A multimodal soft robot tail and its preset time state attraction control method

Abstract

The application discloses a multimodal soft body machine tail and a preset time state attraction control method thereof, and mainly relates to the technical field of unmanned aerial vehicle and robot control. The machine tail comprises a support frame assembly, a driving assembly and a double-soft-arm assembly comprising a built-in bending sensor. The control method is based on a preset time state attraction (PTSAC) strategy, which guides the system state to converge without overshoot within a preset time by constructing a state attraction surface that fuses speed, direction and preset time functions. The application realizes the multimodal reuse of unmanned aerial vehicle in the air, terrestrial habitat grabbing and water surface swimming with a single double-arm structure, greatly reducing the system weight. At the same time, the time complexity of the control method is only O(1), breaking through the algorithm bottleneck of the microcontroller, and significantly improving the robustness and flexibility of the flexible structure under strong disturbance.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) and soft robot control technology, and more specifically, to a method for attracting and controlling the tail of a multimodal soft robot and its preset time state. Background Technology

[0002] In recent years, the application of unmanned aerial vehicles (UAVs) in complex environments has been increasing. Single-mode UAVs are no longer sufficient to meet diverse mission requirements, making amphibious UAVs capable of operating across air, land, and water a research hotspot. However, existing multi-domain UAVs typically employ simple structural stacking methods, such as independently mounting flight propulsion units, land-based walking mechanisms, and underwater propellers to achieve operational capabilities in different environments. This non-modular design results in extremely low component reuse rates, leading to significant weight redundancy. This not only increases energy consumption but also makes it difficult to apply to micro-UAVs, especially biomimetic flapping-wing aircraft, which are extremely sensitive to loads.

[0003] To address the issue of weight redundancy, some studies have attempted to use flexible dual soft robotic arms as the tail of drones, leveraging their deformability to achieve various modes such as flight steering, perching and grasping, and underwater maneuvering. However, the soft arm structure has a highly nonlinear kinematic and dynamic model, making its precise control extremely challenging. Micro-drones typically only carry microcontrollers with extremely low computing power (such as STM32 chips with a maximum clock frequency of 168MHz). Due to this limitation, existing advanced algorithms such as neural network control and model predictive control (MPC) cannot be deployed due to excessive computational load. While traditional preset time sliding mode control (PTSMC) has a fast convergence speed, it is prone to generating huge transient overshoot and control chatter when faced with strong wind disturbances and mode switching shocks. This not only leads to flight instability but also easily causes irreversible physical damage to the fragile soft arm structure.

[0004] Therefore, designing a control method with low computational complexity, strong robustness, and fast convergence without overshoot, while ensuring system lightweighting and multimodal function reuse, is a technical challenge that urgently needs to be solved in the field of micro amphibious unmanned aerial vehicles. Summary of the Invention

[0005] The present invention aims to solve at least one problem existing in the prior art, and provides a method for attracting and controlling the tail of a multimodal software machine and its preset time state.

[0006] To achieve the above objectives, the present invention provides a multimodal soft robot tail assembly, comprising a support frame assembly, a drive assembly, and a dual soft arm assembly, wherein:

[0007] The support frame assembly is used to fix the aircraft body to the aircraft body and to provide an installation reference for the drive assembly and the dual soft arm assembly;

[0008] The drive assembly includes a left servo driver, a right servo driver, and a drive turntable that is fixedly connected to the output shafts of the left servo driver and the right servo driver respectively. The left servo driver and the right servo driver are both fixedly mounted on the support frame assembly.

[0009] The dual soft arm assembly includes an independent left soft arm and a right soft arm. The bottom ends of the left soft arm and the right soft arm are fixedly connected to the support frame assembly, and the top ends of the left soft arm and the right soft arm are fixedly provided with rigid extension members.

[0010] Both the left and right soft arms have flexible traction components penetrating their interiors. One end of each flexible traction component is fixedly connected to the corresponding rigid extension component, and the other end is wound around the corresponding drive turntable. The left and right servo drivers rotate to drive the corresponding drive turntables. By extending and retracting the flexible traction components, the left and right soft arms are pulled to achieve independent or synchronous bending deformation.

[0011] Furthermore, both the left and right soft arms are formed by casting a mixture of silicone elastomer and hollow microspheres; and each of the left and right soft arms is embedded with a bending sensor, which is used to measure and provide feedback on the bending angle of the left and right soft arms in real time.

