A six-dimensional force tactile sensor based on FBG, mediation method and preparation method
By acquiring the center wavelength drift of the FBG sensor for temperature compensation and linear model demodulation, the problems of sensor assembly difficulties and temperature drift were solved, and high-precision six-dimensional force and torque measurement was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN FEIBOSUN ROBOT TECHNOLOGY CO LTD
- Filing Date
- 2026-03-02
- Publication Date
- 2026-06-09
AI Technical Summary
Existing FBG-based six-dimensional force sensors require the arrangement of a large number of gratings in a flexible body or the use of complex three-dimensional winding structures, which leads to assembly difficulties, reduced reliability, and is not conducive to mass production. At the same time, temperature drift is difficult to compensate effectively, affecting measurement accuracy.
A demodulation method based on FBG for a six-dimensional force tactile sensor is adopted. By acquiring the center wavelength drift of the sensing group and the reference fiber optic component, temperature compensation is performed, and a linear model is established. The linear model is solved using the regularized least squares method. Combined with the sensitivity matrix, high-precision six-dimensional force and torque measurement is achieved.
It achieves high-precision and robust real-time demodulation of six-dimensional force and torque with simplified hardware configuration, eliminates temperature interference, and improves the reliability and measurement accuracy of the sensor.
Smart Images

Figure CN122171076A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of tactile sensing technology, specifically to a six-dimensional force tactile sensor based on FBG, a modulation method, and a fabrication method. Background Technology
[0002] Traditional six-dimensional force sensors mostly use strain gauges, capacitors, or piezoelectric structures. Although they are widely used in the field of industrial robots, they have certain limitations in terms of flexible integration, miniaturization, and resistance to electromagnetic interference.
[0003] Fiber Bragg grating (FBG) sensors have advantages such as small size, light weight, resistance to electromagnetic interference and easy reuse, and have been gradually introduced into the fields of multidimensional force and tactile sensing in recent years.
[0004] In the process of realizing this invention, the inventors discovered that existing FBG-based six-dimensional force sensors usually require a large number of gratings to be arranged in a flexible body or to adopt a complex three-dimensional winding structure, which leads to assembly difficulties, reduced reliability, and is not conducive to mass production; at the same time, temperature drift is often difficult to compensate effectively, affecting measurement accuracy. Summary of the Invention
[0005] One of the objectives of this invention is to provide a six-dimensional force tactile sensor based on FBG and its fabrication method, in order to solve the shortcomings of existing six-dimensional force sensors based on FBG, which usually require a large number of gratings to be arranged in a flexible body or to adopt a complex three-dimensional winding structure, resulting in assembly difficulties, reduced reliability, and unfavorable mass production.
[0006] To solve the above-mentioned technical problems, the embodiments of the present invention are implemented as follows: Firstly, a demodulation method for a six-dimensional force tactile sensor based on FBG is provided, the steps of which include: The first center wavelength drift of each sensing fiber in the sensing group and the temperature center wavelength drift of the reference fiber are obtained. The temperature change is obtained by calculating the temperature center wavelength drift, and then the temperature change is compensated for the first center wavelength shift to obtain the wavelength drift vector of the sensing group. The sensitivity matrix is obtained by calibrating the sensing group, and the wavelength drift vector is combined with the sensitivity matrix to establish a linear model; The linear model is solved by regularized least squares method to obtain the estimated values of the six-dimensional force and torque vectors.
[0007] The second aspect also discloses a method for fabricating a six-dimensional force-tactile sensor based on FBG, the steps of which include: Cut the end faces of the sensing fiber and the reference fiber using a fiber optic cleaver, and clean the fiber surface with anhydrous ethanol. Femtosecond lasers were used to photolithographically write gratings on the sensing fiber optic component and the reference fiber optic component. A flexible tactile elastomer mold with fiber optic channels was fabricated, and fiber optic grooves were machined in a rigid substrate for arranging reference fiber optic components. The sensing fiber component is embedded in the flexible tactile elastomer along the prefabricated fiber channel and a slight pretension is applied. At the same time, the reference fiber component is placed horizontally in the fiber groove and fixed. Liquid elastomer is poured into the mold, and after vacuum degassing, it is cured in an environment of 60~80℃. After curing, the flexible tactile elastomer is connected to the rigid substrate, and the sensing fiber optic component and the reference fiber optic lead-out end are connected to the optical path system.
[0008] The third aspect also discloses a six-dimensional force tactile sensor based on FBG, including: a rigid substrate; A flexible elastomer is disposed on the side of the rigid substrate; A sensing group is disposed within the flexible elastic body and is arranged corresponding to the sensing area; The sensing group includes three sensing fiber optic components, which are evenly arranged around the center of the sensing area. One end of each sensing fiber optic component extends from the side of the flexible elastomer close to the rigid substrate to the side away from the rigid substrate, and is used to detect forces and torques from different directions. And, the reference fiber optic component is located within a rigid substrate.
[0009] The beneficial effects of the above-described technical solutions provided in the embodiments of the present invention include at least the following: This invention discloses a six-dimensional force-tactile sensor based on FBGs. Three working FBGs are uniformly distributed circumferentially within a flexible tactile elastomer at a 120° in-plane angle and obliquely arranged relative to a rigid substrate at approximately a 45° angle. This allows the three working FBGs to generate different axial strain responses when the tactile working surface is subjected to force, enabling the sensing of triaxial forces Fx, Fy, and Fz, and triaxial moments Mx, My, and Mz. The ambient temperature change is estimated by compensating for the wavelength drift of the temperature-compensated FBGs, and temperature compensation is applied to the wavelength drift of the three working FBGs to obtain a wavelength drift vector related to the six-dimensional load.
[0010] In this embodiment, synchronous acquisition of multi-source optical signals is achieved by obtaining the center wavelength drift of the sensing group and the reference fiber optic component. The temperature change is calculated based on the temperature center wavelength drift, and the wavelength drift of the sensing group is compensated accordingly, effectively eliminating temperature interference and obtaining the wavelength drift vector of the pure mechanical response. A linear mapping model between Δλ′ and the six-dimensional load is established using the calibrated sensitivity matrix, characterizing the differentiated response characteristics of the three tilted FBGs to the six-dimensional load under a specific geometric arrangement. Finally, the regularized least squares method is used to solve the underdetermined model, ensuring the stability and physical rationality of the solution while avoiding pseudo-inverse ill-conditioning. The entire technical chain, from front-end signal acquisition, mid-end temperature-strain decoupling, to back-end robust solution of the underdetermined system, is progressive and interconnected, jointly supporting high-precision and highly robust real-time demodulation of six-dimensional force / torque using only three working FBGs and one reference FBG with a simplified hardware configuration.
