An artificial surface plasmon structure sensor, a multi-modal perception gripper and application thereof

Abstract

The application discloses an artificial surface plasmon structure sensor, a multi-modal sensing gripper and application thereof, and relates to the technical field of intelligent laboratory automation. The sensor comprises an artificial surface plasmon substrate, the artificial surface plasmon substrate comprises a periodic groove unit, a gradually changing transition section and a coplanar waveguide transmission end arranged in sequence from near to far at two ends of the periodic groove unit, the coplanar waveguide transmission end is connected with a coaxial cable, the coaxial cables on both sides are connected with a signal input end and a signal output end of an external SDR device respectively, and an excitation signal is input to the sensor and a response signal is collected through the SDR device. The application has the functions of gripping force control and substance identification, can safely complete the gripping and identification of a dangerous liquid centrifuge tube without any human intervention, greatly reduces the need for manual contact with harmful substances, and improves the safety and automation level of the intelligent laboratory in dangerous operation.

Description

TECHNICAL FIELD

[0001] The present application relates to an artificial surface plasmon structure sensor, a multi-modal sensing gripper and its application, belonging to the technical field of intelligent laboratory automation. BACKGROUND

[0002] In the field of intelligent laboratory automation, precise material operation and state sensing are the core technical links to realize high-throughput and unmanned experiments, especially in experimental environments involving strong corrosive, toxic or volatile hazardous substances. It is of great significance to replace manual operation with robots to ensure personnel safety and health. In traditional laboratories, centrifuge tube grabbing, pipetting and dispensing operations are performed manually. When contacting acids (such as sulfuric acid and hydrochloric acid), organic solvents (such as ethyl acetate and isopropanol) or other hazardous chemicals, it is extremely easy to cause exposure risks due to liquid splashing, gas evaporation or container breakage, causing health hazards to experimental personnel.

[0003] Currently, most automated grabbing solutions still lack integrated sensing capabilities for grabbing force and material composition. Traditional grippers cannot sense grabbing force in real time, which can cause unstable centrifuge tube grabbing or even breakage, increasing the risk of hazardous material leakage. Material identification often relies on independent external devices (such as spectrometers), which cannot complete composition judgment simultaneously during grabbing actions, and cannot meet the durability requirements of sensors in corrosive environments.

[0004] Therefore, it is urgent to develop an integrated intelligent gripper that can withstand corrosive environments and has grabbing force control and material identification functions, so that it can safely and reliably complete the grabbing, identification and processing of hazardous liquids without human intervention, minimizing the need for human contact with hazardous materials, and improving the safety, automation level and sustainability of intelligent laboratories in dangerous operations. SUMMARY

[0005] The present application aims to overcome the shortcomings of the prior art and provide an artificial surface plasmon structure sensor, a multi-modal sensing gripper and its application, which has grabbing force control and material identification functions, can safely and reliably complete the grabbing, identification and processing of hazardous liquid centrifuge tubes without human intervention, minimizes the need for human contact with hazardous materials, and improves the safety, automation level and sustainability of intelligent laboratories in dangerous operations.

[0006] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0007] In a first aspect, the present application provides an artificial surface plasmon structure sensor, comprising an artificial surface plasmon substrate, the artificial surface plasmon substrate comprising a periodic groove unit and a gradually changing transition section and a coplanar waveguide transmission end arranged in turn at both ends of the periodic groove unit, the coplanar waveguide transmission end is connected with a coaxial cable, and the coaxial cable on both sides is connected with a signal input end and a signal output end of an external software defined radio device respectively, and the artificial surface plasmon structure sensor is inputted with an excitation signal and collected with a response signal by the external software defined radio device.

[0008] Further, both ends of the coplanar waveguide transmission end are connected with the coaxial cable through an IPEX1 connector seat.

[0009] In a second aspect, the present application also provides a multi-modal sensing gripper, comprising an electric gripper and two grippers arranged oppositely at the output end of the electric gripper, the grippers are divided into sensing grippers and positioning grippers, the sensing gripper is integrated with the artificial surface plasmon structure sensor as any one of the first aspect at the gripping contact surface by embedding, and the surface of the artificial surface plasmon structure sensor is attached with an elastic silicone converter.

[0010] Further, the sensing gripper is provided with a guide hole and a wire groove, the wire groove extends from one side of the guide hole to the end of the sensing gripper, and the coaxial cable on one side of the artificial surface plasmon structure sensor passes through the guide hole and is connected with the signal input end of the external software defined radio device, and the coaxial cable on the other side is arranged in the wire groove and extends to the end of the sensing gripper and is connected with the signal output end of the external software defined radio device.