[0012] Furthermore, the feature is that the outer edges of the left and right soft arms are both serrated, and the serrated structure is used to ensure that the left and right soft arms bend in an arc shape when traction is applied and to increase the friction when the object is grasped or suspended.

[0013] Furthermore, the tooth spacing of the serrated structure of the left and right soft arms decreases from the end near the drive turntable to the end near the rigid extension member, and the tooth size of the serrated structure decreases from the end near the drive turntable to the end near the rigid extension member.

[0014] Furthermore, the left soft arm, the right soft arm, and the support frame assembly are all wrapped with waterproof elastic fabric to form a closed, integrated tail fin shape.

[0015] The present invention also provides a preset time state attraction control method for the tail of a multimodal soft machine, applied to the aforementioned tail of a multimodal soft machine, the method comprising the following steps:

[0016] S1. Obtain the desired tail pitch angle and twist angle of the UAV. Based on the predefined inverse kinematic fitting model from the task space to the joint space, calculate the desired bending angles of the left and right soft arms.

[0017] S2. The actual bending angles of the left and right soft arms are acquired in real time, and the trajectory tracking error is calculated by combining the expected bending angle.

[0018] S3. Based on the trajectory tracking error, construct a preset time state attraction function that includes a velocity function, a direction function, and a preset time function, to constrain the convergence direction and convergence rate of the trajectory tracking error;

[0019] S4. Combining the preset time state attraction function and the derivative of the trajectory tracking error, construct a preset time attraction surface, and calculate the attraction error of the system state deviating from the preset time attraction surface;

[0020] S5. Substitute the attraction error into the preset soft arm dynamics equation to calculate the control torque used to control the drive component, so that the system state converges to the origin neighborhood along the preset time attraction surface within the expected convergence time.

[0021] Furthermore, the preset time state attraction function in step S3 Represented as:

[0022]

[0023] In the formula, The preset time function for adaptively updating the gain based on the expected convergence time is expressed as:

[0024]

[0025] In the formula, and To define a time function; For time variables, This is the adaptive switching time parameter.

[0026] It is a velocity function. The direction function; the velocity function and direction function They are represented as follows:

[0027]

[0028]

[0029] In the formula, For trajectory tracking error; , , All are positive numbers. It is a positive real constant close to 0.

[0030] Furthermore, the attraction error in step S4 Represented as:

[0031]

[0032] In the formula, This is the derivative of the trajectory tracking error.

[0033] Furthermore, in step S5, the control torque The calculation formula is expressed as:

[0034]

[0035] In the formula, The control gain is a positive real constant. The mass of the left or right soft arm. The initial length of the flexible traction component that drives the soft arm.

[0036] Furthermore, the direction function constant gain Set to 1000, and the preset time function Adaptive switching time parameters Set to infinitely approach the desired convergence time The constant.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. High space reuse rate, enabling amphibious multimodal operations: This invention breaks the constraints of multi-component stacking, utilizing only a set of lightweight dual soft arms to simulate bird turning during flight (differential bending), bat grasping during roosting (uniform bending and contraction), and fish swimming (cross-flapping) through shape switching. This "one-piece-for-multiple-use" design concept greatly reduces the takeoff weight and system complexity of the UAV.

[0039] 2. Extremely low computational power consumption, breaking through the performance bottleneck of microcontrollers: The Preset Time State Attraction Control (PTSAC) method proposed in this invention uses basic algebraic operations as its core computations, with a stable time complexity of only O(1). High-frequency real-time control can be achieved on mainstream microcontrollers, successfully breaking through the computational power bottleneck of low-performance processors for micro flapping-wing robots, and providing a new approach for deploying advanced control algorithms on resource-constrained platforms.

[0040] 3. Strong robustness and compliance, perfectly eliminating transient overshoot: This invention constructs a pre-defined time attraction surface (PTAS), guiding the system state to converge smoothly like a gravitational field. When encountering strong time-varying disturbances or mode-switching shocks, this method can effectively suppress chattering, reduce trajectory tracking errors by more than three orders of magnitude, completely eliminate overshoot, significantly improve flight stability, and effectively protect the physical structure of the soft arm.

