A three-dimensional deformable soft body driver, manufacturing method, control method and system

Abstract

The present invention proposes a three-dimensional deformable soft actuator, a fabrication method, a control method, and a system. The air chamber at the upper end of the actuator is oriented perpendicularly to the air chamber at the lower end. Both ends can deform horizontally and vertically relative to themselves, enabling the designed three-dimensional deformable soft actuator to generate spatial deformation, thereby increasing the driving force. The entire actuator consists of a single, continuous fluid pathway, reducing the burden of supplying the driving fluid.

Description

Technical Field

[0001] The present invention belongs to the technical field of soft drive, and in particular relates to a three-dimensional deformable soft drive, a manufacturing method, a control method and a system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] As a key development area in the field of robotics, soft robotics technology is gradually moving from experimental environments to practical applications. However, compared to traditional rigid robots, while soft robots have greater environmental adaptability, their driving capabilities lag significantly behind those of widely used rigid robots, a key factor hindering the practical application of soft robots. As the power source for soft robots, soft actuators primarily generate driving force through the deformation of their flexible components under external conditions. However, current soft actuators suffer from problems such as insufficient power, low driving efficiency, and low driving accuracy due to limitations in their structural design and driving methods.

[0004] Specifically, the commonly used actuators for soft robots are small in size, made of soft materials, and deform slowly, resulting in low driving force and difficulty supporting the complex applications of soft robots. Furthermore, soft actuators can usually only bend / stretch within a single plane and cannot perform complex spatial deformations to enhance driving capabilities. Using multiple actuators in collaboration increases the burden on energy supply. When manufacturing actuators with complex structures, 3D printing technology, which restricts material selection, is often used, resulting in limited elasticity of the actuator's own materials and insufficient deformation efficiency. Furthermore, the control of existing soft actuators primarily relies on manual operation and low-precision preset signals. The actuators have low deformation accuracy and cannot achieve precise driving of soft robots. This still falls short of the requirements for autonomous and stable operation of soft robots in complex environments.

[0005] Therefore, how to structurally design the soft actuator to enhance its driving ability and deformation ability; and how to control the soft actuator to improve its driving accuracy are issues that need to be urgently addressed. Summary of the Invention

[0006] In order to overcome the deficiencies of the above-mentioned prior art, the present invention provides a three-dimensional deformable soft driver, a manufacturing method, a control method and a system. The direction of the air chamber at the upper end of the soft driver is perpendicular to the direction of the air chamber at the lower end of the soft driver, so that the upper end and the lower end of the soft driver produce deformation in the horizontal and vertical directions relative to themselves. The designed three-dimensional deformable soft driver can produce spatial deformation, thereby increasing the driving force.

[0007] To achieve the above-mentioned object, a first aspect of the present invention provides a three-dimensional deformable soft actuator, comprising: a soft actuator upper end, a soft actuator lower end, and a connecting chamber;

[0008] The upper end of the soft drive is connected to the lower end of the soft drive via a connecting chamber;

[0009] The bending directions of the upper end of the soft driver and the lower end of the soft driver are perpendicular to each other, and the upper end of the soft driver and the lower end of the soft driver generate deformation relative to the horizontal direction and the vertical direction;

[0010] A fluid passage is passed through the upper end of the soft drive, the lower end of the soft drive and the interior of the connecting chamber.

[0011] A second aspect of the present invention provides a method for manufacturing a three-dimensional deformable soft driver, which is applied to the above-mentioned three-dimensional deformable soft driver and includes the following steps:

[0012] Printing a first mold and a second mold, wherein the first mold has a protrusion and the second mold has a cavity, and assembling the first mold and the second mold;

[0013] pouring liquid silicone rubber into the assembled first and second molds, and then demolding to obtain a driver body consisting of a driver upper end, a soft driver lower end, and a connecting chamber;

[0014] Printing a cavity-enclosed mold, pouring liquid silicone rubber into the cavity-enclosed mold and placing the soft driver body, and then demoulding to obtain a cavity-enclosed driver body;

[0015] The cavity reinforcement mold is printed, liquid silicone rubber is poured into the cavity reinforcement mold, and the cavity-enclosed actuator body is placed on the mold, and then demolding is performed to obtain a three-dimensional deformable soft actuator.

