Software robot model predictive control method and device, electronic equipment and medium

Abstract

The application provides a soft robot model predictive control method and device, electronic equipment and medium, and the method comprises the following steps: obtaining a historical control quantity of a soft controller of a soft robot, and constructing a soft control model of the soft controller according to the historical control quantity; obtaining a real-time control quantity of the soft controller, determining a total disturbance error of the soft control model according to the real-time control quantity, and determining a compensation control quantity of the soft control model according to the total disturbance error; and performing model predictive control on the soft controller according to the soft control model and the compensation control quantity. The application determines the total disturbance error by obtaining the real-time control quantity, determines the compensation control quantity of the soft control model according to the total disturbance error, so as to compensate and correct the error of the control strategy in the soft robot model predictive control process, and improve the accuracy of the soft robot model predictive control.

Description

Technical Field

[0001] This invention relates to the field of soft robot control, and more specifically to a soft robot model predictive control method, device, electronic device, and computer-readable storage medium. Background Technology

[0002] Soft robots, made of flexible materials, are lightweight, small, and highly adaptable to various environments, making them suitable for applications in medical, rescue, and other fields. However, their soft structure presents challenges in precise control and control delays. Therefore, achieving accurate control of soft robots requires trajectory tracking to realize predictive control. Existing control methods have achieved predictive control of soft robots, but due to errors caused by complex environmental changes during robot movement and errors in the modeling process, existing predictive control methods for soft robots still suffer from low accuracy. Summary of the Invention

[0003] In view of this, it is necessary to provide a predictive control method, device, electronic device and computer-readable storage medium for soft robots to solve the technical problems of errors and low accuracy in the predictive control of soft robots in the prior art.

[0004] To address the above problems, this invention provides a predictive control method for soft robot models, comprising:

[0005] Obtain the historical control values ​​of the soft controller of the soft robot, and construct a soft control model of the soft controller based on the historical control values;

[0006] Obtain the real-time control quantity of the software controller, determine the total disturbance error of the software control model based on the real-time control quantity, and determine the compensation control quantity of the software control model based on the total disturbance error.

[0007] The software controller is subjected to model predictive control based on the software control model and the compensation control quantity.

[0008] Furthermore, both the historical control quantity and the real-time control quantity include input control quantity and output control quantity, wherein the input control quantity includes the duty cycle and the output control quantity includes the controller bending angle.

[0009] Furthermore, a software control model for the software controller is constructed based on historical control values, including:

[0010] Construct an initial observation function, which is in the form of an infinite-dimensional function space;

[0011] The initial observation function is converted into an observation function in finite-dimensional subspace form according to the extended mode decomposition algorithm;

[0012] Snapshot pairs are constructed based on the historical control values. The snapshot pairs are then upgraded to form matrix samples. Matrix calculations are performed on the matrix samples to obtain the approximate matrix of the Koopman operator. A software control model is then constructed based on the approximate matrix.

[0013] Further, the initial observation function is transformed into an observation function in finite-dimensional subspace form according to the extended mode decomposition algorithm, including:

[0014] Define a number of basis functions for a finite-dimensional subspace, wherein the basis functions are linearly independent of each other;

[0015] The observation function in finite-dimensional subspace form is obtained by linearly superimposing several basis functions.

[0016] Further, the real-time control input of the software controller is obtained, and the total disturbance error of the software control model is determined based on the real-time control input, including:

[0017] Set the sampling time interval and obtain the real-time control quantity of the software controller by sampling through the extended state observer;

[0018] The predicted bending angle corresponding to the duty cycle of the real-time control quantity is obtained based on the software control model.

[0019] The total disturbance error is obtained based on the real-time control quantity and the predicted bending angle.

[0020] Further, the compensation control quantity of the software control model is determined based on the total disturbance error, including:

[0021] Construct the control law function of the software control model, and determine the compensation control quantity of the software control model based on the control law function of the software control model and the total disturbance error.

