or moment error, while global stability and local safety constraints are added as linear or nonlinear constraints. The solver calculates a set of initial normal and tangential force distributions that satisfy all constraints; this set of values ​​constitutes the initial force distribution scheme. Although this initial scheme is not optimal, it is a feasible solution and provides a starting point for subsequent iterative optimization.

[0100] Preferably, step S3, which involves iterative optimization of the initial force distribution scheme using local-global force balance, includes:

[0101] The initial force distribution scheme is divided into regions to obtain a region block table, which includes core grasping blocks, safety sensitive blocks, and transition support blocks.

[0102] The dual objective function is constructed based on the region partitioning table to obtain the objective function parameter table;

[0103] The initial force distribution scheme is iteratively initialized according to the objective function parameter table to obtain the iterative initial state;

[0104] By fixing the force values ​​of the safety-sensitive block and the transition support block, the core block is optimized based on the initial state of the iteration to obtain the core optimized force distribution;

[0105] By fixing the force values ​​of the core grasping block and the transition support block, the core optimization force distribution is optimized by sensitive block optimization to obtain the sensitive optimization force distribution;

[0106] By fixing the force values ​​of the core grasping block and the security-sensitive block, the support block is optimized to obtain the complete optimized force distribution;

[0107] Based on the objective function parameter table, the complete optimization force distribution is evaluated and convergence is determined to obtain the convergence state.

[0108] Iterative control and instruction generation are performed based on the convergence state and the complete optimization force distribution to obtain a multi-constraint driven instruction set.

[0109] In this embodiment of the invention, based on the structural characteristics of the soft gripper, its contact pattern with the object, and the stress risk map, the contact area between the soft gripper and the object is divided into different functional blocks. The core gripping block is the area that primarily undertakes the gripping task and generates the main gripping force; it typically corresponds to the area with the largest contact area or the highest curvature matching degree between the gripper and the object. The safety-sensitive block is the area marked as high-risk or a warning zone in the stress risk map; it is highly sensitive to changes in mechanical load and requires strict stress control. The transition support block is the area other than the core gripping block and the safety-sensitive block, mainly serving as an auxiliary support and stabilizing element. This block division is dynamically determined based on the gripper structure, contact geometry, and real-time stress risk assessment. Each contact micro-area is labeled with its corresponding functional block category (e.g., using integers 1, 2, and 3 to represent the core gripping block, safety-sensitive block, and transition support block, respectively), forming a region block table.

[0110] A composite optimization objective function J(f) is constructed, which combines global stability and local safety objectives. The global stability objective function G(f) aims to minimize the deviation between the actual applied force distribution and the ideal force distribution required to satisfy global stability, such as minimizing the error in total force or total torque, or minimizing the difference between the normal force distribution scheme and the current scheme: G(f) = Where P_i and F_friction_i are the normal force and frictional force of micro-region i in the current scheme, and P_target_i and F_friction_req_i are the target normal force and required anti-slip frictional force given in the anti-slip force requirement diagram. The local safety objective function L(f) aims to penalize high-stress areas to ensure that the stress does not exceed the safety threshold: L(f) = Where σ_i is the actual stress of micro-region i (obtained from the tactile feature mapping table), σ_safe_i is the safety stress threshold of micro-region i (obtained from the region threshold mapping table), and w_i is a weighting factor. For micro-regions in the safety-sensitive block, w_i is set to a higher value, while w_i is set to a lower value for other regions. The composite objective function J(f) = λG(f) + (1-λ)L(f), where λ is a balancing factor with a value range of [0,1], used to balance global stability and local safety. The initial value of λ is set to 0.5. The specific forms of these objective functions, the weights w_i, and the balancing factor λ constitute the objective function parameter table.

[0111] The initial force distribution scheme obtained from constraint fusion and initial solution generation in step S3 is used as the starting point for iterative optimization. A maximum number of iterations (e.g., 50) and a convergence threshold (e.g., the objective function value changes by less than 0.01%) are set. The initial force distribution scheme, the current λ value, the iteration counter (initialized to 0), and the objective function parameter table are stored as the initial iteration state.