[0016] A third aspect of the present invention provides a control method for a three-dimensional deformable soft body driver, which is applied to the three-dimensional deformable soft body driver as described above, comprising:

[0017] According to the torque balance and the position relationship between the upper end of the soft actuator, the lower end of the soft actuator, and the connecting chamber, the relationship between the bending angle of the upper end of the soft actuator, the lower end of the soft actuator, and the connecting chamber and the input fluid pressure is obtained respectively;

[0018] According to the relationship between the bending angle and the input fluid pressure, a hysteresis inverse model is established;

[0019] The coefficients and thresholds of the established hysteresis inverse model are identified using real-coded genetic algorithm;

[0020] The identified hysteresis inverse model is combined with a sliding membrane controller to control the input fluid pressure of the three-dimensional deformable soft actuator.

[0021] A fourth aspect of the present invention provides a control system for a three-dimensional deformable soft body driver, comprising:

[0022] The bending sensor is used to obtain the bending angle value of the three-dimensional deformable soft actuator and transmit the obtained angle to the host computer;

[0023] The position sensor is used to obtain the displacement change value of the three-dimensional deformable soft body driver and transmit the obtained displacement change to the host computer;

[0024] The host computer performs trajectory tracking calculation based on the obtained bending angle value and displacement change value, and generates a control signal for controlling the pneumatic proportional valve;

[0025] The pneumatic proportional valve is used to control the fluid pressure input by the air pump to the three-dimensional deformation soft body driver according to the control signal of the host computer.

[0026] One or more of the above technical solutions have the following beneficial effects:

[0027] In this invention, by analyzing the characteristics and motion properties of animal limbs with multiple degrees of freedom, the bending directions of the upper and lower ends of the soft actuator are perpendicular to each other. These ends can deform horizontally and vertically relative to themselves, enabling the designed three-dimensional deformable soft actuator to generate spatial deformation, thereby increasing the driving force. Furthermore, the entire 3D deformable soft actuator contains only a single fluid pathway running from beginning to end, reducing the burden of supplying the driving fluid.

[0028] In the present invention, by performing bending analysis on each part of the designed three-dimensional deformable soft driver respectively, the bending characteristics of each part are accurately reflected. According to the relationship between the bending angle and the input fluid pressure, a hysteresis inverse model is established, and the coefficients and thresholds of the hysteresis inverse model are analyzed. The analyzed hysteresis inverse model is combined with the synovial control to achieve precise control of the deformation of the three-dimensional deformable soft driver, and solve the difficult problems of difficult feedback and inaccurate control of the soft driver state.

[0029] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0031] Figure 1 This is a schematic diagram of the structure of a three-dimensional deformable soft body driver in accordance with an embodiment of the present invention;

[0032] Figure 2 Schematic diagram of the mold design and manufacturing process of the three-dimensional deformable soft actuator in the second embodiment of the present invention;

[0033] Figure 3 This is a block diagram of the inner and outer loop tracking control of a three-dimensional deformable soft body actuator based on agent sliding mode based on reinforcement learning in the third embodiment of the present invention;

[0034] Figure 4 Schematic diagram of a control system for a three-dimensional deformable soft body driver in a fourth embodiment of the present invention.

[0035] In the figure, 1. lower end of the soft drive; 2. upper end of the soft drive; 3. connecting chamber; 4. fluid passage; 5. main body mold group; 6. chamber closing mold group; 7. chamber reinforcement mold group; 8. three-dimensional deformation soft drive. DETAILED DESCRIPTION

[0036] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0037] It should be noted that the terms used herein are for describing particular embodiments only and are not intended to limit the exemplary embodiments according to the present invention.