[0022] Further, model predictive control is performed on the software controller based on the software control model and the compensation control quantity, including:

[0023] Establish the cost function of model predictive control, solve the cost function based on the software control model and the compensation control quantity, and determine the optimal control sequence of model predictive control in the prediction time domain;

[0024] The first element of the optimal control sequence is used as the input to the soft controller to control the movement of the soft robot.

[0025] The real-time control quantity of the software controller is obtained by sampling based on the extended state observer, and the predicted bending angle of the software controller is obtained based on the software control model.

[0026] Based on the predicted bending angle and the real-time control quantity of the software controller, the total disturbance error is updated. The optimal control sequence for the next prediction time domain is determined according to the software control model and the total disturbance error. The next prediction time domain partially overlaps with the previous prediction time domain and is later than the previous prediction time domain.

[0027] The present invention also provides a predictive control device for a soft robot model, comprising:

[0028] The model building unit acquires the historical control quantities of the soft controller of the soft robot and constructs a soft control model of the soft controller based on the historical control quantities.

[0029] The control quantity compensation unit acquires the real-time control quantity of the software controller, determines the total interference error of the software control model based on the real-time control quantity, and determines the compensation control quantity of the software control model based on the total interference error.

[0030] The dynamic control unit performs model predictive control on the software controller based on the software control model and the compensation control quantity.

[0031] The present invention also provides an electronic device, including a memory and a processor, wherein,

[0032] The memory is used to store programs;

[0033] The processor, coupled to the memory, is used to execute the program stored in the memory to implement the steps in the soft robot model predictive control method described in any of the above-described methods.

[0034] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the soft robot model predictive control method described in any one of the above claims.

[0035] Compared with existing technologies, the beneficial effects of the above embodiments are as follows: In the soft robot model predictive control method provided by the present invention, firstly, the historical control quantities of the soft robot's soft controller are obtained, and a soft control model of the soft controller is constructed based on the historical control quantities; then, the real-time control quantities of the soft controller are obtained, the total disturbance error of the soft control model is determined based on the real-time control quantities, and the compensation control quantity of the soft control model is determined based on the total disturbance error; finally, model predictive control is performed on the soft controller based on the soft control model and the compensation control quantity. In summary, the present invention achieves the technical effect of improving the accuracy of soft robot model predictive control by obtaining real-time control quantities to determine the total disturbance error and determining the compensation control quantity of the soft control model based on the total disturbance error, thereby compensating and correcting the error of the control strategy during the soft robot model predictive control process. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 A flowchart illustrating an embodiment of the predictive control method for soft robot models provided by the present invention;

[0038] Figure 2 This is a schematic flowchart of an embodiment of step S103 of the present invention;

[0039] Figure 3 A schematic diagram of the structure of an embodiment of the soft robot model prediction and control device provided by the present invention;

[0040] Figure 4 A schematic diagram of the structure of an embodiment of the electronic device provided by the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0042] It should be understood that the accompanying drawings are not drawn to scale. The flowcharts used in this invention illustrate operations implemented according to some embodiments of the invention. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or performed simultaneously. Furthermore, those skilled in the art, guided by the content of this invention, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0043] Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor systems and / or microcontroller systems.

[0044] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0045] This invention provides a predictive control method for a soft robot model, which will be described in detail below.

[0046] Figure 1 This is a schematic flowchart of an embodiment of the predictive control method for soft robots provided by the present invention, as shown below. Figure 1 As shown, the predictive control method for soft robots includes:

[0047] S101. Obtain the historical control values ​​of the soft controller of the soft robot, and construct a soft control model of the soft controller based on the historical control values.

[0048] S102. Obtain the real-time control quantity of the software controller, determine the total interference error of the software control model based on the real-time control quantity, and determine the compensation control quantity of the software control model based on the total interference error.

[0049] S103. Perform model predictive control on the software controller based on the software control model and the compensation control quantity.