[0011] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.
[0012] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description
[0013] Figure 1 A partial cross-sectional view of a six-dimensional force-tactile sensor based on FBG provided for an embodiment of the present invention; Figure 2 This is an embodiment of the present invention. Figure 1 A diagram from another perspective; Figure 3 This is a schematic diagram of an embodiment of the present invention, specifically an induction group. Figure 4 This is a schematic diagram of a demodulation system according to an embodiment of the present invention; Figure 5 This is a superimposed curve of the FBG reflectance spectrum; Figure 6 The calibration relationship curve between Fx and the characteristic wavelength drift Δλ_Fx is shown in the figure. Figure 7 The calibration relationship curve between Fy and the characteristic wavelength drift Δλ_Fy is shown in the figure. Figure 8 The calibration relationship curve between Fz and the characteristic wavelength drift Δλ_Fz is shown in the figure. Figure 9 The calibration relationship curve between Mx and the characteristic wavelength drift Δλ_Mx is shown in the figure. Figure 10The calibration relationship curve between My and the characteristic wavelength drift Δλ_My; Figure 11 The graph shows the calibration relationship between Mz and the characteristic wavelength drift Δλ_Mz.
[0014] Markings in the image: 10. Sensing fiber optic component; 11. Reference fiber optic component; 20. Rigid substrate; 30. Flexible elastomer; 40. Broadband light source; 50. 1×2 coupler; 60. 1×4 coupler; 70. Single-channel FBG demodulator; F. Sensing area; H. Angle. Detailed Implementation
[0015] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0016] In this application, the term "exemplary" is used to mean "used as an example, illustration, or description." Any implementation described as "exemplary" in this application is not necessarily to be construed as being more preferred or advantageous than other implementations. The following description is provided to enable any person skilled in the art to implement and use this application. Details are set forth in the following description for illustrative purposes. It should be understood that those skilled in the art will recognize that this application can be implemented without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of this application with unnecessary detail. Therefore, this application is not intended to be limited to the embodiments shown, but rather to be consistent with the broadest scope of the principles and features disclosed in this application. A six-dimensional force sensor is a key sensor commonly used in industrial production control and robot motion, and its measurement range includes three torque components (Mx, My, and Mz) and three force components (Fx, Fy, and Fz). As a sensor that measures the forces and torques exerted by a robot end effector when it comes into contact with the external environment or grasps a workpiece, the six-dimensional force and torque sensor provides force sensing information for the robot's force and motion control, playing an important role in realizing robot intelligence. It enables precise measurement of the robot's force and motion control, thereby improving the robot's operational accuracy and intelligence level.
[0017] Fiber Bragg Grating (FBG) is an optical sensing unit that operates based on the principle of Bragg grating. The working channel can refer to the optical signal path that is independently identified, distinguished and acquired in the demodulation system.
[0018] Before describing the embodiments of the present invention in detail, the design concept of the present invention will be summarized below.
[0019] This invention provides a demodulation method for a six-dimensional force / tactile sensor based on FBG, the specific implementation of which is as follows. The core of this invention's concept includes: synchronous acquisition of multi-source optical signals by obtaining the center wavelength drift of the sensing group and the reference fiber optic component 11; calculation of temperature change based on the temperature center wavelength drift, and compensation for the wavelength drift of the sensing group accordingly, effectively eliminating temperature interference and obtaining the wavelength drift vector of the pure mechanical response; establishment of a linear mapping model between Δλ′ and the six-dimensional load using the calibrated sensitivity matrix, characterizing the differentiated response characteristics of three tilted FBGs to the six-dimensional load under a specific geometric arrangement; and finally, using the regularized least squares method to solve the underdetermined model, ensuring the stability and physical rationality of the solution while avoiding pseudo-inverse ill-conditioning. The entire technical chain, from front-end signal acquisition, mid-end temperature-strain decoupling, to back-end robust solution of the underdetermined system, is progressive and interconnected, jointly supporting high-precision and highly robust real-time demodulation of six-dimensional force / torque using only three working FBGs and one reference FBG with a simplified hardware configuration.
[0020] For the first aspect, which is based on the same inventive concept, please refer to the appendix. Figure 1 and 2 The diagram also discloses a demodulation method for a six-dimensional force haptic sensor based on FBG, the steps of which include: Step 01: Obtain the first center wavelength drift of each sensing fiber element 10 in the sensing group, and the temperature center wavelength drift of the reference fiber element 11.
[0021] The sensing group can refer to a collection of multiple sensing fiber optic components 10 disposed within the flexible elastomer 30 for sensing six-dimensional mechanical loads. Here, the sensing group includes three or more sensing fiber optic components 10; the following description uses three sensing fiber optic components 10 as an example. The sensing fiber optic component 10 is an FBG. The first center wavelength drift can refer to the offset of the center wavelength of the reflection spectrum of each sensing fiber optic component 10 relative to its initial unloaded and temperature-free state under the combined action of force and temperature. In this embodiment, it is denoted as... , , The reference fiber optic component 11 can refer to a temperature compensation grating disposed within a rigid substrate 20, which is essentially unaffected by mechanical loads and is only sensitive to changes in ambient temperature; the temperature center wavelength drift can refer to the center wavelength shift of the reference fiber optic component 11 under the same ambient temperature change, denoted as . This acquisition process is achieved using a single-channel FBG demodulator in conjunction with a broadband light source and a multiplexed optical path, such as 1×2 and 1×4 fiber couplers. Wavelength division multiplexing (WDM) technology is used to sequentially scan and identify the reflection peak positions of each of the four gratings, thereby extracting the data. , , and .
[0022] In this application, the center wavelength drift of each fiber optic component can be determined using wavelength division multiplexing (WDM) signal separation and peak fitting. Alternatively, it can be determined using reflectance spectral matched filtering and centroid algorithms. Furthermore, it can be determined using multi-peak Gaussian fitting and iterative optimization. Based on any of the above methods, this application obtains the center wavelength drift data of each sensing fiber optic component 10 and the reference fiber optic component 11 under the current operating conditions, serving as the basis for subsequent temperature compensation and mechanical decoupling.