[0112] In the current iteration step, the force distribution values ​​(i.e., normal force P_i and tangential force F_friction_i) of the micro-regions in the safety-sensitive block and transition support block are first fixed in the region segmentation table, and only the force distribution of the micro-regions in the core gripping block is used as the optimization variable. At this time, the optimization objective function simplifies to J_core(f_core)=λG(f)+(1-λ)L(f), where only the force f_core of the core gripping block is variable. To solve this sub-optimization problem, for example, gradient descent is used to adjust the force value of the core gripping block along the negative gradient direction of the objective function, so that the objective function value decreases. During the optimization process, the local friction cone constraints of all regions must still be satisfied. The optimization continues until the force value of the core block converges or the preset number of sub-iterations is reached. The optimized force distribution of the core gripping block, together with the fixed force distributions of the safety-sensitive block and transition support block, constitutes the core optimized force distribution.

[0113] Building upon the previous core block optimization, the force distribution values ​​of the core gripping block and the transition support block are fixed, and only the force distribution of the micro-regions in the safety-sensitive block is considered as the optimization variable. At this point, the optimization objective function is J_sensitive(f_sensitive) = λG(f) + (1-λ)L(f), where only the force f_sensitive of the safety-sensitive block is variable. Solving this sub-optimization problem focuses on the local safety objective L(f), adjusting the force values ​​of the safety-sensitive block, especially reducing the force in high-stress areas, to minimize the local safety objective function. Gradient descent is used for optimization, satisfying the local friction cone constraint. The optimized force distribution of the safety-sensitive block, together with the fixed force distributions of the core gripping block and the transition support block, constitutes the sensitive optimization force distribution.

[0114] Building upon the previous sensitive block optimization, the force distribution values ​​of the core gripping block and the safety sensitive block are fixed, and only the force distribution of the micro-region in the transition support block is considered as the optimization variable. At this point, the optimization objective function is J_support(f_support) = λG(f) + (1-λ)L(f), where only the force f_support of the transition support block is variable. Solving this sub-optimization problem mainly focuses on the global stability objective G(f), adjusting the force values ​​of the transition support block to better meet the overall force balance and torque balance requirements. Gradient descent is used for optimization, satisfying the local friction cone constraint. The optimized force distribution of the transition support block, together with the fixed force distributions of the core gripping block and the safety sensitive block, constitutes the complete optimized force distribution for the current iteration step.

[0115] Calculate the composite objective function value J(f) corresponding to the complete optimized force distribution obtained in the current iteration step. Compare the current objective function value with the objective function value of the previous iteration step. If the rate of change of the objective function value is less than the convergence threshold (0.01%), or the maximum number of iterations (50) is reached, the optimization process is considered to have converged. Simultaneously, evaluate the global stability index (e.g., total force / torque error) and local safety index (e.g., maximum stress exceedance rate) of the current scheme. Based on these evaluation results, dynamically adjust the balance factor λ. For example, if the local safety index is poor (high maximum stress exceedance rate), increase the value of λ in the next iteration to focus more on local safety; if the global stability index is poor (large force / torque error), decrease the value of λ. Record the current λ value, objective function value, and evaluation index to form convergence state information.

[0116] If the optimization process has not yet converged and has not reached the maximum number of iterations, the current complete optimized force distribution is used as the initial force distribution for the next iteration, and the iteration counter is updated to continue the loop from core block optimization to support block optimization. If the optimization process converges or reaches the maximum number of iterations, the complete optimized force distribution scheme obtained in the last iteration is taken as the optimal force distribution scheme. Based on this optimal force distribution scheme, combined with the driving model of the soft gripper (e.g., a pressure-deformation-force model), the control parameters (e.g., air pressure values, voltage values) of each driving unit (e.g., airbag, motor) required to achieve this force distribution are calculated. These driving parameters, execution sequence, timing requirements, and key area monitoring thresholds and emergency adjustment plans extracted from the stress risk map and anti-slip force demand map are formatted into an executable instruction set for the control system, forming a multi-constraint driving instruction set. This instruction set contains all the control information for driving the gripper to achieve safe and stable grasping.

[0117] Preferably, step S4, which involves progressively executing the multi-constraint drive instruction set, includes:

[0118] The software gripper is divided into 8 independent control regions. Each control region is numbered and its relationships are established to obtain the region division scheme.

