A soft robot gripper control method based on strain correction

Abstract

The application discloses a kind of based on strain correction's soft manipulator's gripping control method, it is related to sensor technical field, the method constructs multiple different sample design parameters and establishes finite element model to carry out tensile simulation and obtains tensile simulation result, a large number of simulation results are combined with theoretical analysis, and strain transmission loss parameters that cannot be directly calculated, which affect strain transmission efficiency, can be fitted;When actually carrying out gripping control, according to the actual design parameters of soft manipulator and the actual design parameters of strain sensor, strain transmission loss parameters are combined to determine average strain transmission rate for strain sensor strain measurement value correction, and the strain measurement value after correction is closer to the actual strain of soft manipulator, so the accuracy of controlling the gripping state of soft manipulator according to the strain measurement value after correction is higher, so that the gripping precision of soft manipulator is higher and more difficult to damage goods.

Description

Technical Field

[0001] The present application relates to the field of sensor technology, and in particular to a grasping control method of a soft manipulator based on strain correction. Background Art

[0002] With the development of automatic control technology, robots are widely used in various industries. They can realize actions such as grasping, carrying and operating objects, and are important components for realizing automated operations.

[0003] Traditional manipulators are rigid manipulators made of rigid materials, but they are not ideal for handling small and fragile objects. Therefore, with the development of soft devices, soft manipulators have been developed. The grippers of soft manipulators are generally made of soft materials such as silicone. Compared with rigid manipulators, they have greater flexibility, more flexible deformation capabilities, and higher adaptability. When grasping objects, they have higher grasping accuracy and are less likely to cause damage. Currently, they are widely used in underwater, medical, food, agricultural and other fields.

[0004] However, the operational effectiveness of soft manipulators still depends heavily on the grasping control method. When controlling the grasping of a soft manipulator, the degree of opening and closing of the manipulator's fingertips must be adjusted to provide an appropriate clamping force. This prevents the clamping force from being too small, resulting in a stable grip, while also preventing excessive clamping force from damaging the object. The clamping force of a soft manipulator is closely related to the deformation of its fingertip structure, which can generally be estimated by the strain on its outer surface. Therefore, based on the feedback information from strain sensors attached to the outer surface of the fingertip structure, the deformation of the fingertip structure can be estimated and the current gripping force can be determined. The grasping posture of the soft manipulator can then be adjusted to achieve closed-loop control of the gripping force. However, due to unavoidable measurement errors or strain loss, the feedback information from the strain sensors often differs from the actual strain on the outer surface of the fingertip structure. Due to the insufficient accuracy and precision of the strain sensor feedback information, the grasping control of the soft manipulator is also insufficient. Summary of the Invention

[0005] In response to the above-mentioned problems and technical requirements, this application proposes a grasping control method for a soft manipulator based on strain correction. The technical solution of this application is as follows:

[0006] A strain-corrected soft manipulator grasping control method, the grasping control method comprising:

[0007] Constructing multiple sets of different sample design parameters, and constructing a finite element model of a soft substrate integrated with a strain sensor according to each set of sample design parameters, wherein the strain sensor includes an insulating layer stacked on the soft substrate and a sensitive layer stacked on the insulating layer;

[0008] A tensile simulation is performed on each constructed finite element model to obtain a tensile simulation result. The tensile simulation result includes the tensile strain ε applied to the soft substrate in the finite element model. app And the average strain ε of the sensitive layer of the strain sensor along the tensile direction in the finite element model avg ;

[0009] The strain transmission loss parameters were obtained by fitting all the tensile simulation results;

[0010] According to the actual design parameters of the soft manipulator and the actual design parameters of the strain sensor made on the soft manipulator, the average strain transfer rate α is determined in combination with the strain transfer loss parameter;

[0011] The average strain transfer rate α is used to correct the strain measurement value ε of the strain sensor made on the soft manipulator to obtain the corrected strain measurement value

[0012] The grasping state of the soft manipulator is controlled according to the corrected strain measurement value ε′.