[0011] Further, when the gripping force is in the range of 0-15N, the thickness of the elastic silicone converter is 2mm, and the hardness is A30.

[0012] Further, the gripping contact surface of the positioning gripper is provided with a U-shaped positioning groove, and the U-shaped positioning groove is matched with the outer diameter of the centrifugal tube.

[0013] In a third aspect, the present application also provides an application of the multi-modal sensing gripper as any one of the second aspect, comprising gripping force control of a centrifugal tube and identification of reagent components inside the centrifugal tube.

[0014] Further, the process of the gripping force control of the centrifugal tube comprises:

[0015] The external software defined radio device inputs an excitation signal into the artificial surface plasmon structure sensor and collects a response signal;

[0016] The multi-modal perception gripper is driven to grab the centrifugal tube. When the centrifugal tube contacts the elastic silica gel converter, the elastic deformation of the elastic silica gel converter changes the dielectric environment of the surface plasmon structure sensor surface, so that the response signal produces a phase difference change;

[0017] According to the phase difference-pressure relationship curve, the pressure value between the centrifugal tube and the multi-modal perception gripper is obtained through the phase difference change;

[0018] The gripping force of the multi-modal perception gripper is controlled by using the pressure value between the centrifugal tube and the multi-modal perception gripper;

[0019] The method for obtaining the phase difference-pressure relationship curve comprises:

[0020] A known pressure is applied to the elastic silica gel converter of the multi-modal perception gripper;

[0021] The external software-defined radio device inputs an excitation signal to the surface plasmon structure sensor and collects the response signal, and obtains the corresponding phase difference change from the response signal;

[0022] The phase difference-pressure relationship curve is fitted by synthesizing multiple sets of pressure-phase difference data.

[0023] Further, the process of identifying the reagent component in the centrifugal tube comprises:

[0024] The external software-defined radio device inputs an excitation signal to the surface plasmon structure sensor and collects the response signal;

[0025] The amplitude curve is extracted from the response signal;

[0026] The amplitude curve is used as the input of the pre-trained K-nearest neighbor classification algorithm model, and the pre-test reagent component is obtained through frequency domain analysis output by the K-nearest neighbor classification algorithm model.

[0027] Further, the pre-training method of the K-nearest neighbor classification algorithm model comprises:

[0028] S1, obtaining training set data and test set data, the training set data and the test set data are composed of multiple pairs of known reagent components and corresponding amplitude curves;

[0029] S2, the amplitude curve in the training set data is used as the input, the K-nearest neighbor classification algorithm model is trained, the network parameters are optimized in the training process, until the preset training times are reached, and the trained K-nearest neighbor classification algorithm model is obtained;

[0030] S3, input the amplitude curve in the test set data as an input to test the K-neighbor classification algorithm model, obtain a test result, match the test result with the known reagent component in the test set data, and calculate a prediction success rate, when the prediction success rate is higher than a preset value, obtain a pre-trained K-neighbor classification algorithm model, otherwise, repeat steps S2-S3 until the prediction success rate is higher than the preset value.

[0031] Compared with the prior art, the present application has the following beneficial effects:

[0032] The present application has the functions of gripping force control and substance recognition, the normal distance change between the elastic silica gel converter and the centrifugal tube can be converted into the elastic deformation of the elastic silica gel converter through the elastic silica gel converter, the elastic deformation changes the dielectric environment of the surface plasmon structure sensor surface, and then causes the electromagnetic wave propagation phase shift, so that the indirect measurement of the gripping pressure is realized through the phase difference of the sensor response signal; and the different dielectric constants of different reagent components result in different characteristics on the amplitude curve, so that the reagent component recognition is realized, so that the gripping, recognition and processing of the dangerous liquid centrifugal tube are safely and reliably completed without human intervention, the need for manual contact with harmful substances is minimized, and the safety, automation level and sustainability of the intelligent laboratory in dangerous operation are improved;

[0033] The SSPP sensor of the present application shows stable high-sensitivity sensing performance in the 4.7-4.9 GHz frequency band, the dispersion characteristics can be flexibly customized according to the dielectric properties and mechanical environment of the measured object in actual application, and the multi-modal sensing jaw has a compact and reasonable overall layout, realizing the integration of sensing, wiring and installation functions in a limited space. The structure not only provides reliable mechanical protection and environmental isolation for the surface plasmon structure sensor, but also ensures the integrity of high-frequency signal transmission through the integrated cable management design, and is suitable for high-throughput experimental scenes that require high-frequency sampling, strong interference shielding and complex motion trajectories. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 FIG. 1 is a structural schematic diagram of a surface plasmon structure sensor in an embodiment of the present application;