[0038] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0039] Example 1

[0040] like Figure 1 As shown, this embodiment discloses a three-dimensional deformable soft actuator, comprising: a soft actuator upper end 2, a soft actuator lower end 1, and a connecting chamber 3;

[0041] The upper end 3 of the soft drive is connected to the lower end 1 of the soft drive via a connecting chamber 3;

[0042] The bending directions of the upper end 2 of the soft drive and the lower end 1 of the soft drive are perpendicular to each other, and the upper end 2 of the soft drive and the lower end 1 of the soft drive generate deformation relative to their own horizontal and vertical directions;

[0043] A fluid passage 4 is formed through the interior of the soft drive upper end 2 , the soft drive lower end 1 and the connecting chamber 3 .

[0044] Specifically, the flexible actuator lower end 1, flexible actuator upper end 2, and connecting chamber 3 correspond to the lower limbs, upper limbs, and joints of an animal, respectively. Both the flexible actuator lower end 1 and the flexible actuator upper end 2 are sized grid-like flexible actuators, with their chambers perpendicular to each other and capable of bending horizontally and vertically relative to themselves.

[0045] The connecting chamber 3 is composed of two chambers that are directly connected through a common side wall and are perpendicular to each other, and plays a role in directly connecting the lower end 1 of the soft drive and the upper end 2 of the soft drive.

[0046] Fluid passage 4 provides the driving fluid to the three-dimensional deformable soft actuator. The entire actuator consists of a single, continuous fluid passage, reducing the burden of supplying the driving fluid. Furthermore, the curvature formed by the lower and upper ends of the actuator, 1, 2, increases driving force, improving the operational efficiency of the resulting soft robot.

[0047] Specifically, the number of chambers in the lower end 1 and upper end 2 of the soft drive is different. The short section of the soft drive has a fluid passage opening. The upper end 2 of the soft drive is a grid-like soft drive with a fluid passage opening at one end. It is composed of multiple small soft cubes and one large soft cube. Each soft cube is hollowed out to form five chamber walls and one hollowed-out surface. The hollowed-out surfaces of the multiple small soft cubes are perpendicularly connected to the hollowed-out surface of the large soft cube at equal intervals, forming an internal chamber that includes the small soft cube chambers and the large soft cube chamber. The soft drive lower end 1 and the soft drive upper end 2 have similar structures and can have different numbers of small soft cubes according to different needs. One end of the soft drive lower end 1 is directly connected to the non-fluid passage opening end of the soft drive upper end 2, and the two are perpendicular to each other. At the connection point, the small soft cubes share a chamber wall, and the internal chamber of the large soft cube is connected, forming a three-dimensional deformable soft drive connected by a complete fluid passage. The three-dimensional deformable soft actuator is made of silicone rubber material and can simultaneously produce deformation in the vertical and horizontal directions relative to itself.

[0048] Example 2

[0049] like Figure 2 As shown, this embodiment provides a method for manufacturing a three-dimensional deformable soft driver, comprising the following steps:

[0050] Printing a first mold and a second mold, wherein the first mold has a protrusion and the second mold has a cavity, and assembling the first mold and the second mold;

[0051] pouring liquid silicone rubber into the assembled first and second molds, and then demolding to obtain a driver body consisting of a driver upper end, a soft driver lower end, and a connecting chamber;

[0052] Printing a cavity-enclosed mold, pouring liquid silicone rubber into the cavity-enclosed mold and placing the soft driver body, and then demoulding to obtain a cavity-enclosed driver body;

[0053] The cavity reinforcement mold is printed, liquid silicone rubber is poured into the cavity reinforcement mold, and the cavity-enclosed actuator body is placed on the mold, and then demolding is performed to obtain a three-dimensional deformable soft actuator.