[0041] 4. Optimized Mechanical Distribution and Enhanced Grip Performance: This invention achieves a uniform stress distribution along the length of the flexible arm during bending by using a gradually varying serrated structure on the outer edge (tooth spacing and size decrease from the root to the tip). This effectively avoids stress concentration and significantly improves the structure's durability and tear resistance. Simultaneously, this gradually varying serrated structure forms a more stable mechanical interlock with the contact surface during gripping or suspension, greatly improving gripping stability during multimodal switching and providing reliable protection for the UAV's habitation and operation in complex environments. Attached Figure Description

[0042] Figure 1 This is a front view of the overall structure of the multimodal software machine tail in the embodiment of the present invention in its initial neutral state.

[0043] Figure 2 This is a schematic diagram of the unidirectional bending and retraction state of the dual soft robotic arms (terrestrial dwelling and grasping modes) in an embodiment of the present invention;

[0044] Figure 3 This is a schematic diagram of the differential bending state (aerial flight turning mode) of the dual soft robotic arms in an embodiment of the present invention;

[0045] Figure 4 This is a simulation curve comparing the trajectory tracking and error of the tail pitch and twist angles using preset time state attraction control and traditional preset time sliding mode control in an embodiment of the present invention.

[0046] Figure 5 This is a phase trajectory diagram of the tracking error of the soft robotic arm in an embodiment of the present invention;

[0047] Figure 6 This is a flowchart illustrating the hardware components and manufacturing process of the multimodal software machine tail in this embodiment of the invention.

[0048] Figure 7 This is a schematic diagram of the kinematic coordinate system and pose parameter definition of the soft robotic arm in an embodiment of the present invention;

[0049] Figure 8 This is a system logic block diagram of the preset time state attraction control method in an embodiment of the present invention.

[0050] Explanation of reference numerals in the attached drawings: 1-Support frame assembly; 2-Drive assembly; 3-Dual soft arm assembly; 31-Left soft arm; 32-Right soft arm; 4-Flexible traction component; 5-Rigid extension component; 6-Sawtooth structure. Detailed Implementation

[0051] The techniques described below can be modified in various ways and have multiple embodiments, which are described in detail below with reference to the accompanying drawings. However, this does not mean that the techniques described below are limited to the specific embodiments. It should be understood that the present invention includes all similar modifications, equivalents, and substitutions without departing from the spirit and scope of the techniques described below.

[0052] Example 1

[0053] Combination Figures 1 to 3 As shown, this embodiment provides a multimodal soft robot tail section, which mainly consists of a support frame assembly 1, a drive assembly 2, and a dual soft arm assembly 3.

[0054] The support frame assembly 1 is used to fix it to the aircraft body (such as a micro bionic flapping-wing aircraft) and provide structural support and installation reference for the entire tail section.

[0055] The drive assembly 2 includes a left servo driver, a right servo driver, and a drive turntable fixedly connected to the output shafts of both. Both the left and right servo drivers are fixedly mounted on the support frame assembly 1. Preferably, to meet the torque requirements of multimodal high-frequency flapping, the servo driver uses a waterproof servo motor with a maximum output torque greater than or equal to 4.5 kg·cm, and is powered by an independent lithium polymer battery.

[0056] The dual flexible arm assembly 3 includes an independent left flexible arm 31 and a right flexible continuous arm 32. The bottom ends of both are fixedly connected to the support frame assembly 1, and the top ends of both are fixedly equipped with rigid extension members 5 (e.g., fiber rods). Flexible traction members 4 (e.g., carbon wire) penetrate the interior of both the left and right flexible arms 31 and 32. One end of the flexible traction member 4 is fixedly connected to the rigid extension member 5, and the other end is wound around the corresponding drive turntable. By rotating the drive turntable via a servo driver, the flexible traction member 4 can be extended or retracted, thereby pulling the corresponding flexible arm to achieve independent or synchronous bending deformation.

[0057] To optimize the physical properties of the flexible arms, the left flexible arm 31 and the right flexible arm 32 are cast from a mixture of silicone elastomer (such as Ecoflex 00-30) and hollow microspheres (hollow floats) in a 6:4 mass ratio. This material ratio ensures flexibility while reducing overall weight. Furthermore, the outer edges of the flexible arms are both serrated, which ensures ideal arc-shaped bending under traction and increases friction when grasping or suspending objects.