[0050] Specifically, in the model predictive control method for soft robots provided by this invention, firstly, the historical control quantities of the soft robot's soft controller are obtained, and a soft control model of the soft controller is constructed based on the historical control quantities; then, the real-time control quantities of the soft controller are obtained, the total disturbance error of the soft control model is determined based on the real-time control quantities, and the compensation control quantity of the soft control model is determined based on the total disturbance error; finally, model predictive control is performed on the soft controller based on the soft control model and the compensation control quantity. This invention achieves the technical effect of compensating and correcting errors in the control strategy during the model predictive control process of a soft robot by obtaining the real-time control quantities to determine the total disturbance error and determining the compensation control quantity of the soft control model based on the total disturbance error.

[0051] In a specific embodiment of the present invention, both the historical control quantity and the real-time control quantity include an input control quantity and an output control quantity. The input control quantity includes the duty cycle, and the output control quantity includes the controller bending angle.

[0052] Specifically, the input is an analog signal with a duty cycle of 0-255. When the duty cycle is 0, the solenoid valve of the soft controller used to control the movement of the soft robot is closed, and the air pressure inside the controller is 0. When the duty cycle is 255, the air pressure inside the controller is at its maximum, and the bending angle of the controller also reaches its maximum. Historical control values ​​can be collected experimentally, while real-time control values ​​are obtained by sampling from an extended state observer.

[0053] In a specific embodiment of the present invention, the software control model of the software controller is constructed based on historical control values, including:

[0054] Construct an initial observation function, which is in the form of an infinite-dimensional function space;

[0055] The initial observation function is converted into an observation function in finite-dimensional subspace form according to the extended mode decomposition algorithm;

[0056] Snapshot pairs are constructed based on the historical control values. The snapshot pairs are then upgraded to form matrix samples. Matrix calculations are performed on the matrix samples to obtain the approximate matrix of the Koopman operator. A software control model is then constructed based on the approximate matrix.

[0057] In a specific embodiment of the present invention, converting the initial observation function into an observation function in finite-dimensional subspace form according to the extended mode decomposition algorithm includes:

[0058] Define a number of basis functions for a finite-dimensional subspace, wherein the basis functions are linearly independent of each other;

[0059] The observation function in finite-dimensional subspace form is obtained by linearly superimposing several basis functions.

[0060] Specifically, a continuous nonlinear dynamic system can be defined as follows: Where x and Describing the dynamic state, F is in the state space. The evolution function of the dynamic state with time t. The state space is constructed in matrix form, with the superscript n indicating the matrix dimension. The finite-dimensional nonlinear dynamical system is represented using the Koopman operator as follows: φ t The system state evolves over time, where g is the observable, i.e., the bending angle of the output controller and the corresponding input duty cycle; the Koopman operator K represents the evolution of the observable g over time. This represents the composite function operator, i.e., via φ. t Let the initial observation function be a function that represents the evolution of the observations in the system over time. Then, we can define the initial observation function as a function in infinite-dimensional space.

[0061] Suppose that the finite-dimensional subspace form of the observation function consists of N (N>n) linearly independent basis functions. Composition, the first n basis functions are defined as follows: Then the observation can be written as The observation function can be expressed as a linear superposition of basis functions, where η is the corresponding coefficient. By substituting and simplifying, we can obtain the optimal approximate solution representation of the Koopman operator in finite-dimensional subspace form. in This represents the Muir-Penrose pseudo-inverse symbol.

[0062] It should be noted that the Koopman operator is an infinite-dimensional linear operator that can transform a finite-dimensional nonlinear dynamic system into an infinite-dimensional linear system, effectively capturing nonlinear dynamic characteristics. In order to perform efficient calculations, the embodiment uses the optimal approximate solution of the Koopman operator in finite dimensions.