[0023] In this application, the center wavelength drift of each fiber optic component can be determined using wavelength division multiplexing (WDM) signal separation and peak fitting. Alternatively, it can be determined using reflectance spectral matched filtering and centroid algorithms. Furthermore, it can be determined using multi-peak Gaussian fitting and iterative optimization. Based on any of the above methods, this application obtains the center wavelength drift data of each sensing fiber optic component 10 and the reference fiber optic component 11 under the current operating conditions, serving as the basis for subsequent temperature compensation and mechanical decoupling.
[0024] For example, in this application, under an initial state at room temperature (25°C), a single-channel FBG demodulator measures the center wavelengths of the three sensing fiber optic components 10 to be 1550.499 nm, 1553.522 nm, and 1556.481 nm, respectively, while the reference fiber optic component 11 is measured to be 1559.511 nm. When the ambient temperature rises to 28°C and an external load is applied, the demodulator rescans to obtain the new center wavelengths of the four gratings, and calculates the difference between these new wavelengths and the initial values, thus obtaining the... , , and .
[0025] Step 02: Calculate the temperature change by measuring the temperature center wavelength drift, and then compensate for the temperature change by adjusting the first center wavelength offset to obtain the wavelength drift vector of the sensing group.
[0026] In a further embodiment, the calculation of the temperature change by the temperature center wavelength drift involves substituting the temperature center wavelength drift into a formula to calculate the temperature change of the ambient temperature. The formula is as follows: ,in, This is the temperature sensitivity coefficient. This is the wavelength shift. This refers to the change in temperature. The step of compensating for the temperature change with the first center wavelength shift to obtain the wavelength drift vector of the sensing group is to subtract the temperature change from the total wavelength drift of each sensing fiber element 10 in the sensing group to obtain the wavelength drift vector caused only by the load.
[0027] The temperature change calculated from the temperature center wavelength drift can refer to the temperature-wavelength response relationship of the reference fiber optic component 11, which is then used to calculate the actual temperature change. This is converted to the corresponding ambient temperature change ΔT. This conversion is based on the fundamental principle of FBG temperature sensing, namely, the linear relationship between center wavelength drift and temperature change. Temperature compensation for the first center wavelength shift can be achieved by subtracting the drift component caused by temperature change from the total wavelength drift of each sensing fiber optic element 10, thereby separating the effective wavelength drift caused only by mechanical load. The wavelength drift vector of the sensing group can be a three-dimensional column vector reflecting only the six-dimensional mechanical response after temperature compensation, denoted as... .
[0028] An embodiment may be based on Reverse reasoning based on relationships and substitute This application can also perform channel-by-channel compensation; for example, it can also use a method of joint lookup table and interpolation based on the temperature sensitivity coefficients of multiple channels to perform compensation. Dynamic temperature term stripping is performed; furthermore, this application can also use a method based on online updated temperature drift baseline and sliding window deviation correction to... Real-time temperature compensation is performed. This application obtains the purely mechanical wavelength drift vector after eliminating temperature interference based on any of the above methods. This ensures that it only carries six-dimensional load information: Fx, Fy, Fz, Mx, My, and Mz.
[0029] In this embodiment, a reproducible quantitative inversion mechanism for ΔT is established by combining the temperature center wavelength drift of the reference fiber optic component 11 with the calibration coefficient. Then, based on the differentiated temperature response characteristics of each sensing fiber optic component 10, the measured total drift is specifically subtracted, and the final output is a wavelength drift vector caused only by the load. This collaborative process achieves accurate identification and channel-level separation of temperature crosstalk, ensuring the physical purity and engineering reproducibility of the wavelength drift vector, and providing a pollution-free, high-fidelity original data foundation for subsequent sensitivity matrix modeling and regularization solution in this application.
[0030] For example, this application may use a known reference fiber optic component 11 with a temperature sensitivity coefficient. , measured Then ΔT = 3 ℃; if the temperature sensitivity coefficients of the three sensing optical fibers 10FBG1, FBG2 and FBG3 are respectively , , And its actual measurement , , Then after compensation, , , The final composition is [14.4, 8.6, 20.7]ᵀ pm .
[0031] Step 03: Calibrate the sensing group to obtain the sensitivity matrix, combine the wavelength drift vector with the sensitivity matrix, and establish a linear model.
[0032] The sensitivity matrix obtained by the induction group calibration can refer to the following: during the calibration phase, a known uniaxial load is applied to each independent degree of freedom (Fx, Fy, Fz, Mx, My, Mz) in the six-dimensional load space, and the corresponding wavelength drift response is recorded simultaneously. A 3×6-dimensional mapping matrix S is constructed through linear regression or least squares fitting, where each column characterizes the combined influence weight of a certain load component on the 10-wavelength drift of the three sensing fiber components. Combining the wavelength drift vector with the sensitivity matrix can refer to using the obtained Δλ′ as the observation output and S as the system response parameter to establish a linear equation with the six-dimensional load W as the unknown variable. The linear model can refer to the following: under the assumptions of small deformation and linear elasticity, this mapping relationship can be expressed as... ,in The noise vector is measured. This model demonstrates the differentiated coupling response mechanism of three tilted sensing fiber components 10 under a geometric constraint of 120° circumferential distribution and approximately 45° tilt angle to a six-dimensional load.
[0033] This application can construct S, for example, using a method of uniaxial successive loading and least squares fitting; it can also construct S, for example, using a method of multiaxial collaborative loading and orthogonal experimental design; further, it can also construct S, using a method of physical modeling initial value guidance and iterative correction based on measured data. This application obtains a sensitivity matrix S with good condition number and physical interpretability based on any of the above methods, making... A stable and reversible mathematical relationship is formed between W and W.
[0034] For example, loads of Fx = 1N, 2N, and 3N can be applied sequentially on a calibration platform, and the corresponding Δλ′ samples can be recorded. The first column of S can then be fitted. Similarly, complete the calibration of the remaining five columns, finally obtaining S = [[9.43, 1.02, 5.21, 4.33, 1.15, 3.07], [1.02, 12.56, 4.18, 2.09, 5.32, 4.26], [5.21, 4.18, 11.74, 10.22, 8.41, 6.35]] (unit: pm / unit load). The above results... Substitution This constitutes the linear model to be solved.