[0119] Based on the region partitioning scheme, drive instructions are distributed and mapped to the multi-constraint drive instruction set to obtain the drive execution parameter table;

[0120] Based on the drive execution parameter table and stress risk map, the number of segments for each control region is dynamically determined to obtain the region segment number table;

[0121] A nonlinear segmentation design was performed on the regional segmentation table to obtain a segmentation parameter table; a regional topology diagram was constructed for the eight independent control regions.

[0122] Based on the regional topology diagram and the segmented parameter table, the inter-regional variation constraints are coordinated to obtain the coordinated segmented parameter table;

[0123] Based on the stress risk map, the regional risks are ordered to obtain an execution order list;

[0124] Generate a phased execution plan based on the coordination segmentation parameter table and the execution order list;

[0125] Drive execution and status monitoring are performed according to the phased execution plan to obtain drive execution status data.

[0126] In this embodiment of the invention, the gripper is divided into eight independent control regions based on its physical structure and the layout of the drive units. For example, a three-finger gripper can have each finger divided into three regions: fingertip, middle, and base, for a total of nine regions. Alternatively, it can be divided into eight or more independently controllable regions based on the location of the drive airbag / motor. Each region is assigned a unique number (e.g., 1 to 8). Simultaneously, a topological relationship is constructed between the regions to identify which regions are adjacent and which regions structurally influence each other (e.g., expansion of one region stretches adjacent regions). This relationship can be represented using an adjacency matrix or a graph structure to form a region partitioning scheme.

[0127] The multi-constraint drive instruction set contains the control parameters (such as air pressure and voltage values) required for each drive unit to achieve optimal force distribution. Based on the region division scheme, global or drive unit-specific instructions in the multi-constraint drive instruction set are parsed and distributed to the corresponding control regions. For example, if the instruction set specifies a target air pressure value of P_A for drive unit A, and drive unit A mainly affects regions 1 and 2, then the target air pressure value P_A is mapped to the drive execution parameters for regions 1 and 2. Simultaneously, information such as execution order, timing requirements, and safety boundary values ​​contained in the instruction set are also associated with the corresponding regions. This process transforms high-level drive instructions into specific execution parameters for each independent control region (e.g., the target air pressure for region 1 is P_1, the target air pressure for region 2 is P_2, etc.), forming a drive execution parameter table.

[0128] Progressive driven execution requires breaking down the change process of the target driving parameters in each region into multiple small steps. The number of segments affects the smoothness and safety of the execution. For the difference ΔP_i between the target parameter value and the current parameter value in each region of the driven execution parameter table, the initial number of segments is dynamically determined based on the magnitude of the change. Here, P_threshold is a preset threshold for parameter variation, such as 0.1 atmospheres or 0.5 volts. Simultaneously, combined with the stress risk map, for regions with a high stress exceedance risk index (e.g., risk index > 0.7), the number of segments is increased, for example, N_i = N_i + 2, to make the driving changes in these sensitive regions more slow and precise. The calculated number of segments for each region is recorded to form a region segmentation table.

[0129] Based on the regional segmentation table, a specific parameter change path is designed for each region. A non-linear segmentation strategy is adopted; for example, smaller parameter increments are used in the early and late stages of the driving process, while larger increments are used in the middle stage. This helps to slowly establish contact in the early stage, finely adjust in the later stage, and quickly reach the target value in the middle stage. For example, for N segments of change, the parameter increments of each segment can be distributed according to a sine or exponential function. Simultaneously, to coordinate the driving between regions, the topological relationships between regions are clarified. A regional topology diagram is constructed, showing which regions are directly adjacent (first-level adjacency), which regions are indirectly influenced by other regions (second-level adjacency), and the structural connection strength between them.