[0035] Figure 2 FIG. 2 is a structural schematic diagram of a surface plasmon substrate of a surface plasmon structure sensor in an embodiment of the present application;

[0036] Figure 3 FIG. 3 is a structural schematic diagram of a multi-modal sensing jaw in an embodiment of the present application;

[0037] Figure 4Structure diagram of a sensing jaw in a multi-modal sensing jaw in an embodiment of the present application (structure diagram of a sensing jaw before installation of an artificial surface plasmon structure sensor), wherein a is a structure diagram of a non-contact surface of the sensing jaw, b is a side view structure diagram of the sensing jaw, and c is a structure diagram of a contact surface of the sensing jaw;

[0038] Figure 5 Structure diagram of a positioning jaw in a multi-modal sensing jaw in an embodiment of the present application, wherein a is a side view structure diagram of the positioning jaw, and b is a structure diagram of a contact surface of the positioning jaw;

[0039] Figure 6 Structure diagram of a sensing jaw in a multi-modal sensing jaw in an embodiment of the present application (structure diagram of a sensing jaw after installation of an artificial surface plasmon structure sensor);

[0040] Figure 7 Result diagram of electromagnetic characteristics of an artificial surface plasmon structure sensor in an embodiment of the present application characterized by S parameters and dispersion relations, wherein (a) is a curve diagram of S11 parameters of the artificial surface plasmon structure sensor under different frequency signals, (b) is a curve diagram of normalized propagation constants of the artificial surface plasmon structure sensor under different periodic unit lengths p, (c) is a curve diagram of normalized propagation constants of the artificial surface plasmon structure sensor under different periodic unit heights h, and (d) is a curve diagram of normalized propagation constants of the artificial surface plasmon structure sensor under different SSPP substrate thicknesses t;

[0041] Figure 8 Structure diagram of a test system in Embodiment 3 of the present application;

[0042] Figure 9 Signal transmission diagram between an artificial surface plasmon structure sensor and an external software-defined radio device in an embodiment of the present application;

[0043] Figure 10 Phase difference (θ) of an elastic silicone converter under different hardness and thickness conditions in an embodiment of the present application Fig. 6 is a schematic diagram of the influence of the phase difference of the elastic silicone converter under different forces, wherein (a) is a schematic diagram of the phase difference change of the elastic silicone converter with a hardness of A20 and a thickness of 1.5 mm under different forces, (b) is a schematic diagram of the phase difference change of the elastic silicone converter with a hardness of A30 and a thickness of 1.5 mm under different forces, (c) is a schematic diagram of the phase difference change of the elastic silicone converter with a hardness of A50 and a thickness of 1.5 mm under different forces, (d) is a schematic diagram of the phase difference change of the elastic silicone converter with a hardness of A20 and a thickness of 2 mm under different forces, (e) is a schematic diagram of the phase difference change of the elastic silicone converter with a hardness of A30 and a thickness of 2 mm under different forces, and (f) is a schematic diagram of the phase difference change of the elastic silicone converter with a hardness of A50 and a thickness of 2 mm under different forces;

[0044] Figure 11 Fig. 7 is a schematic diagram of the mechanical sensing performance of the multi-modal sensing gripper in the process of grabbing and placing the empty centrifuge tube and the centrifuge tube filled with water in an embodiment of the present application, wherein a is a schematic diagram of the response characteristic of the phase difference (φ) of the empty centrifuge tube state with time and its relationship with the grabbing force, and b is a schematic diagram of the response characteristic of the phase difference (φ) of the centrifuge tube filled with water with time and its relationship with the grabbing force;

[0045] Figure 12 Fig. 8 is a schematic diagram of the amplitude curve of 10 typical centrifuge tube states (empty tube (Empty), water (Water), acetic acid (AcOH), ethyl acetate (EtOAc), sulfuric acid (H2SO4), ethanol (EtOH), isopropanol (IPA), glycerol (Gly), 783 (washing net water), and hydrochloric acid (HCl)) in an embodiment of the present application;

[0046] Figure 13 Fig. 9 is a schematic diagram of the reagent component identification structure based on the amplitude curve of 10 typical centrifuge tube states (empty tube (Empty), water (Water), acetic acid (AcOH), ethyl acetate (EtOAc), sulfuric acid (H2SO4), ethanol (EtOH), isopropanol (IPA), glycerol (Gly), 783 (washing net water), and hydrochloric acid (HCl)) in an embodiment of the present application through the K nearest neighbor classification algorithm model;

[0047] In the figure: 1-electric gripper, 2-sensing gripper, 3-positioning gripper, 4-artificial surface plasmon structure sensor, 401-periodic slot unit, 402-gradual transition section, 403-coplanar waveguide transmission end, 404-coaxial cable, 5-centrifuge tube, 6-guide hole, 7-mounting hole, 8-wire slot, 9-U-shaped positioning groove, 10-IPEX 1st generation connector, 11-elastic silicone converter. DETAILED DESCRIPTION

[0048] ​​The application will be further described below with reference to the drawings. The following examples are only used to more clearly illustrate the technical solutions of the application, and cannot be used to limit the protection scope of the application.