[0035] In a further embodiment, the step of combining the wavelength drift vector with the sensitivity matrix calibrated by the sensing group and establishing a linear model is as follows: The sensitivity matrix is obtained by applying known loads to the sensing group from six directions. The six directions can refer to the six degrees of freedom components of a six-dimensional force / torque, i.e., three-dimensional translational forces. , , and triaxial torque , , ; In one embodiment, the elements of each column of the sensitivity matrix S are determined by means of applying a uniaxial load and acquiring the corresponding wavelength drift. Alternatively, this application may determine the elements of each column of the sensitivity matrix S by controlling a mechanical loading platform to apply a stepped force / torque along a specified degree of freedom direction according to a preset load sequence, and simultaneously recording the wavelength drift response of the three sensing fiber optic components 10. Furthermore, this application may also determine the elements of each column of the sensitivity matrix S by outputting a load signal with known amplitude and direction through a high-precision six-dimensional force calibration stage, stabilizing the load value under closed-loop feedback, and acquiring steady-state wavelength drift data.
[0036] In this embodiment, a six-dimensional force-tactile sensor based on FBG is fixed on a calibration platform, with the sensing area F facing the loading direction; standard weights or servo actuators are sequentially applied along the positive x-axis with pure force loads of 0.5 N, 1.0 N, 1.5 N, and 2.0 N. The remaining degrees of freedom are strictly constrained; the wavelength drift of the three sensing fiber optic components 10 is synchronously collected after temperature compensation under each loading. , , ;right Load and each Perform linear fitting to obtain the slope , , The first column of the sensitivity matrix S ; and so on, until completion. , , , , A total of six single-axis calibrations were performed, which were then spliced together to form a 3×6 sensitivity matrix S, the mathematical form of which is as follows: .
[0037] In a further embodiment, the rows of the sensitivity matrix correspond to the working channels of the sensing fiber optic device 10, and the columns correspond to the six components of the six-dimensional force. As shown in the sensitivity matrix S above, each row represents a working channel of the sensing fiber optic device 10, i.e., the first row is FBG1, the second row is FBG2, and the third row is FBG3; the columns represent... , , , , There are a total of six single-axis calibrations.
[0038] like Figure 7-11 The single-axis calibration curves for the six axes of FBG1Fx, Fy, Fz, Mx, My, and Mz are shown. The characteristic wavelength drift of the three force components Fx, Fy, and Fz, and the three torque components Mx, My, and Mz are approximately linearly related to the load, which is significantly smaller than the random fluctuations in the combined load experiment. This demonstrates the high sensitivity and high signal-to-noise ratio characteristics of the structure when measuring small torques.
[0039] according to Figure 6-11 The linear fitting results yield the equivalent sensitivity for each axis, i.e., the slope of the characteristic wavelength drift with respect to the load. Table 1 shows the sensitivity fitting values for the six axes: Fx, Fy, Fz, Mx, My, and Mz.
[0040]
[0041] In this embodiment, the wavelength drift vector is combined with the sensitivity matrix to obtain a linear model; the linear model is as follows: ,in, Let W be the compensated wavelength drift vector, and W be the six-dimensional force / torque vector to be determined. This is the sensitivity matrix; To measure the noise vector.
[0042] In an example, the temperature-compensated wavelength drift vector was obtained during an actual measurement. ; Call up the sensitivity matrix S obtained from calibration; Substitute it into the model That is, to construct an overdetermined linear system with 3 equations and 6 unknowns, providing input for subsequent regularized least squares solution.
[0043] In the embodiment, the physical driven calibration of the sensitivity matrix S and the wavelength drift vector S-structure alignment and combination, and noise terms The linear model inherits the experimental reliability of uniaxial calibration while preserving the integrity of the six-dimensional coupled response, avoiding the simplification distortion introduced by theoretical modeling; based on this, a model was developed... The relationship forms the only mathematical bridge from optical measurement signals to six-dimensional mechanical vector calculations, supporting the feasibility and accuracy of regularized least squares solutions.
[0044] Step 04: Solve the linear model using the regularized least squares method to obtain the estimated value of the six-dimensional force / torque vector.
[0045] In a further embodiment, the method of solving the linear model using regularized least squares involves constructing an objective function from the linear model formula using regularized least squares, then differentiating the objective function to derive the analytical solution formula, and finally obtaining the estimated value of the six-dimensional force / torque vector based on the analytical solution formula.
[0046] Regularized least squares can refer to introducing regularized least squares into the traditional least squares objective function. Norm regularization terms are used to suppress noise amplification and matrix ill-conditioned effects during the solution process of underdetermined systems; this method does not rely on pseudo-inverse operations on the sensitivity matrix S, but rather ensures the numerical stability of the solution by constructing a constrained optimization problem; the estimated value of the six-dimensional force / moment vector can refer to the solution obtained from the solution. Each of its components is the optimal unbiased estimate of the true load.
[0047] The derivative of the objective function is as follows: In practice, this can be achieved using several sets of known loads. and the corresponding measurement wavelength drift Constructing the calibration dataset: Let It is a 3×N matrix. Given a 6×N matrix, the sensitivity matrix S can be estimated using the following formula in the least squares sense: .
[0048] For measurements under unknown load conditions, given the temperature-compensated wavelength drift vector Δλ', the six-dimensional load estimation vector Ŵ should satisfy the following: .because It is a 3×6 matrix, and the system of equations is underdetermined.
[0049] To obtain a unique and stable solution from an underdetermined system, the objective function is solved using the Tikhonov regularization framework: , Among them, the first item The second term represents the residual between the model's predicted values and the actual measured values, and is required to be minimized. This is the regularization term. Here, α>0 is the regularization parameter, used to suppress noise and avoid dedivergence.
[0050] When α→0 and When the solution has full row rank, it degenerates into a weighted least squares pseudo-inverse, leading to the derivation of the analytical solution formula: , in, This is the transpose of the sensitivity matrix. It is a 6×6 identity matrix.