[0058] The serrated structure 6 is not uniformly distributed, but rather adopts a gradient design: the tooth spacing and tooth size decrease from the end near the drive assembly 2 (i.e., near the drive turntable) to the end near the rigid extension member 5. Specifically, the tooth spacing and tooth size are larger near the drive end (root), providing stronger structural support and gripping force for the soft arm during the initial bending phase; the tooth spacing and tooth size are smaller near the tip (end), allowing the soft arm end to achieve a finer bending curvature, adapting to more complex gripping or suspension surfaces. When the drive assembly 2 rotates and pulls the internal flexible traction member 4, this gradient serrated structure guides the soft arm to achieve a more uniform and smoother arc-shaped bend along its length, avoiding localized excessive deformation or tearing caused by stress concentration. It also significantly increases the friction between the soft arm and the gripped object (such as a branch or rock wall) or suspension surface (such as a beam), improving reliability and stability in terrestrial habitat and gripping modes.

[0059] Furthermore, the wire diameter of the flexible traction member 4 gradually decreases from the end near the drive assembly 2 (i.e., near the drive turntable) to the end near the rigid extension member 5.

[0060] To form a closed overall tail wing shape and take into account waterproof performance, the left soft arm 31, the right soft arm 32 and the support frame assembly 1 are all wrapped with waterproof elastic fabric.

[0061] To achieve closed-loop control, both the left soft arm 31 and the right soft arm 32 are equipped with bending sensors to measure and provide feedback on the corresponding bending angle in real time.

[0062] Example 2

[0063] Combination Figures 6 to 8 As shown, this embodiment provides a preset time state attraction control method for the tail of a multimodal software machine based on the above-mentioned method. This method is executed on a microcontroller (such as an STM32F407ZGT6), drives a waterproof servo motor via a PWM signal, and receives feedback signals from a bending sensor to form a complete control loop.

[0064] like Figure 8As shown, the control system of this invention adopts a classic closed-loop feedback structure. The controller receives desired tail attitude commands (including desired pitch and twist angles) from the host computer or flight control system. These commands are first converted into desired bending angles of the left and right flexible arms by the inverse kinematics module. Simultaneously, bending sensors embedded within the flexible arms collect the actual bending angles in real time and convert them into actual angle values ​​through a sensor mapping module. The difference between the desired and actual angles yields the trajectory tracking error, which is input to the core preset time-state attraction function (PTSAF) control module. Based on preset convergence time parameters, the control module calculates the required control torque in real time and outputs it to the servo driver, driving the flexible traction component to extend and retract, thereby achieving precise control of the bending deformation of the flexible arms.

[0065] To achieve the conversion from the desired tail posture to the joint angles of the soft arm, a precise kinematic model is first required. For example... Figures 6-7 As shown, this invention establishes a three-dimensional spatial coordinate system and a soft arm joint coordinate system with the center of the support frame component as the origin. Through the geometric relationships of this coordinate system, the mapping relationship between the tail pitch angle and twist angle and the bending angles of the left and right soft arms can be derived.

[0066] Furthermore, within the drive space, the rotation angle of the servo drive (waterproof servo motor) With the The actual bending angle of the soft arm It satisfies the following exact linear mapping relationship:

[0067]

[0068] In the formula, This refers to the perpendicular distance between the internal flexible traction component (carbon wire) and the central axis of the flexible arm. Let be the radius of the drive turntable. This linear relationship makes the calculation of the controller's drive commands extremely simple.

[0069] For real-time feedback from the bending sensor, due to the inherent nonlinearity of the sensor and the limited computing power of the microcontroller, this invention does not employ complex lookup tables or interpolation algorithms. Instead, it uses offline calibration experiments to collect sensor ADC sampling values ​​at different bending angles and employs a sixth-order polynomial for least-squares fitting to establish the sensor ADC sampling value... Compared with the actual bending angle Kinematic mapping between them:

[0070]

[0071] In the formula, to All coefficients are constant fitting coefficients determined through calibration experiments. This polynomial model transforms the complex nonlinear feedback of the sensor into basic algebraic operations with a computational complexity of only O(1), greatly reducing the computational burden of the system and making it possible to achieve high-frequency control on a microcontroller with a main frequency of 168MHz.