[0049] Example 1

[0050] As shown in Figure 1 , the artificial surface plasmon structure sensor provided by the embodiment of the application includes an artificial surface plasmon (SSPP) substrate, the SSPP substrate is implemented by a PCB circuit board, the material of the SSPP substrate is F4R, and the overall size of the SSPP substrate is 48 mm in length, 14 mm in width, and 0.8 mm in thickness.

[0051] In combination with Figure 2 , the SSPP substrate includes a periodic slot unit 401, the periodic slot unit 401 is arranged at the middle part, and gradually changing transition sections 402 for impedance matching and CPW (coplanar waveguide) transmission ends 403 for signal feeding are sequentially arranged on both sides of the periodic slot unit 401 from near to far. The CPW transmission ends are welded with IPEX1 generation connector sub 10, and low-loss transmission of radio frequency signals and stable mechanical connection are realized through welding. The IPEX1 generation connector sub 10 keeps impedance matching with the sensor circuit, the working frequency band covers 1-6 GHz, and the SSPP substrate can be effectively excited and the change of the surface electromagnetic field distribution of the SSPP substrate can be collected. The CPW transmission ends are connected with a coaxial cable 404 through the IPEX1 generation connector sub 10, and the coaxial cable 404 is connected with a signal input end and a signal output end of an external SDR (software defined radio) device.

[0052] In combination with Figure 9 , the signal output end of the SDR device inputs the signal to the SSPP substrate, the signal is converted into a sensor signal after being processed by the SSPP substrate, the sensor signal is output to the signal input end of the SDR device, and the sensor signal is received by the SDR device, so that the excitation and response collection of the sensor signal are realized.

[0053] The electromagnetic characteristics of the artificial surface plasmon structure sensor provided by the embodiment are characterized by S parameters and dispersion relations.

[0054] As shown in a of 7, in a wide frequency range of 1-8 GHz, especially in the frequency bands of 5 GHz and 6 GHz, the reflection loss is lower than -10 dB, which indicates that the structure has good impedance matching characteristics and energy transmission efficiency, effectively suppresses signal reflection, and improves detection sensitivity.

[0055] The dispersion behavior of the sensor can be cooperatively regulated by multiple structural parameters such as the length p of the periodic unit, the height h of the periodic unit (as shown in Figure 2 ), and the thickness t of the SSPP substrate, so as to adapt to the needs of different sensing scenarios. Figure 7As shown in FIG. b, when the period unit length p gradually increases from 2 mm to 1.2 mm, the dispersion curve is obviously moved to the low frequency area, and is all below the light cone line, indicating that the structure has strong slow wave effect and flexible frequency band adaptation ability, while maintaining significant separation from the free space light mode, effectively suppressing electromagnetic wave radiation loss and enhancing field confinement ability.

[0056] Then, the thickness t of the SSPP substrate is accurately controlled to realize systematic optimization of the dispersion characteristics and working performance of the SSPP sensor. The results are as shown in FIG. c. Figure 7 As shown in FIG. c, as the thickness t increases from 0.4 mm to 1.2 mm, the normalized propagation constant βp / π presents regular evolution in the whole interval, and the overall dispersion curve moves to the low frequency, but the west-east range is small, indicating that the thickening of the substrate can fine-tune the dispersion effect. As shown in FIG. d, as the period unit height h gradually increases from a lower size, the dispersion curve of the structure as a whole shifts to the low frequency area, indicating that the slow wave effect is enhanced, and the electromagnetic energy is more concentrated on the surface of the structure, which is conducive to improving the near-field sensing sensitivity. When the period unit height h is 3 mm, the dispersion curve appears above the light cone, indicating that there is no subwavelength effect at this time. Figure 7

[0057] Through multi-parameter collaborative optimization, the SSPP sensor exhibits stable high sensitivity sensing performance in the frequency band of 4.7~4.9 GHz, and the dispersion characteristics can be flexibly customized according to the dielectric properties and mechanical environment of the measured object in actual application, providing a reliable sensing mechanism for liquid composition identification, grabbing force monitoring and multi-physical quantity synchronous sensing in centrifuge tubes under complex experimental conditions.