[0013] The beneficial technical effects of this application are:

[0014] The present application discloses a grasping control method for a soft manipulator based on strain correction. The method uses sample design parameters and performs finite element tensile simulation. Through a large number of simulation results combined with theoretical analysis, the strain transmission loss parameter that cannot be directly calculated and affects the strain transmission efficiency can be fitted. Then, when actually performing grasping control, the average strain transmission rate can be calculated by combining the actual design parameters and the strain transmission loss parameter to correct the strain measurement value of the strain sensor. The corrected strain measurement value is closer to the actual strain of the soft manipulator, so more accurate grasping control can be achieved, making the soft manipulator's grasping accuracy higher and less likely to damage objects. Compared with the traditional method of controlling the manipulator with tactile or visual sensors, this strain feedback control strategy based on mechanical models has the advantages of low cost, high accuracy, and wide applicability. It is suitable for promotion in application scenarios that require precise control of posture and force. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flowchart of a grasping control method according to an embodiment of the present application.

[0016] Figure 2 Schematic diagram of the structure of the fabricated strain sensor.

[0017] Figure 3 It is a structural diagram of the strain sensor and the soft substrate integrated together using a surface integration method.

[0018] Figure 4 It is a structural diagram of the strain sensor and the soft substrate integrated together by embedded integration.

[0019] Figure 5 This is a schematic diagram of the stretching simulation scenario of the finite element model.

[0020] Figure 6 It is a cross-sectional schematic diagram of the xy plane of the finite element model.

[0021] Figure 7 It is a schematic diagram of the cross-section explosion of the zy plane of the finite element model. DETAILED DESCRIPTION

[0022] The specific implementation of this application will be further described below with reference to the accompanying drawings.

[0023] The present application discloses a grasping control method for a soft manipulator based on strain correction. By performing strain correction on the strain measurement value ε obtained by the strain sensor on the outer surface of the fingertip structure of the soft manipulator, the accuracy of sensing the actual strain on the outer surface of the fingertip structure of the soft manipulator is improved, thereby improving the accuracy and precision of the grasping control of the soft manipulator. First, the reasons for the deviation between the strain measurement value obtained by the strain sensor and the actual strain on the outer surface of the fingertip structure of the soft manipulator are analyzed as follows: On the one hand, the strain must be transmitted from the soft manipulator to the strain sensor, and often has to pass through multiple layers of structure inside the strain sensor to be transmitted to the sensitive layer before it can be sensed. The strain will be lost layer by layer during the transmission process, resulting in the strain measurement value being lower than the actual strain. On the other hand, the strain sensor is generally attached to the soft manipulator, and the local strain of the soft manipulator will be constrained by the attached strain sensor, thereby further widening the gap between the strain measurement value and the actual strain.

[0024] The combined influence of the above two reasons leads to a deviation between the strain measurement value and the actual strain. Part of this deviation is related to the design parameters of the soft manipulator and the strain sensor and can be directly calculated. The other part is nonlinear. This application obtains these nonlinear strain transmission loss parameters through finite element simulation. The grasping control method includes the following steps, please refer to Figure 1 Flowchart:

[0025] Step 1: construct multiple sets of different sample design parameters, and construct a finite element model integrating the soft substrate and the strain sensor according to each set of sample design parameters.

[0026] Strain sensors primarily consist of an insulating layer and a sensitive layer. The sensitive layer is used to sense strain, while the insulating layer insulates the sensitive layer. Furthermore, strain sensors also include components such as connecting wires, whose effects on strain are negligible and therefore can be disregarded during modeling and strain correction. Therefore, the design parameters for each sample constructed include those for the soft substrate and the strain sensor. The strain sensor design parameters include those for the sensitive layer and the insulating layer. The design parameters for each component primarily include dimensions and materials.

[0027] To accurately fit strain transfer loss parameters, it's necessary to comprehensively consider different strain transfer scenarios. Therefore, theoretically, it's necessary to construct a combination of design parameters for different soft substrates and strain sensors. However, given that in real-world applications, the design parameters of strain sensors are often predetermined and fixed, and the strain sensors are simply fabricated and applied to different soft manipulators, this approach can be simplified. While the design parameters of the strain sensor remain fixed, the design parameters of different soft substrates can be combined to construct sample design parameters.