[0051] In this embodiment, a six-dimensional demodulation algorithm is used for calculation. The specific steps include: First, during the calibration phase, a sufficient number of sets of six-dimensional loads are collected. and corresponding ,calculate Next, during the online measurement phase, ΔT is calculated based on the FBG wavelength change of the reference fiber optic component 11, and temperature compensation is performed on the FBGs of the three sensing fiber optic components 10 to obtain... Furthermore, based on the pre-selected α and ,according to Calculate the six-dimensional load estimate; finally, optionally... Perform physical constraint corrections, such as saturation and symmetry constraints, to improve the physical consistency of the results.
[0052] For example: we could take the regularization parameter α = 0.01, and substitute S and Δλ′ into... Calculated The results are in high agreement with the actual applied load, verifying the effectiveness of temperature compensation and regularization decoupling.
[0053] In this embodiment, by acquiring the center wavelength drift of the sensing group and the reference fiber optic component 11, synchronous acquisition of multi-source optical signals is achieved; the temperature change is calculated based on the temperature center wavelength drift, and the wavelength drift of the sensing group is compensated accordingly, effectively eliminating temperature interference and obtaining the wavelength drift vector of the pure mechanical response; a sensitivity matrix obtained through calibration is used to establish... A linear mapping model between the model and the six-dimensional load characterizes the differentiated response characteristics of three tilted FBGs under a specific geometric arrangement to the six-dimensional load. Finally, a regularized least squares method is used to solve the underdetermined model, ensuring the stability and physical rationality of the solution while avoiding pseudo-inverse ill-conditioned conditions. The entire technical chain, from front-end signal acquisition and mid-end temperature-strain decoupling to robust solution of the underdetermined system at the back end, is progressive and interconnected, jointly supporting high-precision and robust real-time demodulation of six-dimensional force / torque using only a simplified hardware configuration of three working FBGs and one reference FBG.
[0054] Based on the same inventive concept, a second aspect also discloses a method for fabricating a six-dimensional force-tactile sensor based on FBG, the steps of which include: Step 10: Cut the end faces of the sensing fiber optic component 10 and the reference fiber optic component 11 with a fiber optic cleaver, and clean the fiber optic surface with anhydrous ethanol.
[0055] In this embodiment, cutting the end faces of the sensing fiber optic component 10 and the reference fiber optic component 11 with a fiber optic cleaver can refer to making a flat cut perpendicular to the fiber axis at the end of the sensing fiber optic component 10 and the reference fiber optic component 11 to be connected, in order to ensure the end face quality of the subsequent grating writing area and the subsequent optical path coupling efficiency. Cleaning the fiber surface with anhydrous ethanol can refer to wiping or ultrasonically cleaning the fiber cladding surface with anhydrous ethanol with a concentration of not less than 99.7% to remove residual particles, grease, and oxides from the cutting process, thus avoiding a decrease in grating reflectivity or a shift in center wavelength due to surface contamination during femtosecond laser writing. This step is the fiber pretreatment stage, and its function is to provide a clean, flat, and optically uniform fiber substrate for subsequent femtosecond laser writing, ensuring the integrity of the grating structure and the stability of wavelength accuracy.
[0056] For example: Three 20cm long single-mode optical fibers can be used as sensing optical fiber components 10, and one 20cm long single-mode optical fiber can be used as reference optical fiber components 11; use a high-precision optical fiber cleaver to cut one end of each fiber vertically, and visually inspect the end face for any chipping or cracks; then place the four optical fibers in a beaker containing anhydrous ethanol, gently wipe the cladding surface with a cotton swab, and then dry it with high-purity nitrogen to complete the pretreatment.
[0057] Step 20: Use femtosecond laser to photolithographically write gratings on the sensing fiber optic component 10 and the reference fiber optic component 11.
[0058] In this embodiment, the use of femtosecond lasers to photolithographically write gratings on the sensing fiber element 10 and the reference fiber element 11 refers to focusing a femtosecond laser pulse inside the fiber core and inducing periodic modulation of the refractive index through nonlinear multiphoton absorption, thereby directly writing a Fiber Bragg Grating (FBG) into the fiber. This process does not require stripping the coating layer and has advantages such as high writing accuracy, small heat-affected zone, and good grating stability. The sensing fiber element 10 corresponds to three sensing fiber elements 10FBGs with initial center wavelengths of 1550.499 nm, 1553.522 nm, and 1556.481 nm, respectively; the reference fiber element 11 corresponds to one temperature-compensated FBG with an initial center wavelength of 1559.511 nm; the length of all gratings is controlled within the range of 1–3 mm, and the reflectivity is approximately 70%. The purpose of this step is to precisely construct FBG arrays with distinguishable center wavelengths on different optical fibers, forming the basis for wavelength division multiplexing (WDM) required for subsequent demodulation, while ensuring that each grating has sufficient signal-to-noise ratio and spectral resolution during the sensing process.
[0059] For example: The three pre-processed sensing fiber components 10 are sequentially mounted on a femtosecond laser writing platform. The laser power is set to 2.5mW, the scanning speed to 5μm / s, and the writing length to 2mm. Working FBGs with center wavelengths of 1550.499nm, 1553.522nm, and 1556.481nm are written respectively. Then, the reference fiber component 11 is mounted, and the reference fiber component 11 FBG with a center wavelength of 1559.511nm is written with the same parameters. After writing, the reflection spectrum of each grating is detected using a spectral analyzer to confirm that the center wavelength deviation is less than ±0.01nm and the reflectivity fluctuation range is between 65% and 75%.
[0060] In one embodiment, a preset wavelength interval threshold is set to be greater than the resolution of the demodulator, thereby constraining the writing process parameters of each grating to identify and separate the reflection signals of the four fiber Bragg gratings. Furthermore, the reflection signals of the four fiber Bragg gratings are identified and separated by dynamically allocating the center wavelength positions of the four gratings based on the spectral coverage of the broadband light source and the scanning step size of the demodulator.
[0061] Step 30: Fabricate a flexible tactile elastomer mold with fiber optic channels, and machine fiber optic grooves in the rigid substrate 20 for arranging reference fiber optic components 11.