[0130] To avoid instability in the gripper structure or shear stress on the object due to inconsistent changes in driving parameters between adjacent regions, constraints need to be imposed on the driving changes between regions. Based on the region topology diagram and the segmented parameter table, the changes in driving parameters between adjacent regions are coordinated. For example, the difference in parameter changes between first-level adjacent regions within the same time step is constrained to not exceed a preset threshold: |ΔP_i(k)-ΔP_j(k)|≤C_1, where ΔP_i(k) is the parameter increment of region i at the k-th time step, and C_1 is a constant. For second-level adjacent regions, the constraint can be appropriately relaxed: |ΔP_i(k)-ΔP_j(k)|≤C_2, where C_2>C_1. Simultaneously, a "lead-follow" execution mode is introduced; for example, for regions with lower stress risk, their driving can slightly lead that of high-risk regions, providing a buffer time for adjustments in high-risk regions. Based on these constraints and modes, the segmented parameters of each region are fine-tuned to obtain a coordinated segmented parameter table.

[0131] Using information such as the stress over-limit risk index, stress gradient magnitude, and predicted stress change rate for each region in the stress risk map, the eight control regions are ranked by risk. For example, regions with high current risk indices, large stress gradients, and predicted rapid stress growth are ranked higher on the risk list. The execution order list is used to rank the regions from low to high risk or according to specific collaborative strategies (e.g., starting with the core gripping region and gradually expanding to the support region). This list guides the relative execution priority and timing of the driving instructions for each region during phased execution.

[0132] The coordinated segmented parameter table (containing the segmented parameter increments and timing information for each region) is combined with the execution order list to generate a detailed phased execution plan. This plan specifies the drive parameter values ​​that each control region should achieve at each time step (e.g., every 50ms) throughout the entire drive process. The plan considers the number of segments in each region, nonlinear segmentation design, coordination of inter-region variation constraints, and the execution order of region risks. For example, the plan specifies that at time t_k, region i executes the drive of the m_i-th segment with a target parameter value of P_i(k); region j executes the drive of the m_j-th segment with a target parameter value of P_j(k); and satisfies the timing and variation constraints between regions. This plan is a detailed, time-discrete sequence of instructions used to guide the operation of the underlying drive units.

[0133] The phased execution plan is sent to the underlying drive control module. Based on the plan, the drive control module sends control signals to the corresponding drive units (such as air valves and voltage controllers) in each area. For example, it controls the air valve to open / close to adjust the airbag pressure, or controls the voltage output to drive the electric actuator. During drive execution, the actual operating status of each drive unit is monitored in real time, such as measuring the actual airbag pressure, motor voltage, or current. Simultaneously, the actual deformation state of each area of ​​the soft gripper is monitored using the sensor data collected in step S1 (e.g., calculating local bending angles or elongation rates using strain sensor data). These real-time drive parameter values, actual deformation data, execution time, response delay, and any abnormal states (such as air pressure fluctuations or actuator malfunctions) are recorded to form drive execution status data.

[0134] Preferably, step S4, which involves monitoring contact slippage risk based on the tactile feature tensor, includes:

[0135] The pressure distribution data and friction vector field data at the current moment are extracted from the tactile feature tensor to construct contact feature data;

[0136] The dynamic characteristics of the contact between the gripper and the object are calculated based on the contact characteristic data, including the pressure center position, contact area, key point pressure value and friction direction.

[0137] Micro-slippage detection and analysis were performed based on contact dynamic characteristics to obtain a slippage risk assessment map;

[0138] Pressure anomaly pattern recognition is performed based on contact dynamic characteristics and contact characteristic data to obtain a pressure anomaly area map;

[0139] A local stability assessment map is generated based on the pressure anomaly area map and the slip risk assessment map;

[0140] Calculate the global stability index based on the slip risk assessment diagram and contact characteristic data;

[0141] Contact state monitoring results are generated based on the local stability assessment diagram and the global stability index.

[0142] In this embodiment of the invention, the latest tactile feature tensor is acquired in real time. From the current time slice of this four-dimensional tensor, a two-dimensional pressure distribution map and a two-dimensional friction vector field representing the contact interface between the soft gripper and the object are extracted. The pressure distribution map is a two-dimensional matrix, where each element represents the normal pressure value at the corresponding spatial location. The friction vector field is a two-dimensional vector field, where each location contains a two-dimensional vector representing the magnitude and direction of the tangential force. These two two-dimensional data structures, along with the relevant spatial location information and the original sensor data, are organized together to form the contact feature data at the current moment.