[0028] Strain sensors used in soft manipulators are generally manufactured by direct writing printing. Due to printing accuracy issues, the actual design parameters of the strain sensors printed by direct writing printing equipment according to the theoretical design parameters often deviate from the theoretical design parameters. Therefore, in order to better characterize the characteristics of the strain sensor during modeling and simulation, when constructing sample design parameters based on the design parameters of the strain sensor, the actual design parameters of the strain sensor are used instead of the theoretical design parameters. Therefore, first, a strain sensor is manufactured using direct writing printing technology, including: using the manufacturing material of the insulating layer 11 to print and manufacture the insulating layer 11 according to the size specifications of the insulating layer 11, and then using the manufacturing material of the sensitive layer 12 to print and manufacture the sensitive layer 12 on the insulating layer 11 according to the size specifications of the sensitive layer 12, and using conductive material to print and manufacture single-ended electrodes 13 at both ends of the sensitive layer 12, and connecting a metal lead 14 to each single-ended electrode 13 for signal lead-out. The overall structure of the strain sensor thus printed is shown as follows: Figure 2 In one embodiment, the insulating layer 11 is made of epoxy resin, the sensitive layer 12 is made of carbon paste, the single-ended electrode 13 is made of silver paste, and the metal lead 14 is made of copper wire.

[0029] Then, the actual design parameters of the strain sensor are measured and determined, including the geometric characteristic parameters and basic mechanical parameters of the strain sensor:

[0030] (1) For the geometric characteristic parameters of the strain sensor, different characterization methods are used for the insulating layer 11 and the sensitive layer 12: the width w of the sensitive layer 12 in the fabricated strain sensor is determined by observation using an ultra-depth of field microscope. f , thickness t f The width w of the insulating layer 11 in the strain sensor was determined by optical microscope observation. b and thickness t b .

[0031] (2) For the basic mechanical parameters of the strain sensor, a similar method is used for the insulating layer 11 and the sensitive layer 12: the material of the sensitive layer 12 is printed into a standard tensile test and a tensile test is performed to determine the Young's modulus E of the sensitive layer. f and the shear modulus G of the sensitive layer f The Young's modulus E of the insulating layer is determined by printing the insulating layer 11 into a standard tensile test material and performing a tensile test. b and the shear modulus of the insulation layer G b .

[0032] Therefore, the actual design parameters of the strain sensor determined include: the thickness of the sensitive layer t f , the width of the sensitive layer w f , half the length L of the sensitive layer, Young's modulus E of the sensitive layer f , shear modulus G of the sensitive layer f The thickness of the insulating layer t b , the width of the insulation layer w b , Young's modulus E of the insulation layer b , shear modulus G of the insulation layer b . And the Young's modulus E of the sensitive layer f and shear modulus G f Both are related to the material of the sensitive layer, the Young's modulus E of the insulating layer b and shear modulus G b All are related to the material of the insulation layer.

[0033] Then, multiple sets of different sample design parameters are constructed based on the actual design parameters of the strain sensor. Each set of sample design parameters includes the actual design parameters of the strain sensor and the sample design parameters of the soft substrate. The sample design parameters of the soft substrate in each set of sample design parameters are different. The sample design parameters of the soft substrate in each set of sample design parameters include: the thickness of the soft substrate t s , the width of the soft substrate w s , Young's modulus E of soft substrate s and the shear modulus G of the soft substrate s Similarly, the Young's modulus E of the soft substrate s and shear modulus G sBoth are related to the material of the soft substrate.

[0034] Considering that the width of the soft manipulator is always greater than the width of the insulating layer in actual application, the width of the soft substrate is also always greater than the width of the insulating layer in the modeling and simulation stage. When only the width of the soft substrate is changed, it is found that when the width of the soft substrate is greater than 1.5 times the width of the insulating layer, changing the width of the soft substrate has no effect on the strain transfer. Therefore, a further simplified method can directly fix the width of the soft substrate to a value greater than 1.5 times the width of the insulating layer, and by changing the thickness t of the soft substrate s and / or materials to construct different sample design parameters.