[0058] Embodiment 2

[0059] As shown in FIG. c, as the thickness t increases from 0.4 mm to 1.2 mm, the normalized propagation constant βp / π presents regular evolution in the whole interval, and the overall dispersion curve moves to the low frequency, but the west-east range is small, indicating that the thickening of the substrate can fine-tune the dispersion effect. As shown in FIG. d, as the period unit height h gradually increases from a lower size, the dispersion curve of the structure as a whole shifts to the low frequency area, indicating that the slow wave effect is enhanced, and the electromagnetic energy is more concentrated on the surface of the structure, which is conducive to improving the near-field sensing sensitivity. When the period unit height h is 3 mm, the dispersion curve appears above the light cone, indicating that there is no subwavelength effect at this time. Figure 3 As shown in FIG. c, as the thickness t increases from 0.4 mm to 1.2 mm, the normalized propagation constant βp / π presents regular evolution in the whole interval, and the overall dispersion curve moves to the low frequency, but the west-east range is small, indicating that the thickening of the substrate can fine-tune the dispersion effect. As shown in FIG. d, as the period unit height h gradually increases from a lower size, the dispersion curve of the structure as a whole shifts to the low frequency area, indicating that the slow wave effect is enhanced, and the electromagnetic energy is more concentrated on the surface of the structure, which is conducive to improving the near-field sensing sensitivity. When the period unit height h is 3 mm, the dispersion curve appears above the light cone, indicating that there is no subwavelength effect at this time.

[0060] The gripper is divided into a sensing gripper 2 and a positioning gripper 3, and the artificial surface plasmon structure sensor 4 of embodiment 1 is embedded in the gripping contact surface of the sensing gripper 2. The profile of the sensing gripper 2 matches the shape of the artificial surface plasmon structure sensor 4, and the embedding flatness of the sensor is ensured by precise machining.

[0061] In combination with Figure 6 ​The artificial surface plasmon structure sensor 4 is provided with an elastic silica gel converter 11 on the surface, and when the centrifugal tube 5 contacts the elastic silica gel converter 11, the elastic silica gel converter 11 converts the normal distance between the centrifugal tube 5 and the elastic silica gel converter 11 into the elastic deformation of the silica gel body, the elastic deformation changes the dielectric environment of the surface of the SSPP substrate, and then causes the phase shift of electromagnetic wave propagation.

[0062] As shown in Figure 4 , the sensing jaw 2 is provided with a wire groove 8 and a guide hole 6, which are used to regularize the coaxial cable 404 and prevent it from being squeezed and disturbed in mechanical movement. The diameters of the wire groove 8 and the guide hole 6 need to be matched with the outer diameter of the coaxial cable 404, and in this embodiment, the diameters of the wire groove 8 and the guide hole 6 are both 4 mm. The wire groove 8 extends from one side of the guide hole 6 to the end of the sensing jaw 2. One side of the artificial surface plasmon structure sensor 4 is connected with the signal input end of the external software defined radio device through the coaxial cable 404 passing through the guide hole 6, and the other side of the artificial surface plasmon structure sensor 4 is connected with the signal output end of the external software defined radio device through the coaxial cable 404 arranged in the wire groove 8 and extending to the end of the sensing jaw 2.

[0063] As shown in Figure 5 , the U-shaped positioning groove 9 is arranged at the gripping contact surface of the positioning jaw 3, and the U-shaped positioning groove 9 is matched with the outer diameter of the centrifugal tube 5. In this embodiment, the opening width of the U-shaped positioning groove 9 is 14.47 mm, which can meet the stable positioning requirements of various specifications of the centrifugal tube 5.

[0064] In the initial stage of clamping, the centrifugal tube 5 is clamped in the U-shaped positioning groove 9, which can effectively prevent the centrifugal tube 5 from tilting and shaking.

[0065] The overall layout of the multi-modal sensing jaw of this embodiment is compact and reasonable, and the functions of sensing, wiring and installation are integrated in a limited space. This structure not only provides reliable mechanical protection and environmental isolation for the artificial surface plasmon structure sensor 4, but also guarantees the integrity of high-frequency signal transmission through the integrated cable management design, which is suitable for high-throughput experimental scenes that require high-frequency sampling, strong interference shielding and complex motion trajectories.

[0066] Next, the coupling influence law of the Shore hardness (A20, A30, A50) and the thickness (1.5 mm, 2 mm) of the elastic silica gel converter 11 on the corresponding relationship between the sensor phase difference (θ) ) and the grasping force is studied through experiments, which provides key design basis for high-precision force perception.