[0143] The contact characteristic data is analyzed to extract macroscopic and local features describing the contact interface. The pressure center location of the contact interface is calculated, i.e., the weighted average location of all contact pressures: C = (∑_iP_i × r_i) / ∑_iP_i, where P_i is the pressure value of micro-region i, and r_i is the position vector of micro-region i. The effective contact area is calculated, i.e., the total area of ​​micro-regions where the pressure value exceeds a preset threshold (e.g., 0.01 MPa). Key contact points are identified, such as pressure peak points, friction peak points, or high-risk points in the stress risk map, and the specific pressure values ​​and friction vectors of these points are recorded. The overall distribution of the friction vector field is analyzed, such as calculating the total friction vector and the average friction direction. These calculated macroscopic and local features constitute the contact dynamic characteristics.

[0144] Microslip is a localized, minute relative motion preceding macro-slip and is a significant precursor to grip instability. Microslip signs at the contact interface can be detected using contact dynamics and contact characteristic data. One approach is to analyze whether the local friction coefficient reaches or approaches a critical value: for each contact micro-region, calculate its local friction coefficient μ_local_i = |F_friction_i| / P_i. If μ_local_i is close to or exceeds the effective friction coefficient μ_effective_i (obtained from the anti-slip force demand map), it indicates that the region is on the verge of microslip. Another method is to monitor changes in the direction of the local friction force vector: during stable gripping, the direction of the local friction force is usually related to the object's gravity or expected direction of motion; if the friction force direction in a certain region deflects abnormally or fluctuates rapidly, it indicates the occurrence of microslip. Rapid, unexpected changes in local pressure distribution, which accompany the redistribution of contact points caused by microslip, can also be analyzed. Based on these analyses, a micro-slip risk index is calculated for each micro-region of the contact interface (e.g., based on the value of μ_local_i / μ_effective_i or the severity of the change in the direction of friction), forming a two-dimensional or three-dimensional slip risk assessment map.

[0145] Monitor the contact interface for abnormal pressure distribution patterns, such as excessively high local pressure (leading to object damage) or excessively low local pressure (leading to loss of contact). Compare the pressure distribution map from the contact feature data with expected or historically successful pressure distribution patterns, for example, using image comparison algorithms or statistical methods. Identify areas where pressure values ​​significantly deviate from expected values ​​(e.g., exceeding a certain multiple of the average pressure or exceeding historical safe ranges). Pay particular attention to the location and magnitude of pressure peaks, as well as the uniformity of pressure distribution. Combine this with stress risk mapping to identify whether pressure anomalies occur in safety-sensitive blocks or high-risk areas. Mark the identified pressure anomaly areas to create a pressure anomaly area map.

[0146] By integrating the pressure anomaly map (reflecting abnormal local load conditions) and the slip risk assessment map (reflecting the uncertainty of local contact states), a local stability assessment of the gripper-object contact interface is performed. Anomaly areas in the pressure anomaly map and high slip risk areas in the slip risk assessment map are marked as locally unstable regions. Different local stability risk levels are assigned based on the specific anomaly type (excessively high / low pressure, degree of micro-slippage). For example, a region exhibiting both high pressure and high micro-slippage risk is rated as the highest level of local instability. These local stability risk levels are then mapped onto the spatial regions corresponding to the gripper contact interface to form a local stability assessment map.

[0147] Assess the overall stability of the entire gripping system. One global stability metric is the gripping torque margin, which is the margin between the stabilizing torque generated by the force applied by the gripper and the disturbance torques (such as gravitational torque or external disturbance torques) that cause instability in the object. Calculate the total force and torque applied by the gripper to the object using the pressure distribution and friction vector field from the contact characteristic data. Perform a balance analysis with the object's gravity, inertial forces, etc. Calculate the distance of the gripping torque relative to the friction cone boundary; this distance can serve as an indicator of anti-slip stability. For example, calculate whether the gripping force and torque are within the safety margin range inside the friction cone. Another global stability metric is the rate of change of contact area or the degree of fluctuation in the pressure center position. If the contact area decreases rapidly or the pressure center position drifts drastically, it indicates overall gripping instability. These calculated numerical indicators constitute the global stability metrics.