[0035] After constructing the sample design parameters, a finite element model can be constructed according to each set of sample design parameters. When actually making the strain sensor on the soft manipulator, there are two main sensor integration methods: surface integration and embedded integration. The surface integration method refers to integrating the strain sensor directly on the surface of the soft manipulator, while the embedded integration method refers to embedding the strain sensor in a groove opened on the surface of the soft substrate. Regardless of the sensor integration method used, the strain transmission path and the strain measurement value are basically the same due to the combined influence of the above two aspects. Therefore, in order to comprehensively consider the characterization of different scenarios to fit more accurate parameters, a finite element model is constructed according to each set of sample design parameters, in which the soft substrate and the strain sensor are integrated using different sensor integration methods. The sensor integration methods include surface integration and embedded integration. The strain sensor 1 in the finite element model constructed using the surface integration method is located on the surface of the soft substrate 2, as shown in FIG. Figure 3 As shown. The strain sensor 1 in the finite element model constructed by embedded integration is embedded in the groove opened on the surface of the soft substrate 2, as shown Figure 4 shown.

[0036] Step 2: Perform stretching simulation on each constructed finite element model to obtain the stretching simulation results. Figure 3 As an example of the surface integration method shown in the figure, the stretch simulation scene is as follows Figure 5 As shown, for the convenience of description, a coordinate system xyz is established, the length direction of the sensitive layer in the strain sensor 1 is the z-axis, the width direction of the sensitive layer in the strain sensor 1 is the x-axis, the xz plane is located on the upper surface of the sensitive layer, and the direction perpendicular to the xz plane and vertically downward along the horizontal plane is the y-axis. The cross-sectional schematic diagram of the xy plane of the finite element model is shown as follows Figure 6 shown.

[0037] Perform tensile simulation on the finite element model and obtain the tensile simulation results of each tensile simulation, including the tensile strain ε applied to the soft substrate in the finite element model app And the average strain ε of the sensitive layer of the strain sensor along the tensile direction in the finite element modelavg Specifically: fix one end of the soft substrate 2 in the finite element model along the length direction of the sensitive layer in the strain sensor 1, and apply uniform tension to the other end to generate tensile strain ε app , that is, a tensile strain ε is applied to the soft substrate 2 along the z-axis direction app During the stretching simulation, the tensile strain ε(z) of the sensitive layer 12 in the strain sensor along the stretching direction is measured, and the tensile strain ε(z) is calculated according to The average strain ε of the sensitive layer 12 along the tensile direction in the strain sensor is calculated avg , l is half of the length L of the sensitive layer in the strain sensor.

[0038] Step 3: All tensile simulation results are integrated to obtain the strain transmission loss parameters.

[0039] First, the strain transfer efficiency is analyzed. In order to study the influence of some main physical parameters on the measurement accuracy of the sensor, the following three basic premises are proposed: (1) The materials of each structural layer in the finite element model, including the soft substrate, insulating layer and sensitive layer, are isotropic and linear elastic. (2) There is no relative slip between the two adjacent structural layers. (3) The length L of the sensitive layer is much larger than the thickness of each structural layer. The cross-sectional explosion diagram of the yz plane of the unit segment structure with a length of dz cut from the entire finite element model is shown as follows: Figure 7 As shown. According to the basic equation of elasticity problem, the equilibrium equation can be obtained by force analysis:

[0040]

[0041] Among them, σ f is the normal stress of the sensitive layer, σ b is the normal stress of the insulating layer, σ s is the normal stress of the soft substrate. τ bf is the shear stress at the interface between the sensitive layer and the insulating layer, τ bs is the shear stress at the interface between the insulating layer and the soft substrate.