[0063] Based on historical control values, for each k∈{1,2,…,K}, the k-th historical control value is constructed as a snapshot pair sample (c[k],e[k]). Here, c[k] is the input duty cycle, and e[k] is the output driver bending angle. Let c[k] = x[k], where x[k] represents the input at the k-th measurement, then e[k] = φ Ts (x[k])+ε[k], where ε[k] represents the measurement noise, and Ts represents the sampling time, i.e., the dynamic experimental data expressed as input and output. The snapshot samples are then upgraded to a K×N matrix as follows:

[0064]

[0065] Cost function for constructing a software control model:

[0066]

[0067] Furthermore, by minimizing the solution in the closed form of the cost function, the optimal approximate solution of the Koopman operator can be further represented. in It can be obtained using the following formula:

[0068]

[0069]

[0070] In the process of finding the optimal approximate solution of the Koopman operator, u[k] is added to the snapshot pair, where u[k] represents the predicted model input control quantity, and it is constructed in the following form:

[0071]

[0072] And further, it can be written as a matrix of dimension K×(N+m):

[0073]

[0074] Where K represents the number of snapshot pairs, N represents the dimension of the processed dynamic state representation, and m represents the number of model inputs.

[0075] By further simplifying and solving the matrix, the Koopman operator approximation matrix can be obtained. Represented as:

[0076]

[0077] in, and By decomposing the matrix Obtain, and

[0078] A soft control model can be constructed using the Koopman approximation matrix, which can then be used for model prediction and design of soft controllers.

[0079] In a specific embodiment of the present invention, obtaining the real-time control quantity of the software controller and determining the total disturbance error of the software control model based on the real-time control quantity includes:

[0080] Set the sampling time interval and obtain the real-time control quantity of the software controller by sampling through the extended state observer;

[0081] The predicted bending angle corresponding to the duty cycle of the real-time control quantity is obtained based on the software control model.

[0082] The total disturbance error is obtained based on the real-time control quantity and the predicted bending angle.

[0083] Specifically, to compensate for modeling errors and errors caused by external uncertainties in the software controller, this embodiment of the invention uses an extended state observer to sample the real-time control quantities during the operation of the software controller according to a set sampling time interval. Based on the sampled real-time control quantities and the predicted bending angle obtained through the software control model, the total disturbance error of the software control model is calculated by comparing the bending angle in the real-time control quantities with the predicted bending angle.

[0084] In a specific embodiment of the present invention, determining the compensation control quantity of the software control model based on the total disturbance error includes:

[0085] Construct the control law function of the software control model, and determine the compensation control quantity of the software control model based on the control law function of the software control model and the total disturbance error.

[0086] Specifically, defining the total disturbance at time kth as d(k), the state equation of the soft control model can be redefined as:

[0087]

[0088] Where, C = [I n×n O n×(N-n) ], O represents the zero matrix, I represents the identity matrix, and the subscripts represent the dimensions of the matrix. x[k] is the state representation of the dynamic system, and u[k] is the predicted input control quantity.

[0089] The control law of the software control model can be expressed as u[k] = u MPC [k]+u d [k], where u MPC [k] is the input control quantity calculated by the software controller, u d [k] is the compensation control quantity for the total disturbance, and the value of the compensation control quantity for the total disturbance is calculated based on the state equation of the software control model.

[0090] In a specific embodiment of the present invention Figure 2 This is a schematic diagram of the model predictive control process according to an embodiment of the present invention, such as... Figure 2 As shown, model predictive control of the software controller is performed based on the software control model and the compensation control quantity, including:

[0091] S201. Establish the cost function of model predictive control, solve the cost function based on the software control model and the compensation control quantity, and determine the optimal control sequence of model predictive control in the prediction time domain.

[0092] S202. Use the first element of the optimal control sequence as the input of the soft controller to control the movement of the soft robot;

[0093] S203. The real-time control quantity of the software controller is obtained based on the sampling of the extended state observer, and the predicted bending angle of the software controller is obtained based on the software control model.

[0094] S204. Based on the predicted bending angle and real-time control quantity of the software controller, update the total disturbance error. Determine the optimal control sequence for the next prediction time domain according to the software control model and the total disturbance error. The next prediction time domain partially overlaps with the previous prediction time domain, and the next prediction time domain is later than the previous prediction time domain.

