Method and system for controlling non-contact operation tasks of a micro robot

Abstract

The application provides a control method and system for non-contact operation tasks of a magnetic micro robot, and the method comprises the following steps: acquiring a current position state, wherein the current position state comprises a current position of a target object and a relative distance between the magnetic micro robot and the target object; a motion model performs position state prediction according to the current position state to obtain a predicted position state at a next moment; a magnetic field control parameter is generated by taking a safe distance as a constraint and minimizing an error between an ideal position state and the predicted position state as an objective, wherein the safe distance refers to a target distance between the magnetic micro robot and the target object, and the magnetic field control parameter comprises a magnetic field direction and a magnetic field frequency; and the magnetic field is controlled according to the magnetic field control parameter to drive the magnetic micro robot to push the target object to move along a planned path. In this way, an accurate and flexible control method for non-contact operation tasks of the magnetic micro robot is realized.

Description

Technical Field

[0001] This application relates to the field of microrobot control technology, and in particular to a control method and system for non-contact operation tasks of a magnetically controlled microrobot. Background Technology

[0002] Magnetically controlled microrobots (magnetically driven microrobots) can navigate using external magnetic fields and move autonomously in complex and unstructured environments, showing broad application prospects in fields such as drug delivery, diagnosis, and therapeutic intervention.

[0003] To enhance the autonomy of magnetically controlled microrobots, the positioning, navigation, and control of magnetic particles can be achieved through direct manipulation, contact manipulation, and non-contact manipulation methods. Direct manipulation drives the object by imparting magnetism, but this can damage its inherent properties and biological activity. Contact manipulation involves the robot directly contacting the target object, but introducing physical contact can lead to problems such as object deformation, solid contamination, and difficulties in separation. Non-contact manipulation indirectly moves the target object by generating eddies or laminar flows through the robot. While non-contact manipulation overcomes the limitations of the above two methods, existing methods suffer from drawbacks such as limited range of motion, difficulty in precise control, and poor flexibility. Summary of the Invention

[0004] In view of the above problems, embodiments of this application provide a control method and system for non-contact operation tasks of a magnetically controlled microrobot, so as to overcome the above problems or at least partially solve the above problems.

[0005] A first aspect of this application discloses a control method for a non-contact operation task of a magnetically controlled microrobot, the method comprising:

[0006] Obtain the current position status, which includes: the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object;

[0007] The motion model predicts the position state based on the current position state to obtain the predicted position state at the next moment. The motion model is constructed using a neural network model.

[0008] With the constraint of maintaining a safe distance and the objective of minimizing the error between the ideal position state and the predicted position state, magnetic field control parameters are generated. The safe distance refers to the target distance for maintaining non-operational contact operations. The magnetic field control parameters include: magnetic field direction and magnetic field frequency.

[0009] The magnetic field is controlled according to the magnetic field control parameters to drive the magnetically controlled microrobot to push the target object along the planned path.

[0010] Optionally, the method further includes:

[0011] Calculate the state error value between the current position state and the corresponding predicted position state;

[0012] The model parameters of the motion model are updated online based on the state error value to obtain the updated motion model;

[0013] The motion model predicts the position state based on the current position state to obtain the predicted position state at the next moment, including:

[0014] The updated motion model predicts the position state based on the current position state to obtain the predicted position state at the next moment.

[0015] Optionally, the motion model is constructed using a neural network model according to the following steps:

[0016] The position state definition and control input definition in the global coordinate system are determined. The position state definition includes: the position of the target object and the relative distance between the magnetically controlled microrobot and the target object. The control input definition includes: the projected angular velocity on the first coordinate axis and the projected angular velocity on the second coordinate axis of the global coordinate system. The magnetic field rotates in the plane formed by the first coordinate axis and the second coordinate axis.

[0017] Based on the defined position state and the defined control input, an initial motion model is constructed;

[0018] The motion model parameters are determined based on the neural network model, and the initial motion model is transformed into a local coordinate system to obtain the motion model. The local coordinate system takes the line connecting the center point of the magnetically controlled microrobot and the center point of the target object as the normal direction.

[0019] Optionally, the neural network model includes a parameter determination network; determining motion model parameters based on the neural network model, and transforming the initial motion model to a local coordinate system to obtain the motion model, includes:

[0020] The average radius of the magnetically controlled microrobot and the target object is normalized to the relative distance to obtain the normalized relative Euclidean distance.

[0021] A parameter determination network is constructed, which is trained using offline collected location state data and magnetic field parameter values ​​as training samples.

[0022] The network is determined based on the normalized relative Euclidean distance and the parameters, and the diagonal matrix generated by directional decoupling is obtained.

[0023] The motion model is obtained based on the rotation matrix from the global coordinate system to the local coordinate system, the diagonal matrix generated by the direction decoupling, and the control input definition.