[0042] Since the length L of the sensitive layer is much larger than the thickness of each structural layer, that is, L>>t f , L>>t b , L>>t s , assuming that the shear stress in different structural layers changes linearly along the vertical direction y, the shear stress τ of the sensitive layer along the y direction f (y), shear stress τ of the insulating layer changing along the y direction b (y), shear stress τ of the soft substrate along the y direction s The linear distribution of (y) is:

[0043]

[0044] For linear elastic materials, the constitutive model and geometric equations of the material under small-scale deformation are:

[0045]

[0046] Among them, σ is the normal stress of the material, τ is the shear stress of the material, ε is the normal strain of the material, γ is the shear strain of the material, E is the Young's modulus of the material, G is the shear modulus of the material, and u is the displacement of the material. Substituting the shear stress of each structural layer along the y direction into the constitutive model and geometric equation, the displacement u of the sensitive layer along the y direction is derived by integration. f (y) and the displacement u of the soft substrate along the y direction s (y) is:

[0047]

[0048] Since there is no relative slip between two adjacent structural layers due to perfect bonding, the average displacement u of the sensitive layer can be obtained by assuming continuity of the interface displacement between the structural layers. f , average displacement of the insulating layer u b and the average displacement u of the soft substrate s They are:

[0049]

[0050] The average displacement u of the soft base is given by the continuity of the displacement at the interface between the structural layers. s and the average displacement u of the sensitive layer f The difference between them is:

[0051]

[0052] Since the structural layers of the strain sensor are deformed synchronously, it can be considered that the strain gradients of different structural layers are of the same magnitude. The strain differential equation can be derived by integrating the above formulas:

[0053]

[0054] The expression of shear lag coefficient k is:

[0055]

[0056] The general solution of the strain differential equation above is usually given by e with unknown coefficients. kx and e -kx Composition, namely:

[0057] ε f (x) = a1 + a2ekx +a3e -kx (9)

[0058] Combined with the boundary conditions, the general solution of the above equation is:

[0059] ε f (x) = a1 - λa2e kx -λa3e -kx (10)

[0060] The parameter λ is used to estimate the overall reinforcement effect of the strain sensor on the soft substrate. a1, a2, and a3 are constants determined by the boundary conditions and can be determined based on the boundary conditions:

[0061]

[0062] Substituting the above forms of a1, a2 and a3 into the general solution formula, we can obtain the strain distribution ε of the sensitive layer along the tensile direction, i.e., the z-axis direction: f (z) can be written as:

[0063]

[0064] Then the average strain ε of the sensitive layer 12 along the tensile direction can be further obtained avg It can be written as:

[0065]

[0066] in,

[0067]

[0068] Where β is the effect of the strain sensor fabrication area on the overall modulus of the structure, and F is the tensile force applied to the soft substrate. a is the local elastic modulus of the area on the soft substrate where the strain sensor is attached, S a is the sum of the actual areas of each structural layer in the finite element model on the yz plane, so:

[0069]

[0070] β is affected by the design parameters of the soft substrate, the insulating layer of the strain sensor, and the sensitive layer. It is mainly affected by the design parameters of the soft substrate and the sensitive layer, and can be written as:

[0071]

[0072] In the above formula, S f =w f ×t f is the area of ​​the sensitive layer on the yz plane, Sb =w b ×t b is the area of ​​the insulating layer on the yz plane. The parameters a and b are Related parameters, and the specific forms of parameters a and b are:

[0073]

[0074] Among them, c1, c2, c3, c4, c5 and c6 are the strain transmission loss parameters that need to be fitted.

[0075] Combining equations (13) to (17), we can obtain:

[0076]

[0077] In order to facilitate analysis, the parameters η, S, ξ, and m related to the sample design parameters of the finite element model are introduced and defined as follows:

[0078]

[0079] Simplifying formula (18) we can get:

[0080]

[0081] Based on the above theoretical analysis, it can be seen that each time the finite element model is stretched, the parameters η, S, ξ, and m can be determined using equations (19) and (8) according to the sample design parameters of the finite element model for this stretching simulation, and the tensile strain ε applied to the soft substrate in the finite element model during this stretching simulation can be expressed as app and the average strain ε of the sensing layer along the tensile direction avg Substituting this into the equation, we can calculate the actual strain transfer efficiency during the tensile simulation. Then, according to the above function form, the strain transmission loss parameters c1, c2, c3, c4, c5 and c6 can be fitted using the results of a large number of tensile simulations.