[0148] By integrating local stability assessment maps (providing detailed risk information for each area of ​​the contact interface) and global stability indicators (providing an overview of the overall gripping status), the final contact status monitoring results are generated. The contact status monitoring results are a comprehensive report containing the following information: 1) Overall gripping stability status (e.g., stable, attentive, unstable), based on global stability indicators; 2) A list of locally unstable areas, indicating which areas of the gripper have abnormal pressure or micro-slip risks, and their risk levels (obtained from the local stability assessment maps); 3) Potential slip directions or pressure overload directions (obtained from the slip risk assessment map and the abnormal pressure area map); 4) Predictions of future contact status evolution trends (e.g., how long until macroscopic slip occurs if the current state continues). This result provides real-time, multi-dimensional contact status feedback information for subsequent fine-tuning compensation calculations and emergency response.

[0149] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0150] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. An intelligent control method based on the Internet of Things, characterized in that, Includes the following steps: Step S1: Perform multimodal data acquisition and preprocessing on the contact process between the soft gripper and the object to obtain a time-synchronized dataset; construct a tactile feature tensor containing pressure, friction, and strain based on the time-synchronized dataset; Step S2: Calculate and map the stress gradient based on the tactile feature tensor to obtain the stress gradient field and region threshold mapping table; The stress gradient threshold is adaptively adjusted based on the region threshold mapping table and the stress gradient field. Specifically, the region sensitivity of the soft gripper is classified to obtain a region sensitivity classification table; the similarity of the currently gripped object is calculated based on historical gripping data to obtain a similar case set; time weights are assigned to the similar case set to obtain a weighted case set; and a baseline threshold is calculated for the weighted case set to obtain a region baseline threshold table. Stress change rate monitoring is performed based on stress gradient field and tactile feature tensor to obtain stress change rate map; sensitive areas of change rate are identified on stress change rate map to obtain rapid change area marking map; threshold dynamic adjustment of rapid change area marking map is performed on region threshold mapping table to obtain preliminary adjustment threshold table; The initial threshold table is spatially smoothed to obtain a smoothed threshold table; an adaptive threshold distribution map is generated based on the smoothed threshold table. Stress risk prediction is performed based on the adaptive threshold distribution map, specifically by assessing stress over-limit risk based on the adaptive threshold distribution map to obtain the stress risk index distribution. Stress evolution trend is predicted based on the stress risk index distribution and tactile feature tensor, resulting in a stress evolution prediction map; A stress risk map is constructed based on the stress risk index distribution, stress gradient field, and stress evolution prediction map. Step S3: Analyze the contact points based on the stress risk map to obtain a contact point characteristic table; The anti-slip force is calculated based on the contact point feature table. Specifically, the object characteristic parameters are obtained, and the velocity-dependent friction coefficient of the contact point is calculated in combination with the contact point feature table to obtain the velocity-corrected friction coefficient. The surface roughness effect is analyzed based on the speed-corrected friction coefficient and the contact point characteristic table to obtain the effective friction coefficient; the contact area state is analyzed based on the effective friction coefficient and the object characteristic parameters to obtain contact state partition data. Calculate the minimum normal force value based on the contact state partition data; The normal force distribution is optimized based on the minimum normal force value to obtain a normal force allocation scheme; Construct an anti-slip force requirement diagram based on the normal force distribution scheme and contact state zoning data; Local-global constraint fusion is performed based on the stress risk map and the anti-slip force demand map. Specifically, global stability constraints are constructed based on the anti-slip force demand map and the contact point characteristic table; high-risk areas are extracted from the stress risk map to construct local safety constraints. By fusing local safety constraints and global stability constraints and generating initial solutions, an initial force distribution scheme is obtained. The initial force distribution scheme is subjected to local-global force balance iterative optimization, specifically: the initial force distribution scheme is divided into regions to obtain a region block table, which includes core grasping blocks, safety sensitive blocks and transition support blocks; a dual objective function is constructed based on the region block table to obtain the objective function parameter table; The initial force distribution scheme is iteratively initialized according to the objective function parameter table to obtain the iterative initial state. The force values ​​of the safety sensitive block and the transition support block are fixed, and the core block is optimized to obtain the core optimized force distribution. The force values ​​of the core grasping block and the transition support block are fixed, and the core optimized force distribution is optimized by the sensitive block to obtain the sensitive optimized force distribution. The force values ​​of the core grasping block and the safety sensitive block are fixed, and the sensitive optimized force distribution is optimized by the support block to obtain the complete optimized force distribution. The complete optimized force distribution is evaluated for objective and convergence is judged according to the objective function parameter table to obtain the convergence state. Iterative control and instruction generation are performed based on the convergence state and the complete optimized force distribution to obtain the multi-constraint driving instruction set. Step S4: Perform progressive drive execution on the multi-constraint drive instruction set to obtain drive execution status data; evaluate the deformation effect data of the gripper in real time; perform contact slip risk monitoring based on the tactile feature tensor to obtain contact status monitoring results; Fine-tuning compensation instructions are generated based on deformation effect data and contact state monitoring results; the drive feedback loop is driven in real time based on drive execution state data and fine-tuning compensation instructions to achieve intelligent control of the gripper.