[0054] In this embodiment, the main structure of a three-dimensional deformable soft actuator was analyzed and disassembled into three parts for sequential fabrication. To facilitate demolding, accurate fabrication, and improved fluid pressure tolerance, the corresponding molds were designed into three sets through precise dimensional adjustments: a main body mold set, a cavity-sealing mold set, and a cavity-reinforcement mold set. The main body mold set includes a combinable bottom raised mold and an upper cavity mold. The cavity-sealing mold set includes a recessed mold corresponding to the actuator surface shape. The cavity-reinforcement mold set includes a recessed mold corresponding to the other surface shape of the actuator.

[0055] The main mold assembly 5 is composed of a bottom raised mold and an upper cavity mold, and is used to form the main body and cavity frame of the three-dimensional deformable soft drive. The base of the bottom raised mold is a long flat cube, and the outer side of the upper cavity mold is a rectangular frame with a certain height. The upper part / interior of both contain two-part structures that form the lower end 1 and the upper end 2 of the soft drive respectively. In order to form the internal cavity of the soft drive upper end 2, a plurality of vertically fixed flat small cubes and a flat large cube are provided above the corresponding part of the bottom raised mold base. The long sides of the flat large cubes are parallel to the long sides of the base, and the width and height of the flat small cubes are connected to the long and height sides of the flat large cube at equal intervals. The two are perpendicular to each other and the long and wide sides are connected to the base. The overall structure corresponding to the soft drive lower end 1 on the bottom raised mold base is similar to and vertical to the above-mentioned overall structure, is specially the length and width face of flat small cube and is connected to the length and width face of flat large cube at equal intervals, and the two are perpendicular to each other and only have the length and width face of flat large cube to be horizontally connected with the base.For forming the chamber wall of soft drive upper end 2, the height and width face of a plurality of flat cubes are vertically connected to the frame inner wall of upper cavity mold at equal intervals, and the long side and opposite side inner wall leave a certain distance.For forming the chamber wall of soft drive lower end 1, the height and width face of a plurality of flat cubes are connected to the frame inner wall of upper cavity mold at equal intervals, and the other end is directly connected to the opposite side inner wall, forms multiple cavities. Upper cavity mold can be directly nested on the bottom raised mold, makes the flat cube of bottom be positioned at each cavity center of upper cavity mold, and the chamber sidewall of the soft drive upper end 2 that demoulding obtains and the chamber bottom surface of soft drive lower end 1 all are hollowed out like this, and the hollowed-out surfaces of the two are on the same plane. The cavity sealing mold set 6 is a recessed mold with a certain depth, comprising a complete flat surface. This flat surface is formed by the sidewalls of the upper end 2 of the soft actuator and the bottom surface of the lower end 1 of the soft actuator. This mold set 6 is used to seal the hollowed-out surface of the main body formed after the main body mold set is demolded. The cavity reinforcement mold set 7 is a recessed mold with a certain depth, comprising a complete flat surface. This flat surface is formed by the bottom surface of the upper end 2 of the soft actuator and the side surface of the lower end 1 of the soft actuator. This mold set 7 is used to reinforce the bottom surface of the upper end 2 of the soft actuator, thereby improving the fluid-carrying capacity of the soft actuator.

[0056] In this embodiment, first, liquid Dragon Ski n 30 silicone rubber is prepared, the main body mold group 5 is assembled, the liquid silicone rubber is poured into the main body mold group 5 and waited for solidification; then, the main body mold group 5 is demolded, the liquid silicone rubber is poured into the cavity closing mold group 6 and the main part of the three-dimensional deformable soft driver is placed thereon in a corresponding shape, and waited for solidification; then, the cavity closing mold group 6 is demolded, the liquid silicone rubber is poured into the cavity reinforcement mold group 7 and the three-dimensional deformable soft driver with a closed cavity formed thereon is placed thereon in a corresponding shape, and waited for solidification; finally, the cavity reinforcement mold group 7 is demolded to obtain the manufactured three-dimensional deformable soft driver 8 physical object.