[0082] Step 4: Determine the average strain transfer rate α based on the actual design parameters of the soft manipulator and the strain sensors fabricated on it, combined with the strain transfer loss parameter. Strain sensors are fabricated on the soft manipulator using either surface integration or embedded integration, and are typically located at the fingertips, where the strain is greater and more easily detected.

[0083] In practical application, the actual design parameters of the soft manipulator that need to be obtained include: the actual design parameters of the soft manipulator include the thickness t of the soft manipulator s ′, width w of the soft manipulator s′, Young's modulus E of the soft manipulator s ′ and the shear modulus G of the soft manipulator s ′. Among them, the thickness of the soft manipulator t s ′ is the thickness of the soft manipulator at the location where the strain sensor is made, and the Young's modulus E of the soft manipulator s ′ and shear modulus G s ′ is related to the material of the soft manipulator. When the strain sensor is made at the fingertip of the soft manipulator, the width w of the soft manipulator is s ′ is the width of the fingertip where the strain sensor is located. In addition, in actual application, the width of the soft manipulator w s ' is usually greater than 1.5 times the width of the insulation layer. As mentioned above, the width of the soft manipulator w s ′ has little effect on strain transfer.

[0084] According to the actual design parameters of the soft manipulator and the actual design parameters of the strain sensor manufactured on the soft manipulator, the parameters η0, S0, ξ0, and m0 are calculated using the following formula (21) according to the calculation form of formula (19) and formula (18):

[0085]

[0086] in, E a ′ is the local elastic modulus of the area where the strain sensor is attached to the soft manipulator and

[0087] S a ′=w s ′×t s ′+w b ×t b +w f ×t f .

[0088] Finally, the calculated parameters η0, S0, ξ0, m0 and the fitted strain transfer loss parameters c1, c2, c3, c4, c5 and c6 are substituted into formula (20) to obtain the average strain transfer rate α:

[0089]

[0090] Step 5: Use the average strain transfer rate α to correct the strain measurement value ε of the strain sensor made on the soft manipulator to obtain the corrected strain measurement value

[0091] Step 6, controlling the grasping state of the soft manipulator according to the corrected strain measurement value ε′, is similar to the conventional practice of directly controlling the grasping state of the soft manipulator according to the strain measurement value, and this application will not elaborate on this.

[0092] The above description is only a preferred embodiment of the present application, and the present application is not limited to the above embodiments. It is understood that other improvements and variations directly derived or imagined by those skilled in the art without departing from the spirit and concept of the present application should be considered to be included in the scope of protection of the present application.

Claims

1. A grasping control method for a soft manipulator based on strain correction, characterized in that: The grasping control method comprises: Constructing multiple sets of different sample design parameters, and constructing a finite element model integrating a soft substrate and a strain sensor according to each set of sample design parameters, wherein the strain sensor includes an insulating layer stacked on the soft substrate and a sensitive layer stacked on the insulating layer; wherein constructing the multiple sets of different sample design parameters includes: using direct writing printing technology to produce the strain sensor, and measuring and determining the actual design parameters of the strain sensor; constructing multiple sets of different sample design parameters according to the actual design parameters of the strain sensor, each set of sample design parameters includes the actual design parameters of the strain sensor and sample design parameters of the soft substrate, and the sample design parameters of the soft substrate in each set of sample design parameters are different; in each set of sample design parameters: the sample design parameters of the soft substrate include: the thickness t of the soft substrate s , the width of the soft substrate w s , Young's modulus E of soft substrate s and the shear modulus G of the soft substrate s The actual design parameters of the strain sensor include: the thickness of the sensitive layer t f , the width w of the sensitive layer f , half the length of the sensitive layer l, Young's modulus of the sensitive layer E f , shear modulus G of the sensitive layer f ;Thickness of the insulation layer t b , the width of the insulation layer w b , Young's modulus E of the insulation layer b , shear modulus G of the insulation layer b ; Among them, the Young's modulus E of the soft substrate s and shear modulus G s Both are related to the material of the soft substrate, the Young's modulus E of the sensitive layer f and shear modulus G f Both are related to the material of the sensitive layer, the Young's modulus E of the insulating layer b and shear modulus G b All are related to the material of the insulation layer; A tensile simulation is performed on each constructed finite element model to obtain a tensile simulation result, wherein the tensile simulation result includes the tensile strain ε applied to the soft substrate in the finite element model. app And the average strain ε of the sensitive layer of the strain sensor along the tensile direction in the finite element model avg ; The strain transmission loss parameters are obtained by fitting all the tensile simulation results, including fitting the strain transmission loss parameters c1, c2, c3, c4, c5 and c6 according to the tensile simulation results obtained from all the tensile simulations in the following function form: The parameters η, S, ξ, and m are determined according to the sample design parameters of the finite element model for the tensile simulation and are: in, E a is the local elastic modulus of the area on the soft substrate where the strain sensor is attached and S a =w s ×t s +w b ×t b +w f ×t f , S f =w f ×t f , S b =w b ×t b ; Determine an average strain transfer rate α based on actual design parameters of the soft manipulator and actual design parameters of the strain sensor fabricated on the soft manipulator, combined with the strain transfer loss parameter; The average strain transfer rate α is used to correct the strain measurement value ε of the strain sensor made on the soft manipulator to obtain the corrected strain measurement value The grasping state of the soft manipulator is controlled according to the corrected strain measurement value ε′.