2. The intelligent control method based on the Internet of Things according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: The sensor array installed on the soft gripper collects signals during the contact process with the object to obtain raw sensor signals, which include pressure signals, friction signals and strain signals. Step S12: Perform signal filtering and noise cancellation on the original sensor signal to obtain the filtered signal dataset; Step S13: Perform sensor data time synchronization on the filtered signal dataset to obtain a time-synchronized dataset; Step S14: Construct the contact pressure distribution map, contact friction vector field, and internal strain field of the soft body based on the time synchronization dataset; Step S15: Organize the spatial positional relationship of the contact pressure distribution map, the contact friction vector field, and the internal strain field of the soft body to obtain the tactile feature tensor.

3. The intelligent control method based on the Internet of Things according to claim 1, characterized in that, Step S2, which involves calculating and mapping the stress gradient based on the tactile feature tensor, includes: Extract the tactile feature mapping table from the tactile feature tensor; Calculate the stress gradient field based on the tactile feature mapping table; Stress concentration regions are identified in the stress gradient field to obtain a stress concentration region marking map; Historical data correlation analysis was performed on the stress concentration area marker map to obtain a historical correlation feature set; Based on the historical correlation feature set and stress concentration area labeling map, the regional differential threshold is calculated to obtain the regional threshold mapping table.

4. The intelligent control method based on the Internet of Things according to claim 1, characterized in that, Step S4, which involves progressively executing the multi-constraint drive instruction set, includes: The software gripper is divided into 8 independent control regions. Each control region is numbered and its relationships are established to obtain the region division scheme. Based on the region partitioning scheme, drive instructions are distributed and mapped to the multi-constraint drive instruction set to obtain the drive execution parameter table; Based on the drive execution parameter table and stress risk map, the number of segments for each control region is dynamically determined to obtain the region segment number table; A nonlinear segmentation design was performed on the regional segmentation table to obtain a segmentation parameter table; a regional topology diagram was constructed for the eight independent control regions. Based on the regional topology diagram and the segmented parameter table, the inter-regional variation constraints are coordinated to obtain the coordinated segmented parameter table; Based on the stress risk map, the regional risks are ordered to obtain an execution order list; Generate a phased execution plan based on the coordination segmentation parameter table and the execution order list; Drive execution and status monitoring are performed according to the phased execution plan to obtain drive execution status data.

5. The intelligent control method based on the Internet of Things according to claim 1, characterized in that, Step S4, which involves monitoring contact slip risk based on the tactile feature tensor, includes: The pressure distribution data and friction vector field data at the current moment are extracted from the tactile feature tensor to construct contact feature data; The dynamic characteristics of the contact between the gripper and the object are calculated based on the contact characteristic data, including the pressure center position, contact area, key point pressure value and friction direction. Micro-slippage detection and analysis were performed based on contact dynamic characteristics to obtain a slippage risk assessment map; Pressure anomaly pattern recognition is performed based on contact dynamic characteristics and contact characteristic data to obtain a pressure anomaly area map; A local stability assessment map is generated based on the pressure anomaly area map and the slip risk assessment map; Calculate the global stability index based on the slip risk assessment diagram and contact characteristic data; Contact state monitoring results are generated based on the local stability assessment diagram and the global stability index.

Images