[0062] In this embodiment, the flexible tactile elastomer mold with fiber optic channels can refer to a female mold used for molding the flexible tactile elastomer. It has three inclined fiber optic channels inside, with the channel axes forming an angle H of 30°-60° with the bottom surface of the mold, i.e., the future mating surface between the flexible elastomer 30 and the rigid substrate 20 (approximately 45°). The three channels are evenly distributed circumferentially within a plane of 120° in the mold cross-section. Machining the fiber optic groove in the rigid substrate 20 can refer to milling a horizontal, straight fiber optic groove onto the rigid substrate 20 made of metal or hard alloy using CNC precision machining. The groove width is slightly larger than the outer diameter of the fiber, for example, 260 μm, and the groove depth is adapted to the fiber cladding diameter, used for precise positioning and fixing of the reference fiber optic component 11. The purpose of this step is to establish a dual-reference spatial positioning system: the inclined channels define the spatial configuration of the three sensing fibers, and the horizontal fiber optic groove defines the spatial reference of the reference fiber. These two elements work together to ensure the repeatability and manufacturing consistency of the overall sensor geometry, which is a physical prerequisite for achieving six-dimensional force decoupling modeling.
[0063] It should be noted that, in order to ensure that the sensing fiber optic component 10 is embedded in the flexible tactile elastomer, the central axis of the fiber optic channel is not parallel to the plane of the rigid substrate 20, i.e., a non-zero angle H is formed between its axis and the plane; this angle H is an acute angle, which can be 30°, 45°, or 60°, but is 45°; this tilted orientation allows the sensing fiber optic component 10 to simultaneously respond to normal compression / tension deformation along the z-direction and shear / bending deformation along the x and y directions when subjected to force. Three tilted fiber optic channels are provided inside the flexible tactile elastomer for mounting and fixing the sensing fiber optic component 10. For example, a flexible elastomer 30 mold master plate made of ABS material is designed and 3D printed. Its internal cavity is cylindrical with a size of Φ12 mm × 10 mm. Three stainless steel guide pins with a diameter of 200 μm are embedded inside. The angle H between the pin axis and the bottom surface is 45°, and they are 120° to each other in the top view projection. The master plate is then used to make a PDMS mold. At the same time, a horizontal fiber optic groove with a length of 18 mm, a width of 260 μm, and a depth of 125 μm is CNC machined on a 6061 aluminum alloy substrate. Φ1.5 mm positioning pin holes are machined at both ends of the groove. Finally, the PDMS mold is precisely assembled with the aluminum alloy substrate through the positioning pins to form an integrated packaging reference platform.
[0064] Step 40: Embed the sensing fiber component 10 into the flexible tactile elastomer along the prefabricated fiber channel and apply a slight pretension. At the same time, place the reference fiber component 11 horizontally in the fiber groove and fix it. Pour liquid elastomer into the mold, degas it under vacuum, and then cure it in an environment of 60~80℃.
[0065] Embedding the sensing fiber optic component 10 into the flexible tactile elastomer along the prefabricated fiber optic channel and applying slight pretension can refer to sequentially inserting the three segments of sensing fiber optic component 10, which have been written into the FBG, into the three inclined channels inside the mold, so that the far end of the fiber is close to the top of the mold (i.e., the future tactile working surface position), and the near end is led out from the bottom of the mold, and a constant tension of 10-50mN is applied to the lead-out end to eliminate fiber slack and ensure that it is always in an effective sensing state during the deformation of the elastomer; placing the reference fiber optic component 11 horizontally in the fiber optic groove and fixing it can refer to placing the reference fiber optic component 11, which has been written into the FBG, straight into the horizontal fiber optic groove of the rigid substrate 20, and fixing both ends with UV adhesive dots; pouring liquid elastomer into the mold, degassing under vacuum, and curing in an environment of 60-80℃ can refer to injecting liquid silicone rubber or polyurethane precursor into the cavity formed by the mold and the substrate, evacuating to ≤10Pa to remove air bubbles, and then placing it in a constant temperature oven and keeping it at 60-80℃ for 2-4 hours to complete cross-linking and curing. The purpose of this step is to achieve integrated encapsulation of the sensing fiber and the flexible elastomer 30, as well as rigid integration of the reference fiber and the rigid substrate 20. Through pre-tension control and temperature-time synergistic curing process, a composite structure with Shore A hardness of 20-40, strong interfacial bonding and no debonding defects is obtained while ensuring the thermal stability of FBG.
[0066] For example, three sensing fiber optic components 10 with FBG already written are inserted into three inclined channels of the PDMS mold. The distance between the far end of the fiber and the top of the mold is 1 mm, and the near end is led out from the bottom of the mold and connected to a micro tension controller. The pretension is set to 35 mN. The reference fiber optic component 11 is placed horizontally in the fiber groove of the aluminum alloy substrate. NOA61 UV glue is applied to both ends and cured by UV irradiation for 30 seconds. Then the mold and the substrate are assembled in place. The degassed KE-106 silicone rubber precursor is injected into the cavity using an injection pump. After vacuuming for 15 minutes, it is transferred to a 75°C oven for curing for 3 hours. After curing, it is taken out and observed that the fiber has no obvious bending, the elastomer has no bubbles, and the interface has no debonding.
[0067] Step 50: After curing, connect the flexible tactile elastomer to the rigid substrate 20, and connect the sensing fiber optic component 10 and the reference fiber optic lead-out end to the optical path system.
[0068] In this embodiment, connecting the flexible tactile elastomer to the rigid substrate 20 after curing can refer to physically assembling the flexible tactile elastomer to the rigid substrate 20 with the integrated reference fiber optic component 11 after the flexible tactile elastomer has been cured and separated from the mold. The connection method includes bonding or mechanical fixing. Bonding can use two-component epoxy resin or silicone, apply a pressure of 0.1 to 0.3 MPa at the interface and cure at room temperature for 24 hours. Mechanical fixing can use M2 micro screws with countersunk holes for fastening, with no less than two screws distributed on both sides of the substrate. Connecting the sensing fiber optic component 10 and the reference fiber optic lead-out end to the optical path system can refer to leading out the four fiber optic lead-out ends through a bend-resistant protective sleeve and forming a complete optical path system with a broadband light source, a 1×2 fiber optic coupler, a 1×4 fiber optic coupler and a single-channel FBG demodulator through a standard FC / APC interface. The purpose of this step is to complete the final structural integration and functional closed loop of the sensor, ensuring the optical path independence, mechanical stability and signal readability of the sensing fiber and the reference fiber, so that the prepared sensor can be directly connected to the demodulation system described above to carry out six-dimensional force measurement.