2. The grasping control method according to claim 1, characterized in that: The finite element model of the soft substrate and strain sensor integrated together is constructed according to the design parameters of each set of samples, including: According to the design parameters of each set of samples, finite element models were constructed in which the soft substrate and strain sensor were integrated using different sensor integration methods. The sensor integration methods include surface integration and embedded integration. The strain sensor in the finite element model constructed using the surface integration method was located on the surface of the soft substrate, while the strain sensor in the finite element model constructed using the embedded integration method was embedded in the groove opened on the surface of the soft substrate.

3. The grasping control method according to claim 1, characterized in that: The tensile simulation results obtained by performing tensile simulation on each constructed finite element model include: The soft substrate in the finite element model is fixed at one end along the length direction of the sensitive layer in the strain sensor, and a uniform tensile force is applied to the other end to generate a tensile strain ε. app , measure the tensile strain ε(z) of the sensitive layer in the strain sensor along the tensile direction, and follow The average strain ε of the sensitive layer along the tensile direction in the strain sensor is calculated avg , l is half the length of the sensitive layer in the strain sensor.

4. The grasping control method according to claim 1, characterized in that: Determining the average strain transfer rate α includes: The parameters η0, S0, ξ0, and m0 are determined according to the actual design parameters of the soft manipulator and the actual design parameters of the strain sensor made on the soft manipulator, and the average strain transfer rate is calculated by combining the strain transmission loss parameters c1, c2, c3, c4, c5, and c6 obtained by fitting.

5. The grasping control method according to claim 4, characterized in that: The actual design parameters of the soft manipulator include the thickness t of the soft manipulator s ′, width w of the soft manipulator s ′, Young's modulus E of the soft manipulator s ′ and the shear modulus G of the soft manipulator s '; Determining the parameters η0, S0, ξ0, and m0 according to the actual design parameters of the soft manipulator and the actual design parameters of the strain sensor manufactured on the soft manipulator includes calculating according to the following formula: in, E a ′ is the local elastic modulus of the area where the strain sensor is attached to the soft manipulator and S a ′=w s ′×t s ′+w b ×t b +w f ×t f , S f =w f ×t f , S b =w b ×t b .

6. The grasping control method according to claim 1, characterized in that: The measuring and determining of actual design parameters of the strain sensor includes: The width w of the sensitive layer in the fabricated strain sensor was determined using an ultra-depth of field microscope. f , thickness t f The width w of the insulating layer in the strain sensor was determined by optical microscope observation. b and thickness t b ; The material of the sensitive layer is printed into a standard tensile test and a tensile test is performed to determine the Young's modulus E of the sensitive layer. f and the shear modulus G of the sensitive layer f ; Use the material of the insulation layer to print a standard tensile test to perform a tensile test to determine the Young's modulus E of the insulation layer b and the shear modulus of the insulation layer G b .

Images