[0057] Example 3

[0058] like Figure 3 As shown, this embodiment provides a control method for a three-dimensional deformable soft body driver, comprising:

[0059] According to the torque balance and the position relationship between the upper end of the soft actuator, the lower end of the soft actuator, and the connecting chamber, the relationship between the bending angle of the upper end of the soft actuator, the lower end of the soft actuator, and the connecting chamber and the input fluid pressure is obtained respectively;

[0060] According to the relationship between the bending angle and the input fluid pressure, a hysteresis inverse model is established;

[0061] The coefficients and thresholds of the established hysteresis inverse model are identified using real-coded genetic algorithm;

[0062] The identified hysteresis inverse model is combined with a sliding membrane controller to control the input fluid pressure of the three-dimensional deformable soft actuator.

[0063] In this example, segmented modeling is performed based on elastic dynamics and kinematic transformations. The connecting chamber 3 is considered a "joint," and the torque balance of the chamber itself is considered. The lower end 1 and upper end 2 of the 3D deformable soft actuator are then treated as "flexible links," and the overall torque balance is calculated for each. Once these individual components are modeled independently, kinematic transformations are used to describe the entire 3D deformable soft actuator model, creating a complete deformation model.

[0064] Specifically, first consider the lower end 1 and the upper end 2 of the soft drive, and use the following stress tensor calculation equation to obtain the stress tensor of the system in each direction, which is then used to calculate the torque generated by the soft body deformation:

[0065]

[0066] Among them, s ij is the stress tensor, W is the strain energy of the system, I1, I2, I3 are the strain invariants of the system in each direction, g ij is the Cauchy-Green strain tensor.

[0067] Since the strain torque of the soft material and the load torque of the fluid pressure are balanced relative to the reference bottom surface of the actuator, the following derivation relationship is formed:

[0068]

[0069] Among them, M 应变 is the strain moment of the soft material, M 流体 The load moment of the fluid pressure is derived from the moment balance of formula (2), which shows the direct relationship between the input fluid pressure and the bending angle of the actuator.

[0070] Analyzing the connected chamber 3, the following equilibrium equation exists for a single chamber:

[0071] M 侧 +M 底 +M 顶 =M 流体 , (3)

[0072] That is, relative to the same plane, the chamber side wall M 侧 、Top M 顶 With the bottom M 底 The sum of the stress moments is equal to the load moment of the fluid pressure.

[0073] The stress moment of the sidewall and top of the connecting chamber 3 can be calculated by integrating the stress tensor and volume deformation, while the stress moment of the bottom of the chamber can be directly calculated from its Young's modulus and inertia, thereby clarifying the relationship between the input fluid pressure of the connecting chamber 3 and the bending angle.

[0074] In this embodiment,

[0075] Consider a direct expression of the hysteresis inverse model:

[0076]

[0077] Among them, p(t) is the fluid pressure of the input driver, q(t) is the bending angle of the three-dimensional deformable soft body driver, a, b, c are the coefficients that need to be identified, and the input and output at both ends are the bending angle of the three-dimensional deformable soft body driver and the input fluid pressure, respectively. It is the product of the density function and the inverse hysteresis operator, and includes the operator threshold that needs to be identified.

[0078] During the identification process, there will be actual measured experimental data, including the input and output of the system (here for the inverse model, the input bending angle and output pressure). The output pressure corresponding to the actual input bending angle is the experimental value. The predicted value is the output pressure calculated theoretically when the same bending angle is input in the identification algorithm. The purpose of identification is to minimize the difference between the predicted value and the experimental value, so that the theoretically calculated value is consistent with the actual value. By setting the minimum sum of the squares of the difference between the experimental value and the predicted value during the identification process as the objective function, the real number coded genetic algorithm is used to identify and obtain the required inverse model coefficients and thresholds, and an accurate hysteresis inverse model is obtained and compensated. This method can also be used to identify the hysteresis characteristics of inner ring proportional valves.