[0067] Results Figure 10As shown, the phase difference under different hardness and thickness combinations presents obvious changes with the increase of applied pressure, but the response characteristics are significantly different: under the same thickness, the silicone transducer with lower hardness (such as A20) shows higher phase sensitivity in the low force interval (0-5N), but its response tends to be saturated, while the transducer with higher hardness (such as A50) presents a wider linear response range and stronger overload capacity in the high force interval (5-15N); under the same hardness, the thinner (1.5mm) transducer has higher initial sensitivity, but its effective range is smaller, while the thicker (2mm) transducer can significantly expand the force perception range and improve the linearity.

[0068] The force mapping model under different parameter combinations can be established by linear and polynomial fitting, respectively. A30 hardness and 2mm thickness combination exhibits the best comprehensive performance in the 0-15N range, with good sensitivity and wide linear range; the research results show that by matching the force detection range and precision requirements of the target application, the hardness and thickness parameters of the silicone transducer can be selected accordingly to realize high-precision and high-reliability calculation of the grasping force, providing a key technical basis for the wide application of the sensing gripper in precise operation, fragile object grasping and automation experiments.

[0069] Embodiment 3

[0070] Based on embodiment 2, the application of a multi-modal sensing gripper is provided, which includes grasping and internal reagent component identification for centrifuge tubes, specifically:

[0071] The process of grasping force control includes:

[0072] The external software defines the input excitation signal of the radio device to the plasmonic sensor 4 and collects the response signal of the sensor.

[0073] The electrically driven gripper 1 drives the gripper to grasp the centrifuge tube, when the centrifuge tube 5 contacts the elastic silicone transducer 11, the elastic silicone transducer 11 can convert the normal distance change between the elastic silicone transducer 11 and the centrifuge tube 5 into elastic deformation of the elastic silicone transducer 11, which changes the dielectric environment of the surface of the plasmonic sensor 4, and further causes the phase shift of electromagnetic wave propagation, resulting in the change of the phase difference of the output signal of the plasmonic sensor 4.

[0074] According to the phase difference change of the sensor response signal, combined with the pre-calibrated phase difference-pressure relationship curve, the pressure value between the centrifuge tube 5 and the multi-modal sensing gripper can be obtained.

[0075] Finally, the driving force of the electric gripper 1 is adjusted in real time by using the measured pressure value between the centrifugal tube 5 and the multi-modal sensing gripper, so as to realize real-time control of the grabbing force of the centrifugal tube 5.

[0076] In the embodiment, the calibration method of the phase difference-pressure relationship curve is to apply a known pressure to the elastic silicone converter 11, collect the response signal of the sensor, obtain the corresponding phase difference change from the response signal, and fit a plurality of groups of pressure-phase difference to obtain the phase-pressure relationship curve.

[0077] The process of identifying the reagent components in the centrifugal tube 5 includes:

[0078] The external software defines the radio device to input a sweep excitation signal of 4.7-4.9 GHz to the plasmonic sensor 4 and collect the response signal of the sensor (the principle is that different reagents have different dielectric properties (dielectric constant, conductivity, etc.), so that the amplitude and resonance frequency of the sweep curve are different).

[0079] The amplitude curve is extracted from the response signal, and the amplitude curve is input into the pre-trained K-nearest neighbor classification algorithm (KNN algorithm) for frequency domain analysis to output the pre-test reagent component.

[0080] In the embodiment, the pre-training method of the KNN algorithm includes:

[0081] S1, obtain training set data and test set data, and the training set data and the test set data are composed of a plurality of pairs of known reagent components and corresponding amplitude curves (because different reagent components have different dielectric constants, the characteristics (such as mean, standard deviation, peak-to-peak value, main frequency position, high and low frequency energy distribution, morphological oscillation characteristics, and specific frequency point amplitude) on the amplitude curve are also different).

[0082] S2, input the amplitude curve in the training set data as input, train the KNN algorithm, and optimize the network parameters in the training process until a preset training number is reached, in the embodiment, each reagent component is repeatedly trained for 40 groups, and the trained KNN algorithm is obtained.

[0083] S3, input the amplitude curve in the test set data as input, test the KNN algorithm, obtain the test result, match the test result with the known reagent component in the test set data, when the prediction success rate is higher than a preset value, the pre-trained KNN algorithm is obtained, otherwise, repeat steps S2-S3 until the prediction success rate is higher than the preset value.

[0084] The performance of the multimodal sensing gripper provided in this embodiment is tested below. The test system, as shown in Figure 8, consists of a sensing and measurement unit, a control unit, and a signal processing unit, forming a closed-loop sensing and operating system. The sensing and measurement unit is the multimodal sensing gripper provided in this embodiment, responsible for real-time acquisition of dielectric and mechanical signals during the gripping process. The control unit is based on a microcontroller and receives instructions from the microcontroller through a relay module. It controls the 24V DC power supply to drive the electric gripper to complete gripping, lifting, and placement actions, and transmits and receives radio frequency signals using an SDR.