[0095] Specifically, based on the software control model and the compensation control quantity, the control law function for controlling the input of the software controller is obtained as follows:

[0096] The cost function of model predictive control is established based on the control law function and state equation of the software control model. Furthermore, for optimization purposes, the cost function of model predictive control is rewritten as follows:

[0097]

[0098] st UV+Ms0≤c

[0099]

[0100] Where P, p, L, V, M, and c are intermediate matrices defined for simplified matrix operations, and It is a positive semi-definite matrix. This represents the optimal control sequence obtained from the model predictive control solution.

[0101] After solving the cost function of model predictive control to obtain the optimal control sequence, the first element in the optimal control sequence is used as the input of the soft controller to control the movement of the soft robot.

[0102] Simultaneously, the extended state observer detects the real-time control quantity of the sampled software controller and uses the software control model to predict the predicted bending angle of the software controller.

[0103] The predicted bending angle is compared with the acquired real-time control quantity, the total disturbance error is updated in real time, and the optimal control sequence is updated in real time based on the updated total disturbance error to ensure the dynamic balance of the number of elements in the optimal control sequence. This allows for accurate prediction and control of the soft robot's motion trajectory and real-time compensation and correction of any possible errors.

[0104] To better implement the soft robot model predictive control method in the embodiments of the present invention, the present invention also provides a soft robot model predictive control device 300, based on the soft robot model predictive control method, such as... Figure 3 As shown, it includes:

[0105] Model building unit 301 acquires the historical control quantities of the soft controller of the soft robot and constructs a soft control model of the soft controller based on the historical control quantities.

[0106] The control quantity compensation unit 302 acquires the real-time control quantity of the software controller, determines the total interference error of the software control model based on the real-time control quantity, and determines the compensation control quantity of the software control model based on the total interference error.

[0107] The dynamic control unit 303 performs model predictive control on the software controller based on the software control model and the compensation control quantity.

[0108] The soft robot model prediction control device 300 provided in the above embodiments can realize the technical solutions described in the above embodiments of the soft robot model prediction control method. The specific implementation principles of each module or unit can be found in the corresponding content in the above embodiments of the soft robot model prediction control method, which will not be repeated here.

[0109] Based on the predictive control method of soft robot models, this invention also provides an electronic device, such as... Figure 4 As shown, Figure 4 This is a schematic diagram of an embodiment of the electronic device provided by the present invention. The electronic device 400 includes a processor 401, a memory 402, and a computer program stored in the memory 402 and executable on the processor 401. When the processor 401 executes the program, it implements the soft robot model predictive control method as described above.

[0110] In a preferred embodiment, the electronic device further includes a display 403 for displaying the process of the processor 401 executing the soft robot model predictive control method as described above.

[0111] The processor 401 may be an integrated circuit chip with signal processing capabilities. The processor 401 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP) or an application-specific integrated circuit (ASIC). It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can also be a microprocessor or any conventional processor.

[0112] The memory 402 may be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Secure Digital (SD card), Flash Card, etc. The memory 402 stores programs, and the processor 401 executes these programs upon receiving execution instructions. The process definition methods disclosed in any of the foregoing embodiments of the present invention can be applied to the processor 401, or implemented by the processor 401.

[0113] The display 403 can be an LED display, an LCD display, or a touch screen display, etc. The display 403 is used to display various information from the electronic device 400.

[0114] Understandable Figure 4 The structure shown is only a schematic diagram of one possible structure of electronic device 400. Electronic device 400 may also include more than one of the following: Figure 4 Show more or fewer components. Figure 4 The components shown can be implemented using hardware, software, or a combination thereof.

[0115] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the soft robot model predictive control method as described above.

[0116] Generally, computer instructions for implementing the methods of the present invention can be carried on any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media can include any computer-readable medium except for signals themselves that are temporarily propagating.