[0072] To achieve precise torque control, a dynamic model of the flexible arm needs to be established. Considering the bending deformation characteristics of the flexible arm, this invention equates it to constrained link motion and models it based on Lagrangian mechanics.

[0073] set up For the mass of a single soft arm, The initial length of the flexible traction component. The total kinetic energy of the system. Total gravitational potential energy The derivations are as follows:

[0074]

[0075]

[0076] Based on the Lagrange equations, the dynamic equations under ideal conditions can be derived:

[0077]

[0078] However, external time-varying disturbances inevitably exist in real systems. (such as gusts of wind, water flow impact) and unmodeled dynamic uncertainties (e.g., material hysteresis, friction, etc.). After considering these factors, the first... The complete dynamic equations for the soft arm are:

[0079]

[0080] In the formula, This is the control torque required to drive the motor. Let be the acceleration due to gravity. This equation clearly reveals the nonlinear relationship between the system state and the control torque, providing a theoretical basis for subsequent controller design.

[0081] The core innovation of this invention lies in proposing a novel Preset Time State Attraction Controller (PTSAC). This controller aims to solve the problems of overshoot and chattering that traditional control methods are prone to under strong disturbances, and achieves smooth convergence of errors without overshoot within a preset time.

[0082] Define the trajectory tracking error as ,in For the desired bending angle, This represents the actual bending angle. The control objective is to design the control torque. This causes tracking error Within the preset convergence time It converges to the neighborhood of the origin without overshooting during the entire process.

[0083] To achieve this goal, the present invention first constructs a preset time-state attraction function (PTSAF). Its expression is:

[0084]

[0085] This function consists of three core parts, which control the convergence rate, convergence direction, and convergence time, respectively:

[0086] 1. Velocity function Used to control the rate of error convergence and prevent sudden changes in control torque. Its expression is:

[0087]

[0088] in, For trajectory tracking error; , All are positive numbers. This is a positive real constant close to 0, used to prevent the denominator from being zero. This function provides a large convergence rate when the error is large, and the convergence rate gradually decreases as the error approaches zero, thus achieving smooth convergence.

[0089] 2. Direction function This is used to constrain the convergence direction of the error, ensuring that the error always approaches the origin. To guarantee the function's differentiability throughout and avoid chattering caused by the sign function, this invention uses the arctangent function to approximate the ideal sign function.

[0090]

[0091] In the formula, As a positive constant, the larger the value, the closer the function is to the sign function. By choosing a sufficiently large value... (e.g., 1000) can achieve fast symbolic response characteristics while ensuring smoothness.

[0092] 3. Preset time function This function is used to adaptively update the gain in real time based on the expected convergence time. It employs a piecewise design.

[0093]

[0094] In the formula, and To define a time function; For time variables, This is the adaptive switching time parameter.

[0095] Furthermore, a preset time attraction surface (PTAS) and the attraction error caused by the system state deviating from this attraction surface are defined. for:

[0096]

[0097] In the formula, This is the derivative of the trajectory tracking error.

[0098] By combining the above dynamic equations, the control torque ultimately input to the servo drive can be obtained through inverse solving. for:

[0099]

[0100] In the formula, The control gain is a positive real constant. The mass of the left or right soft arm. The initial length of the flexible traction component that drives the soft arm. This is used to precisely compensate for the physical inertia of the soft arm.

[0101] In actual flight and swimming modes, through extensive experimental optimization and theoretical analysis, the key parameters of the controller proposed in this invention are preferably selected as follows:

[0102] Velocity function parameters: .in, The convergence speed gain is determined when the error is large. Deceleration characteristics when control error approaches zero It is a very small positive real number used to prevent the denominator from being zero and to ensure numerical stability.

[0103] Direction function parameters: = 1000. This value is large enough to allow the arctangent function to quickly approximate the sign function when the error is non-zero, while maintaining the function's differentiability throughout, thus avoiding chattering caused by the discontinuity of the sign function in traditional sliding mode control.

[0104] Preset time function parameters: for the preset expected convergence time (as set) Adaptive switching time parameters It needs to be set to infinitely close to constants (such as) This design ensures that the system operates within [0, ... Maintaining high gain within the interval to achieve fast convergence, Then switch to low gain to maintain steady-state accuracy and suppress noise interference.