[0085] First, the mechanical sensing performance of the multimodal sensing gripper in grasping empty and water-filled centrifuge tubes was verified through three consecutive repeatable tests. The experimental results clearly showed the phase difference in the two states. The response characteristics that change over time and their mapping relationship with gripping force, such as Figure 11 As shown, during the grasping of an empty centrifuge tube (left side), the phase difference variation range is small (0.00~0.35 rad), corresponding to a low grasping force (0~3 N). The response curve is smooth and has good repeatability, indicating that the empty centrifuge tube has little impact on the phase of the SSPP sensor. During the grasping of a water-filled centrifuge tube (right side), the phase difference variation range increases significantly (0.00~0.8 rad), corresponding to a grasping force range of 0~6.94 N. This is because the high dielectric constant of water enhances the field-matter interaction, and the increased liquid mass increases the mechanical load. The system responds by... The force mapping model calculates the pressure value in real time, demonstrating a higher response amplitude and consistent repeatability. In both states, the phase difference rises rapidly when the gripping action occurs, remains stable during the holding phase, and quickly returns to zero after release, with no significant hysteresis in the dynamic response. This result proves that the sensing gripper described in this invention can effectively distinguish objects with different loads and dielectric properties, and achieves high-precision, high-repeatability gripping force monitoring through a unified mechanical sensing mechanism, providing reliable technical support for automated operations involving multiple material states in intelligent laboratories.

[0086] Then, 10 typical centrifuge tube states were identified, including empty tube, water, acetic acid (AcOH), ethyl acetate (EtOAc), sulfuric acid (H2SO4), ethanol (EtOH), isopropanol (IPA), glycerol (Gly), 783 (screen cleaning solution), and hydrochloric acid (HCl).

[0087] The amplitude curve extracted from the sensor response signal is as follows: Figure 12As shown, the amplitude response of each centrifuge tube state in the 4.8-5.2 GHz frequency band shows significant differences. The empty tube has the lowest amplitude response and flat curve in the entire frequency band, while different reagents show unique resonance frequency and amplitude characteristics due to their different dielectric properties: high dielectric constant liquids such as water and glycerol induce a significant resonance peak near 4.9 GHz, and polar liquids such as sulfuric acid and acetic acid show a wide amplitude response in the 5.0-5.1 GHz range; all test states have distinguishable spectral fingerprints in the 4.8-5.2 GHz operating frequency band, and the amplitude of multiple characteristic frequency points, the overall curve shape and resonance characteristics in this frequency band are extracted. In order to improve the sweep efficiency, the first trough to peak position, i.e. the characteristic region of 4.78-4.94 GHz, is selected for sweep operation.

[0088] Finally, based on the amplitude curves of the 10 centrifuge tube states obtained by sweep testing, the K nearest neighbor classification algorithm model is used to identify the empty tube (Empty), water (Water), acetic acid (AcOH), ethyl acetate (EtOAc), sulfuric acid (H2SO4), ethanol (EtOH), isopropanol (IPA), glycerol (Gly) 783 (washing water) and hydrochloric acid (HCl).

[0089] By extracting multi-dimensional features including resonance frequency amplitude, specific frequency point response, curve steepness, frequency band width, etc. to construct a feature vector, and using the pre-trained KNN algorithm model for reagent component prediction. The results are shown in Figure 13 As shown, the 10 states show good clustering in the feature space, with clear boundaries between each class, and the classification accuracy is above 95%. Among them, the empty tube is easy to distinguish due to its weakest dielectric response, water, glycerol and other high dielectric constant liquids are clustered in a specific area due to their significant resonance peak, sulfuric acid, hydrochloric acid and other strong polar liquids form independent classes due to their unique loss characteristics, and ethanol, isopropanol, ethyl acetate and other organic solvents are effectively separated according to their differences in dielectric relaxation frequency.

[0090] The KNN algorithm, with its characteristics of no complex model assumption and strong adaptability to multi-modal data, fully excavates the dielectric information captured by the SSPP structure sensor, and proves the effectiveness of multi-dimensional features in complex liquid identification. The results show that the sensing system described in the present application can stably and reliably identify common liquids and states in the laboratory, providing core algorithm support for unmanned and high-throughput identification of centrifuge tube contents in intelligent and automated experimental environments.

[0091] The above only describes the preferred embodiments of the present application, and it should be noted that for those skilled in the art, without departing from the technical principles of the present application, several improvements and modifications can be made, which should also be considered as the protection scope of the present application.