[0069] For example: Place the cured and demolded PDMS flexible elastomer 30 on top of an aluminum alloy substrate, uniformly coat the contact surface of the two with a layer of SE1700 adhesive with a thickness of about 50μm, apply a pressure of 0.2MPa and let it stand for 24 hours; then bundle the four optical fiber leads, insert them into Φ3mm stainless steel braided armor sleeves, install FC / APC connectors at the ends, and connect them to a demodulation system consisting of an ASE broadband light source (1520–1570nm), 1×2 couplers (split ratio 50:50), 1×4 couplers (split ratio 25:25:25:25), and a single-channel FBG demodulator (wavelength resolution 1pm) to complete the assembly of the whole machine.
[0070] The aforementioned fabrication method enables high-precision, high-consistency, and mass-producible integrated packaging of three tilting sensing fiber components 10 and one horizontal reference fiber component 11 within a flexible tactile elastomer and a rigid substrate 20. Specifically, fiber end-face treatment and surface cleaning ensure grating writing quality, femtosecond laser writing enables precise construction of multi-wavelength FBGs, the mold with tilting channels and the substrate with horizontal grooves jointly establish a dual spatial configuration reference, pre-tension control and gradient curing processes balance fiber mechanical stability and material cross-linking integrity, and the flexible body-substrate connection and optical path docking complete the functional closed loop from device to system. This fabrication method directly produces the physical entity of the aforementioned six-dimensional force tactile sensor, serving as the material foundation and technological guarantee for the entire invention to move from design drawings to engineering applications.
[0071] Based on the same inventive concept, a third aspect also discloses a six-dimensional force tactile sensor based on FBG, comprising: a rigid substrate 20; A flexible elastomer 30 is disposed on the side of the rigid substrate 20; The sensing group is located within the flexible elastomer 30 and is positioned corresponding to the sensing area F; The sensing group includes three sensing fiber optic components 10, which are evenly arranged around the center of the sensing area F. One end of the sensing fiber optic component 10 extends from the side of the flexible elastomer 30 close to the rigid substrate 20 to the side away from the rigid substrate 20, and is used to detect the force and torque from different directions. The reference fiber optic component 11 is disposed within the rigid substrate 20.
[0072] In this embodiment, a physical topology of three tilts and one horizontal structure is constructed: three sensing fiber optic components 10 are evenly distributed circumferentially at 120° within a flexible elastomer 30 and extend obliquely at a 45° tilt angle to form a sensing array with differentiated responses to six-dimensional loads. A reference fiber optic component 11 is independently buried within a rigid substrate 20 to form a dedicated monitoring channel for temperature drift. The two are physically isolated in space and have a clear division of labor in function.
[0073] Working process: When an external load is applied to the surface of the flexible elastomer 30, the flexible elastomer 30 undergoes local deformation, which causes the three tilted sensing fiber components 10 to produce axial strain of different degrees, thereby causing their respective center wavelengths to drift. Since the three sensing fiber components 10 have different spatial orientations, their coupling responses to the six degrees of freedom loads Fx, Fy, Fz, Mx, My, and Mz are different, thus forming a distinguishable information combination in the wavelength domain. At the same time, the rigid substrate 20 has a much lower coefficient of thermal expansion than the flexible elastomer 30, and its deformation is extremely small under the same temperature change. This makes the built-in reference fiber component 11 almost unaffected by mechanical strain, and its wavelength drift only reflects the change in ambient temperature, thereby achieving a natural decoupling between mechanical signals and temperature signals.
[0074] For example: 304 stainless steel is selected as the rigid substrate material 20, and a substrate with dimensions of 25mm × 25mm × 3mm is processed. A horizontal fiber groove with a depth of 0.3mm and a width of 0.25mm is drilled on one side of the substrate. A flexible elastomer mold 30 is prepared using silicone rubber with a Shore A hardness of 30. Three fiber channels with a 45° angle H to the substrate plane and circumferential angles of 0°, 120°, and 240° are pre-fabricated inside the mold. Three single-mode optical fibers are respectively etched with FBGs with center wavelengths of 1550.499nm, 1553.522nm, and 1556.481nm in a femtosecond laser system. The grating length is 2mm and the reflectivity is approximately 70%. Three sensing fiber components 10 are embedded in the uncured silicone rubber along the inclined channels. After applying a pretension of 0.3N, vacuum degassing is performed at 70℃. Curing in an oven for 2 hours; a reference FBG with a center wavelength of 1559.511nm and a length of 1.5mm is horizontally placed into the fiber optic slot of the substrate and fixed with UV adhesive; after the flexible elastomer 30 has cured, it is bonded to the rigid substrate 20 with two-component epoxy adhesive. The four fiber leads are connected to the optical path system consisting of a broadband light source, a 1×2 coupler, a 1×4 coupler, and a single-channel FBG demodulator after being protected by a sheath. In this structure, when a six-dimensional force / torque is applied in any direction by the fingertip, the three sensing FBGs output differentiated wavelength drift, and the reference FBG synchronously outputs temperature drift. The two are input to the demodulation process in concert, and finally output the estimated values of Fx, Fy, Fz, Mx, My, and Mz.
[0075] In this embodiment, the rigid substrate 20 provides a stable platform with a low coefficient of thermal expansion, and the reference fiber optic component 11 can accurately separate temperature drift, thereby avoiding the impact of temperature cross-sensitivity on the demodulation accuracy of the six-dimensional force. The flexible elastomer 30 has three inclined and circumferentially uniformly distributed fiber optic channels inside. The three sensing fiber optic components 10 generate differentiated axial strains when subjected to force, improving the completeness and solvability of the six-dimensional load information. One end of the sensing fiber optic component 10 extends outward from the side of the substrate to the tactile working surface, enhancing the spatial orientation sensing capability of external contact forces. The reference fiber optic component 11 is completely isolated from the sensing group in physical space, ensuring the independence and reliability of the temperature compensation channel. This structure is an adapted hardware carrier for implementing the demodulation method, solving key technical problems in existing multi-dimensional force sensors such as complex structure, difficulty in suppressing temperature drift, and information redundancy or incompleteness.