[0105] Control gain: This gain is used to adjust the attraction error. The convergence rate is high, and a larger gain can accelerate the approach speed of the system state to the preset time attraction surface (PTAS), but it needs to be considered in conjunction with hardware limiting to avoid saturation.

[0106] To protect the servo motor from damage caused by excessive impact, and to prevent irreversible physical tearing or material fatigue of the flexible arm due to overload, this invention incorporates a robust safety protection mechanism at the hardware level. Specifically, the controller calculates the control torque... Before being output to the servo driver, the amplitude is limited by the hardware program. Within the safe range. The selection of this limit value comprehensively considers the tensile strength of the software material, the rated torque of the servo motor, and the actual load requirements under multimodal operating conditions, ensuring both the effectiveness of the control algorithm and the physical safety of the system.

[0107] Example 3

[0108] To further verify the effectiveness of the control algorithm proposed in this invention, this embodiment underwent rigorous comparative testing in a simulation environment. An existing preset time sliding mode control (PTSMC) algorithm was introduced as a control group. To ensure fairness in the comparison, both algorithms were implemented on the same hardware platform (a 168MHz STM32F407 chip) and used the same sampling frequency and control cycle.

[0109] The initial state of the system is zero (i.e.) Under the condition of explicitly injecting complex time-varying disturbances into the system. and unmodeled dynamic uncertainty This simulates extremely harsh outdoor airflow and water flow environments. Among them:

[0110] The derivative of the actual bending angle; the external disturbance is set to... Simulate periodic environmental disturbances; unmodeled dynamic uncertainties are set as follows: To simulate the nonlinear properties of materials.

[0111] Subsequently, trajectory tracking comparison tests were conducted, focusing on disturbance rejection capability, overshoot elimination capability, and computational efficiency.

[0112] 1. Disturbance immunity and overshoot elimination capability

[0113] like Figure 4As shown in the error comparison curves, when faced with strong external disturbances, the traditional PTSMC algorithm produces severe transient overshoot and control chattering. Overshoot not only leads to flight instability but also causes additional physical stress on the soft arm structure, which can easily lead to material fatigue or even tearing during long-term operation.

[0114] In comparison, the PTSAC algorithm of this invention exhibits superior control performance. The root mean square error (RMS) of trajectory tracking of the left soft arm is significantly lower than that of the control group. sudden drop The right soft arm then from Down to The error level was reduced by three orders of magnitude. More importantly, the entire tracking process curve was smooth and oscillating, completely eliminating the overshoot phenomenon commonly found in traditional sliding mode control, greatly protecting the soft arm from physical impacts, and extending the system's service life.

[0115] 2. Convergence verification of the attraction surface

[0116] like Figure 5 The tracking error phase trajectory diagram shown further reveals the convergence mechanism of the present invention. In the diagram, the horizontal axis represents the tracking error. The vertical axis represents the error derivative. By introducing a pre-defined time-state attraction function (PTSAF), the system state is forcibly constrained in the phase space. Regardless of the initial state, the system state can smoothly converge to a minimal neighborhood of the origin within a pre-defined time interval along the pre-defined attraction surface (PTAS) trajectory. The phase trajectory curve is smooth and continuous, without oscillations or crossing phenomena, fully verifying the effectiveness and superiority of the pre-defined time-state attraction surface.

[0117] 3. Computational complexity and real-time performance

[0118] In a test of the total computation time for a microprocessor executing 5000 control cycles, the total time consumed by this invention (PTSAC) was only [amount missing]. Superior to the control group (PTSMC) This result demonstrates that the present invention achieves superior control performance while maintaining an extremely low time complexity of O(1). For a microcontroller with a main frequency of only 168MHz, the average time for each control cycle is less than 6 microseconds, fully meeting the requirements for high-frequency real-time control and successfully breaking through the computing power bottleneck of low-performance processors for micro flapping-wing robots.

[0119] In summary, the preset time state attraction control method proposed in this invention is significantly superior to traditional control methods in multiple dimensions such as computational efficiency, anti-disturbance capability, and overshoot elimination, providing an ideal solution for tail control of micro amphibious unmanned aerial vehicles.

[0120] Although the present invention has been described in detail above with general descriptions and specific embodiments, some modifications or improvements can be made to it. The above descriptions are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Other changes and modifications made by those skilled in the art without departing from the spirit and scope of the present invention are still included within the scope of protection of the present invention.