[0117] Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0118] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A predictive control method for a soft robot model, characterized in that, include: The process involves obtaining historical control variables of the soft robot's soft controller, including both historical and real-time control variables, such as input and output control variables. The input control variables include the duty cycle, and the output control variables include the controller's bending angle. An initial observation function is constructed, which is in an infinite-dimensional function space. This initial observation function is then converted into a finite-dimensional subspace observation function using an extended mode decomposition algorithm. Snapshot pairs are constructed based on the historical control variables. These snapshot pairs are then upgraded to matrix samples. Matrix calculations are performed on these matrix samples to obtain an approximate matrix of the Koopman operator. Finally, a soft control model is constructed based on this approximate matrix. The sampling time interval is set, and the real-time control quantity of the software controller is obtained by sampling through the extended state observer; the predicted bending angle corresponding to the duty cycle of the real-time control quantity is obtained according to the software control model; the total disturbance error is obtained according to the real-time control quantity and the predicted bending angle; the control law function of the software control model is constructed, and the compensation control quantity of the software control model is determined based on the control law function of the software control model and the total disturbance error. Establish the cost function of model predictive control, solve the cost function based on the soft control model and the compensation control quantity, and determine the optimal control sequence of model predictive control in the prediction time domain; use the first element of the optimal control sequence as the input of the soft controller to control the movement of the soft robot. The real-time control quantity of the software controller is obtained based on the sampling of the extended state observer, and the predicted bending angle of the software controller is obtained based on the software control model. The total disturbance error is updated based on the predicted bending angle of the software controller and the real-time control quantity. The optimal control sequence for the next prediction time domain is determined according to the software control model and the total disturbance error. The next prediction time domain partially overlaps with the previous prediction time domain and is later than the previous prediction time domain.

2. The predictive control method for soft robot models according to claim 1, characterized in that, The step of converting the initial observation function into an observation function in finite-dimensional subspace form according to the extended mode decomposition algorithm includes: Define a number of basis functions for a finite-dimensional subspace, wherein the basis functions are linearly independent of each other; The observation function in finite-dimensional subspace form is obtained by linearly superimposing several basis functions.

3. A predictive control device for a soft robot model, characterized in that, include: The model building unit is used to acquire the historical control quantities of the soft controller of the soft robot. Both the historical control quantities and the real-time control quantities include input control quantities and output control quantities. The input control quantities include the duty cycle, and the output control quantities include the controller bending angle. An initial observation function is constructed, which is in an infinite-dimensional function space form. The initial observation function is then converted into an observation function in a finite-dimensional subspace form using an extended mode decomposition algorithm. Snapshot pairs are constructed based on the historical control quantities. These snapshot pairs are then upgraded to matrix samples. Based on these matrix samples, matrix calculations are performed to obtain an approximate matrix of the Koopman operator. Finally, a soft control model is constructed based on this approximate matrix. The control quantity compensation unit is used to set the sampling time interval, obtain the real-time control quantity of the software controller by sampling through the extended state observer; obtain the predicted bending angle corresponding to the duty cycle of the real-time control quantity according to the software control model; obtain the total disturbance error according to the real-time control quantity and the predicted bending angle; construct the control law function of the software control model; and determine the compensation control quantity of the software control model based on the control law function of the software control model and the total disturbance error. The dynamic control unit is used to establish the cost function of model predictive control, solve the cost function based on the soft control model and the compensation control quantity, and determine the optimal control sequence of model predictive control in the prediction time domain; the first element of the optimal control sequence is used as the input of the soft controller to control the motion of the soft robot. The real-time control quantity of the software controller is obtained based on the sampling of the extended state observer, and the predicted bending angle of the software controller is obtained based on the software control model. The total disturbance error is updated based on the predicted bending angle of the software controller and the real-time control quantity. The optimal control sequence for the next prediction time domain is determined according to the software control model and the total disturbance error. The next prediction time domain partially overlaps with the previous prediction time domain and is later than the previous prediction time domain.

4. An electronic device, characterized in that, Including memory and processor, among which, The memory is used to store programs; The processor, coupled to the memory, is used to execute the program stored in the memory to implement the steps in the soft robot model predictive control method according to any one of claims 1 to 2.

5. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by a processor, implements the soft robot model predictive control method as described in any one of claims 1-2.

Images