[0076] For inventions based on the same inventive concept, please refer to Figure 4 The fourth aspect also discloses a demodulation system for a tactile sensor, including a six-dimensional force tactile sensor based on pattern interference fiber. Broadband light source, 1×2 fiber optic coupler, 1×4 fiber optic coupler, and single-channel FBG demodulator.
[0077] In this embodiment, the optical path operation of the demodulation system is as follows: Broadband light emitted from a broadband light source is split into two paths via a 1×2 fiber coupler. One path enters the sensing branch, while the other serves as a reference light and does not participate in sensing. The sensing branch optical signal enters a 1×4 fiber coupler and is distributed to four independent paths, each injected into one of four FBGs. Each FBG generates strong reflection at its corresponding center wavelength, while the remaining wavelengths are transmitted. The reflected light from the four FBGs returns along the original path to the 1×4 fiber coupler, where spatial beam combining is completed at its input port. The combined multi-wavelength reflected light is then fed into a single-channel FBG demodulator via a 1×2 fiber coupler. The demodulator identifies the positions of each reflection peak using a built-in spectral analysis module and calculates the center wavelength values of the four FBGs sequentially or simultaneously. , , , and its relative to the initial value , , , drift amount , , , .
[0078] The embodiment achieves efficient spectral multiplexing between the broadband light source and four FBGs, and completes the separation, acquisition, and synchronous analysis of the four grating signals on a single-channel hardware platform using a two-stage coupler structure. Since only a single-channel demodulator is needed to read the wavelengths of all four gratings, it saves 75% of the photoelectric detection hardware cost compared to the traditional four-channel solution, and significantly reduces the system size, power consumption, and multi-channel calibration complexity. The system architecture is strictly matched with the physical sensors defined in the above embodiment, forming a complete closed loop from optical perception to mechanical calculation. It is the core hardware carrier for the engineering implementation of this invention for robotic tactile applications.
[0079] Based on the same inventive concept, the fifth aspect also discloses an application method of a six-dimensional force tactile sensor based on FBG, which is applied in wearable health monitoring, robot tactile sensing and human-computer interaction and other application scenarios.
[0080] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. This disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims. Thus, if these modifications and variations of the invention fall within the scope of the claims of the invention and their equivalents, the invention is also intended to include these modifications and variations.
Claims
1. A demodulation method for a six-dimensional force tactile sensor based on FBG, characterized in that the steps include... include: The first center wavelength drift of each sensing fiber in the sensing group and the temperature center wavelength drift of the reference fiber are obtained. The temperature change is obtained by calculating the temperature center wavelength drift, and then the temperature change is compensated for the first center wavelength shift to obtain the wavelength drift vector of the sensing group. The sensitivity matrix is obtained by calibrating the sensing group, and the wavelength drift vector is combined with the sensitivity matrix to establish a linear model; The linear model is solved by regularized least squares method to obtain the estimated values of the six-dimensional force and torque vectors.
2. The demodulation method according to claim 1, characterized in that, The method of calculating the temperature change by substituting the temperature center wavelength drift into the formula is as follows: ,in, This is the temperature sensitivity coefficient. This is the wavelength shift. This refers to the change in temperature. The step of compensating for the temperature change in the first center wavelength shift to obtain the wavelength drift vector of the sensing group is to subtract the temperature change from the total wavelength drift of each sensing fiber in the sensing group to obtain the wavelength drift vector caused only by the load.
3. The demodulation method according to claim 1, characterized in that, The step of combining the wavelength drift vector with the sensitivity matrix calibrated by the sensing group and establishing a linear model is as follows: The sensitivity matrix is obtained by applying known loads to the sensing group from six directions. The wavelength drift vector is combined with the sensitivity matrix to obtain a linear model; The linear model is as follows: ,in, Let W be the compensated wavelength drift vector, and W be the six-dimensional force / torque vector to be determined. This is the sensitivity matrix; To measure the noise vector.
4. The demodulation method according to claim 3, characterized in that: The rows of the sensitivity matrix correspond to the working channels of the sensing fiber optic device, and the columns correspond to the six components of the six-dimensional force.
5. The demodulation method according to claim 1, characterized in that, The method of solving the linear model using regularized least squares involves constructing an objective function from the linear model formula using regularized least squares, then differentiating the objective function to derive the analytical solution formula, and finally obtaining the estimated value of the six-dimensional force / torque vector based on the analytical solution formula.
6. A method for fabricating a six-dimensional force-tactile sensor based on FBG, characterized in that the steps include... include: Cut the end faces of the sensing fiber and the reference fiber using a fiber optic cleaver, and clean the fiber surface with anhydrous ethanol. Femtosecond lasers were used to photolithographically write gratings on the sensing fiber optic component and the reference fiber optic component. A flexible tactile elastomer mold with an optical fiber channel was fabricated, and an optical fiber groove was machined in a rigid substrate 20 for arranging reference optical fiber components. The sensing fiber component is embedded in the flexible tactile elastomer along the prefabricated fiber channel and a slight pretension is applied. At the same time, the reference fiber component is placed horizontally in the fiber groove and fixed. Liquid elastomer is poured into the mold, and after vacuum degassing, it is cured in an environment of 60~80℃. After curing, the flexible tactile elastomer is connected to the rigid substrate 20, and the sensing fiber optic component and the reference fiber optic lead-out end are connected to the optical path system.
7. The preparation method according to claim 6, characterized in that, The center wavelengths of the sensing fiber optic component and the reference fiber optic component are different.
8. The preparation method according to claim 6, characterized in that, The optical fiber channels are inclined to the plane of the rigid substrate and are evenly distributed along the center of the sensing area.
9. The preparation method according to claim 6, characterized in that, The connection between the flexible tactile elastomer and the rigid substrate includes bonding or mechanical fixation.
10. A six-dimensional force-tactile sensor based on FBG, characterized in that, include: Rigid substrate; A flexible elastomer is disposed on the side of the rigid substrate; A sensing group is disposed within the flexible elastic body and is arranged corresponding to the sensing area; The sensing group includes three sensing fiber optic components, which are evenly arranged around the center of the sensing area. One end of each sensing fiber optic component extends from the side of the flexible elastomer close to the rigid substrate to the side away from the rigid substrate, and is used to detect forces and torques from different directions. And, the reference fiber optic component is located within a rigid substrate.