[0079] It should be noted that the three-dimensional deformable soft actuator designed in this embodiment has its own hysteresis characteristic, which refers to the relationship between the input air pressure and the bending angle mentioned above. This is the outer loop. The inner loop is for the proportional valve that supplies air to the soft actuator. The proportional valve itself also has hysteresis characteristics, which refers to the relationship between its input voltage and output air pressure. When used in practice, the input air pressure and bending angle of the soft actuator mentioned above are replaced with the input voltage and output air pressure of the proportional valve, respectively.

[0080] In this embodiment, the following proxy sliding mode control output is designed:

[0081]

[0082] Where k and l are adjustable control parameters, n is the integral operator, sat(·) is the saturation function, F is the output of the sliding mode controller based on the control pressure (i.e., the air pressure supplied to the soft actuator), and s is the sliding surface. The input of the proxy sliding mode control is the error between the soft actuator's current position and the target position, and the output is the control pressure (i.e., the air pressure supplied to the soft actuator).

[0083] This control input effectively limits excessive input fluid pressure when tracking errors are large, achieving control compliance. To ensure control accuracy and test stability, a reinforcement learning parameter adjustment method that considers debugging stability is employed to limit update time and ensure the stability of the learning process. Furthermore, a baseline controller is prepared to replace parameters that exceed the baseline score during the learning process, preventing unstable control output from damaging the 3D deformable soft actuator during debugging. The baseline controller input is the error between the soft actuator's current and target positions, and the output is the control air pressure.

[0084] A reinforcement learning parameter adjustment method that considers debugging stability is adopted, and a deterministic policy gradient algorithm (DPG) is used for reinforcement learning training. Specifically, a conservative and stable PID control with fixed parameters is introduced as the baseline controller and the training process is initialized to obtain an initial reward score. To avoid controller instability during parameter adjustment, when the reward score of the PID controller during parameter change is greater than the initial reward score, a conservative and stable PID controller is used instead to ensure stability during parameter adjustment and prevent unstable control output from damaging the three-dimensional deformable soft body actuator during debugging. After executing the control algorithm, the system's feedback state (control air pressure), PID parameters, and reward score are returned to the DPG algorithm for further training until control is stable. In addition, the PID parameter update should be performed after the control is completed to prevent instability caused by changing the PID parameters during the control process.

[0085] Example 4

[0086] like Figure 4 As shown, this embodiment provides a control system for a three-dimensional deformable soft body driver, including:

[0087] The bending sensor is used to obtain the bending angle value of the three-dimensional deformable soft actuator and transmit the obtained angle to the host computer;

[0088] The position sensor is used to obtain the displacement change value of the three-dimensional deformable soft body driver and transmit the obtained displacement change to the host computer;

[0089] The host computer performs trajectory tracking calculation based on the obtained bending angle value and displacement change value, and generates a control signal for controlling the pneumatic proportional valve;

[0090] The pneumatic proportional valve is used to control the fluid pressure input by the air pump to the three-dimensional deformation soft body driver according to the control signal of the host computer.

[0091] In this embodiment, one end of the pneumatic proportional valve is connected to an air pump, which continuously provides it with gas of sufficient pressure. Under the action of the control signal, the pneumatic proportional valve accurately adjusts the opening of the internal solenoid valve air circuit, allowing gas of a specific pressure to pass through and be supplied to the three-dimensional deformable soft actuator. The three-dimensional deformable soft actuator bends under the action of air pressure, and the corresponding bending angle and position changes are transmitted to the communication module in the form of a voltage signal. The communication module converts the voltage signal into the corresponding angle value and displacement value and feeds it back to the host computer. The host computer performs calculations in the trajectory tracking control algorithm based on the feedback signal, generates a new control signal, and finally forms a complete closed-loop control system. In particular, under the action of this control system, the three-dimensional deformable soft actuator can become the driving leg of a soft crawling robot, or the bending finger of a soft gripper, with multifunctional application scenarios.