Claims

1. Use of a multi-modal sensing gripper, characterized in that, The application relates to a centrifugal tube internal reagent component recognition method and a centrifugal tube gripping force control method. The process of the centrifugal tube gripping force control comprises the following steps: An external software-defined radio device inputs an excitation signal to the artificial surface plasmon structure sensor and collects a response signal thereof; The multi-modal perception gripper is driven to grip the centrifugal tube, and when the centrifugal tube is in contact with the elastic silica gel converter, the dielectric environment of the surface of the artificial surface plasmon structure sensor is changed through the elastic deformation of the elastic silica gel converter, so that the response signal produces a phase difference change; According to a phase difference-pressure relationship curve, the pressure value between the centrifugal tube and the multi-modal perception gripper is obtained through the phase difference change; The gripping force of the multi-modal perception gripper is controlled by using the pressure value between the centrifugal tube and the multi-modal perception gripper. The method for obtaining the phase difference-pressure relationship curve comprises the following steps: A known pressure is applied to the elastic silica gel converter of the multi-modal perception gripper; An external software-defined radio device inputs an excitation signal to the artificial surface plasmon structure sensor and collects a response signal thereof, and the corresponding phase difference change is obtained from the response signal; A phase difference-pressure relationship curve is fitted by comprehensively using a plurality of groups of pressure-phase difference data; The artificial surface plasmon structure sensor comprises an artificial surface plasmon substrate, the artificial surface plasmon substrate comprises a periodic groove unit and a gradually changing transition section and a coplanar waveguide transmission end arranged in sequence from near to far at two ends of the periodic groove unit, the coplanar waveguide transmission ends are connected with coaxial cables, the coaxial cables on the two sides are connected with a signal input end and a signal output end of an external software-defined radio device respectively, and the external software-defined radio device inputs an excitation signal to the artificial surface plasmon structure sensor and collects a response signal thereof; The multi-modal perception gripper comprises an electric gripper and two grippers oppositely arranged at an output end of the electric gripper, the two grippers are a sensing gripper and a positioning gripper respectively, the sensing gripper is integrated with the artificial surface plasmon structure sensor in an embedded mode at a gripping contact surface thereof, and the surface of the artificial surface plasmon structure sensor is attached with an elastic silica gel converter.

2. Use of a multi-modal sensing gripper according to claim 1, characterized in that, The process of the centrifugal tube internal reagent component recognition comprises the following steps: An external software-defined radio device inputs an excitation signal to the artificial surface plasmon structure sensor and collects a response signal thereof; An amplitude curve is extracted from the response signal; The amplitude curve is taken as an input of a pre-trained K nearest neighbor classification algorithm model, and a pretest reagent component is output through frequency domain analysis of the K nearest neighbor classification algorithm model.

3. Use of a multi-modal sensing jaw according to claim 2, characterized in that, The pre-training method of the K nearest neighbor classification algorithm model comprises the following steps: S1, obtaining training set data and test set data, the training set data and the test set data are composed of a plurality of pairs of known reagent components and corresponding amplitude curves; S2, taking the amplitude curves in the training set data as inputs, training the K nearest neighbor classification algorithm model, optimizing network parameters in the training process until a preset training number is reached, and obtaining a trained K nearest neighbor classification algorithm model. S3, input the amplitude curve in the test set data as an input to test the K-neighbor classification algorithm model, obtain a test result, match the test result with the known reagent component in the test set data, and calculate a prediction success rate, when the prediction success rate is higher than a preset value, obtain a pre-trained K-neighbor classification algorithm model, otherwise, repeat steps S2-S3 until the prediction success rate is higher than the preset value.

4. Use of a multi-modal sensing gripper according to claim 1, characterized in that, The two ends of the coplanar waveguide transmission end are connected with coaxial cables through IPEX1 connector sub.

5. Use of a multi-modal sensing gripper according to claim 1, characterized in that, The sensing jaw is provided with a guide hole and a wire slot, the wire slot extends from one side of the guide hole to the end of the sensing jaw, one side of the surface plasmon polariton sensor coaxial cable passes through the guide hole and is connected with the signal input end of the external software defined radio device, and the other side coaxial cable is arranged in the wire slot and extends to the end of the sensing jaw and is connected with the signal output end of the external software defined radio device.

6. Use of a multi-modal sensing gripper according to claim 1, characterized in that, When the gripping force is in the range of 0-15N, the thickness of the elastic silica gel converter is 2mm, and the hardness is A30.

7. Use of a multi-modal sensing gripper according to claim 1, characterized in that, The U-shaped positioning groove is matched with the outer diameter of the centrifugal tube.

Images