[0024] Optionally, the model parameters of the motion model are updated online based on the state error value to obtain an updated motion model, including:

[0025] The weight values ​​of the parameter determination network are updated based on the state error value to obtain the updated parameter determination network;

[0026] The network is determined based on the normalized relative Euclidean distance and the updated parameters, and the updated diagonal matrix generated by directional decoupling is obtained.

[0027] Based on the diagonal matrix generated by the updated directional decoupling, the model parameters of the motion model are updated online to obtain the updated motion model.

[0028] Optionally, with the constraint of maintaining a safe distance and the objective of minimizing the error between the ideal position state and the predicted position state, magnetic field control parameters are generated, including:

[0029] Based on the ideal position state and the predicted position state, a cost function is constructed, the cost function having the optimization objective of minimizing the error between the ideal position state and the predicted position state;

[0030] Determine the first constraint condition, which is used to constrain the maximum value of the control input;

[0031] A second constraint condition is determined, which is used to constrain the minimum relative distance between the magnetically controlled microrobot and the target object;

[0032] A third constraint condition is determined, which is used to constrain the maximum relative distance between the magnetically controlled microrobot and the target object;

[0033] The cost function is solved using the first constraint, the third constraint, and the third constraint as constraints to obtain the magnetic field control parameters.

[0034] Optionally, the planned path is obtained through the following steps:

[0035] Obtain the starting position and ending position of the target object;

[0036] The nearest point is determined by using a turning angle distance metric. The starting position is used as the root node, and the path is iteratively generated towards the ending position by randomly sampling and adding leaf nodes to obtain a path expansion tree. The turning angle distance metric is determined based on the Euclidean distance between the two points and the turning angle.

[0037] The initial planned path from the starting position to the ending position is extracted from the path expansion tree, and the initial planned path is smoothed to obtain the planned path.

[0038] A second aspect of this application discloses a control system for a non-contact operation task of a magnetically controlled microrobot, used to execute the steps of the control method for a non-contact operation task of a magnetically controlled microrobot described in the first aspect of this application. The system includes:

[0039] Image acquisition equipment used to acquire microscopic images of magnetically controlled microrobots and target objects;

[0040] A computing processing device is used to identify the current position state based on the microscopic image and output magnetic field control parameters based on the current position state. The current position state includes: the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object. The magnetic field control parameters include: magnetic field direction and magnetic field frequency.

[0041] An electromagnetic device is used to generate a magnetic field according to magnetic field control parameters to drive the magnetically controlled microrobot to push the target object along a planned path.

[0042] A third aspect of this application discloses a control device for non-contact operation tasks of a magnetically controlled microrobot, the device comprising:

[0043] The acquisition module is used to acquire the current position status, which includes: the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object;

[0044] The prediction module is used by the motion model to predict the position state of the next moment based on the current position state. The motion model is constructed by a neural network model.

[0045] The generation module is used to generate magnetic field control parameters with the constraint of maintaining a safe distance and the goal of minimizing the error between the ideal position state and the predicted position state. The safe distance refers to the target distance for maintaining non-operational contact operations. The magnetic field control parameters include: magnetic field direction and magnetic field frequency.

[0046] The control module is used to control the magnetic field according to the magnetic field control parameters, so as to drive the magnetically controlled microrobot to push the target object to move along the planned path.

[0047] A fourth aspect of this application discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the control method for non-contact operation tasks of the magnetically controlled microrobot described in the first aspect of this application.

[0048] The embodiments of this application have the following advantages:

[0049] In this embodiment, the current position state is obtained, which includes the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object. A motion model predicts the position state at the next moment based on the current position state, and the motion model is constructed using a neural network model. Magnetic field control parameters are generated with the constraint of maintaining a safe distance and the objective of minimizing the error between the ideal position state and the predicted position state. The safe distance refers to the target distance between the magnetically controlled microrobot and the target object. The magnetic field control parameters include the magnetic field direction and magnetic field frequency. The magnetic field is controlled according to the magnetic field control parameters to drive the magnetically controlled microrobot to propel the target object along a planned path.

[0050] By constructing a motion model using neural networks, the problem of difficulty in obtaining motion models due to the uncertainty of dynamic parameters is solved. Based on this motion model, the predicted position state at the next moment can be obtained from the current position state, realizing real-time closed-loop control of the target object. This allows the magnetically controlled microrobot to achieve precise operation and tracking in different environments by adjusting the magnetic field control parameters in real time. Furthermore, the magnetic field control parameters are generated with the constraint of maintaining a safe distance and the goal of minimizing the error between the target position state and the predicted position state, achieving non-contact operation of the target object and avoiding damage caused by direct contact. Based on these magnetic field control parameters, precise actuation of the target object can be achieved. In addition, by using the magnetic field frequency and direction as magnetic field control parameters, the micro-magnetically controlled robot can move in any direction by changing the magnetic field direction and frequency, thereby expanding the movement range of the micro-magnetically controlled robot and improving control flexibility. Thus, precise and flexible control of the magnetically controlled microrobot for non-contact operation tasks is achieved. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a schematic diagram of a non-contact operation task of a magnetically controlled microrobot provided in an embodiment of this application;

[0053] Figure 2 This is a flowchart illustrating the steps of a control method for a non-contact operation task of a magnetically controlled microrobot provided in an embodiment of this application.