[0092] In this embodiment, the bending sensor is a bending angle sensing element that can be attached to the surface of the three-dimensional deformable soft actuator and provides feedback on the bending angle of the three-dimensional deformable soft actuator through changes in internal resistance during bending.

[0093] The position sensor is a position coordinate sensing element that can be attached to one end of a three-dimensional deformable soft actuator to feedback the changes in the actual coordinates of its endpoint position in the calibration coordinate system.

[0094] The pneumatic proportional valve is a fluid opening control device that can accurately change the opening of the internal electromagnetic switch in real time under the action of the control signal, allowing the fluid to pass through it at a specific pressure, thereby providing accurate driving fluid pressure for the soft actuator.

[0095] The communication module is a signal conversion component that can collect, store and convert the feedback signals of each sensor into a signal form that can be processed by the control host. It is connected between each sensor and the control host.

[0096] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. A control method for a three-dimensional deformable soft body driver, characterized in that: The three-dimensional deformable soft actuator comprises: a soft actuator upper end, a soft actuator lower end, and a connecting chamber; The upper end of the soft drive is connected to the lower end of the soft drive via a connecting chamber; The bending directions of the upper end of the soft driver and the lower end of the soft driver are perpendicular to each other, and the upper end of the soft driver and the lower end of the soft driver generate deformation in the horizontal direction and the vertical direction relative to themselves; A fluid passage is formed through the upper end of the soft driver, the lower end of the soft driver and the interior of the connecting chamber; The control method comprises: According to the torque balance and the position relationship between the upper end of the soft actuator, the lower end of the soft actuator, and the connecting chamber, the relationship between the bending angle of the upper end of the soft actuator, the lower end of the soft actuator, and the connecting chamber and the input fluid pressure is obtained respectively; According to the relationship between the bending angle and the input fluid pressure, a hysteresis inverse model is established; The coefficients and thresholds of the established hysteresis inverse model are identified using real-coded genetic algorithm; The identified hysteresis inverse model is combined with a sliding membrane controller to control the input fluid pressure of the three-dimensional deformable soft actuator.

2. The control method of a three-dimensional deformable soft body driver according to claim 1, wherein: Also includes: A hysteresis inverse model is established based on the relationship between the control voltage of the pneumatic proportional valve of the three-dimensional deformable soft body actuator and the output air pressure of the three-dimensional deformable soft body actuator; the coefficients and thresholds of the established hysteresis inverse model are identified using a real-coded genetic algorithm; The identified hysteresis inverse model is combined with the sliding membrane controller to control the control voltage of the pneumatic proportional valve of the three-dimensional deformable soft body actuator.

3. The control method of a three-dimensional deformable soft body driver according to claim 1, wherein: The parameters of the agent synovial controller are updated using reinforcement learning parameter adjustment method.

4. The control method of a three-dimensional deformable soft body driver according to claim 1, wherein: The connecting chamber includes an upper chamber and a lower chamber, the upper chamber and the lower chamber are connected via a side wall, and the chamber directions of the upper chamber and the lower chamber are perpendicular to each other.

5. The control method of a three-dimensional deformable soft body driver according to claim 1, wherein: An opening is formed on the end surface of the upper end of the soft driver or the lower end of the soft driver, and the opening is connected to the fluid passage.

6. A control system for a three-dimensional deformable soft body driver, using a control method for a three-dimensional deformable soft body driver according to any one of claims 1 to 5, characterized in that: include: The bending sensor is used to obtain the bending angle value of the three-dimensional deformable soft actuator and transmit the obtained angle to the host computer; The position sensor is used to obtain the displacement change value of the three-dimensional deformable soft body driver and transmit the obtained displacement change to the host computer; The host computer performs trajectory tracking calculation based on the obtained bending angle value and displacement change value, and generates a control signal for controlling the pneumatic proportional valve; The pneumatic proportional valve is used to control the fluid pressure input by the air pump to the three-dimensional deformation soft body driver according to the control signal of the host computer.

Images