Claims

1. A multimodal soft machine tail section, characterized in that, This includes a support frame assembly, a drive assembly, and a dual-soft-arm assembly, wherein: The support frame assembly is used to fix the aircraft body to the aircraft body and to provide an installation reference for the drive assembly and the dual soft arm assembly; The drive assembly includes a left servo driver, a right servo driver, and a drive turntable that is fixedly connected to the output shafts of the left servo driver and the right servo driver respectively. The left servo driver and the right servo driver are both fixedly mounted on the support frame assembly. The dual soft arm assembly includes an independent left soft arm and a right soft arm. The bottom ends of the left soft arm and the right soft arm are fixedly connected to the support frame assembly, and the top ends of the left soft arm and the right soft arm are fixedly provided with rigid extension members. Both the left and right soft arms have flexible traction components penetrating their interiors. One end of each flexible traction component is fixedly connected to the corresponding rigid extension component, and the other end is wound around the corresponding drive turntable. The left and right servo drivers rotate to drive the corresponding drive turntables. By extending and retracting the flexible traction components, the left and right soft arms are pulled to achieve independent or synchronous bending deformation.

2. The multimodal soft machine tail section according to claim 1, characterized in that, Both the left and right soft arms are formed by casting a mixture of silicone elastomer and hollow microspheres; and each of the left and right soft arms is embedded with a bending sensor, which is used to measure and provide feedback on the bending angle of the left and right soft arms in real time.

3. The multimodal soft machine tail section according to claim 1, characterized in that, The outer edges of both the left and right soft arms are serrated. The serrated structure is used to ensure that the left and right soft arms bend in an arc shape when traction is applied and to increase the friction when the object is grasped or suspended.

4. The multimodal soft machine tail section according to claim 3, characterized in that, The tooth spacing of the serrated structures of the left and right soft arms decreases from the end near the drive turntable to the end near the rigid extension member, and the tooth size of the serrated structures decreases from the end near the drive turntable to the end near the rigid extension member.

5. The multimodal soft machine tail section according to claim 1, characterized in that, The left soft arm, the right soft arm, and the support frame assembly are all wrapped in waterproof elastic fabric to form a closed, integrated tail fin shape.

6. A method for attracting and controlling the tail of a multimodal software machine based on a preset time state, characterized in that, Utilizing the multimodal soft machine tail according to any one of claims 1-5, the method comprises the following steps: S1. Obtain the desired tail pitch angle and twist angle of the UAV. Based on the predefined inverse kinematic fitting model from the task space to the joint space, calculate the desired bending angles of the left and right soft arms. S2. The actual bending angles of the left and right soft arms are acquired in real time, and the trajectory tracking error is calculated by combining the expected bending angle. S3. Based on the trajectory tracking error, construct a preset time state attraction function that includes a velocity function, a direction function, and a preset time function, to constrain the convergence direction and convergence rate of the trajectory tracking error; S4. Combining the preset time state attraction function and the derivative of the trajectory tracking error, construct a preset time attraction surface, and calculate the attraction error of the system state deviating from the preset time attraction surface; S5. Substitute the attraction error into the preset soft arm dynamics equation to calculate the control torque used to control the drive component, so that the system state converges to the origin neighborhood along the preset time attraction surface within the expected convergence time.

7. The control method according to claim 6, characterized in that, The preset time state attraction function in step S3 Represented as: ； In the formula, The preset time function for adaptively updating the gain based on the expected convergence time is expressed as: ； In the formula, and To define a time function; For time variables, For adaptive switching time parameters; Let velocity be the function. The direction function; the velocity function and direction function They are represented as follows: ； ； In the formula, For trajectory tracking error; , , All are positive numbers. It is a positive real constant close to 0.

8. The control method according to claim 7, characterized in that, The attraction error in step S4 Represented as: ； In the formula, This is the derivative of the trajectory tracking error.

9. The control method according to claim 8, characterized in that, In step S5, the control torque The calculation formula is expressed as: ； In the formula, The control gain is a positive real constant. The mass of the left or right soft arm. The initial length of the flexible traction component that drives the soft arm.

10. The control method according to claim 9, characterized in that, The direction function constant gain Set to 1000, and the preset time function Adaptive switching time parameter Set to infinitely approach the desired convergence time The constant.

Images