[0054] Figure 3 This is a schematic diagram illustrating the motion of a magnetically controlled microrobot pushing a target object without contact, as provided in an embodiment of this application.

[0055] Figure 4 This is a flowchart of the steps of a control method for a non-contact operation task of a magnetically controlled microrobot provided in an embodiment of this application;

[0056] Figure 5 This is a performance comparison analysis result of different motion models provided in the embodiments of this application;

[0057] Figure 6 This application provides a comparison of tracking results for different planned trajectories in its embodiments.

[0058] Figure 7 This application provides a comparison of tracking results for controllers based on different motion models.

[0059] Figure 8 This is a schematic diagram of the structure of a control system for a non-contact operation task of a magnetically controlled microrobot provided in an embodiment of this application;

[0060] Figure 9 This is a schematic diagram of the structure of a control device for a non-contact operation task of a magnetically controlled microrobot provided in an embodiment of this application;

[0061] Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0062] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0063] In related technologies, non-contact operation methods are divided into two categories: one is to use gradient magnetic fields to translate and propel a robot, thereby generating laminar flow along the robot's boundary, and using this laminar flow to propel the target object. Specifically, in one method of translating and propelling a robot, the microrobot uses the fluid velocity field generated by the fluid boundary layer to induce non-contact propulsion of the target object; in another method, the microrobot directly contacts the microobject in the forward thrust direction, and non-contact displacement is achieved in the lateral thrust direction using the fluid flow field generated by the translating microrobot. However, the effective range of the translation mode is limited, making long-distance transportation of objects difficult to achieve.

[0064] Another approach utilizes eddy currents generated when a robot rotates or rolls in a uniform magnetic field to propel a target object. Specifically, in one method for rolling microrobots, eddy currents above the microrobot's body are used to lift the object; in another, movable micro-eddy currents are used to control individual micro- and nano-sized objects, allowing for capture and manipulation in fluid environments without significant limitations imposed by material properties; and in a magnetically driven peanut-shaped hematite colloidal motor, both rolling and swaying motions in fluids are possible, enabling non-contact transport of single cells. While these eddy current non-contact modes offer greater flexibility, achieving closed-loop control using eddy currents is challenging, leading to reduced control accuracy.

[0065] To address the limitations of related technologies, this application provides a control method for non-contact operation tasks of a magnetically controlled microrobot. (Refer to...) Figure 1 As shown, Figure 1 This is a schematic diagram of a non-contact operation task of a magnetically controlled microrobot provided in an embodiment of this application. Specifically, it utilizes a flexible non-contact operation scheme based on rotation. By controlling the direction and frequency of the magnetic field, the magnetically controlled microrobot rotates under the influence of the magnetic field, generating a repulsive field and an eddy current field, thereby inducing fluid motion to drive the target object. Compared with existing non-contact methods, this method utilizing local pressure is more stable and easier to control. Therefore, this push-based non-contact operation method is proposed to achieve precise closed-loop control.

[0066] Considering that motion model analysis from magnetic field input to target velocity output in such a non-contact operation mode is a complex fluid-solid coupling problem involving bidirectional influences between the magnetically controlled microrobot, the target object, and the surrounding fluid, and that the uncertainty of dynamic parameters increases the challenge of obtaining the motion model, this application's embodiments are based on data-driven methods. A neural network is used to effectively estimate the motion model, which can predict the next position state based on the current position state. Under the constraint of maintaining a safe distance, the goal is to minimize the error between the ideal and predicted position states to obtain magnetic field control parameters. These parameters are then used to control the magnetic field, driving the magnetically controlled microrobot to propel the target object along a planned path. By adjusting the magnetic field control parameters in real time, the magnetically controlled microrobot can achieve precise manipulation and tracking in different environments, realizing precise and flexible control of the non-contact operation task of the magnetically controlled microrobot. The control method for the non-contact operation task of the magnetically controlled microrobot of this application will be described in detail below.

[0067] Reference Figure 2 As shown, Figure 2 This is a flowchart illustrating the steps of a control method for a non-contact operation task of a magnetically controlled microrobot provided in an embodiment of this application. Figure 2 As shown, the control method for non-contact operation tasks of a magnetically controlled microrobot provided in this application embodiment may include steps S210 to S240:

[0068] Step S210: Obtain the current position status, which includes: the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object.

[0069] In this embodiment, to achieve non-contact operation of the magnetically controlled microrobot, only the position control of the target object needs to be considered, while maintaining a non-contact state between the magnetically controlled microrobot and the target object. Therefore, the position of the target object and the relative distance between the magnetically controlled microrobot and the target object are used as the position state. Obtaining the current position state specifically includes: acquiring a microscopic image of the current magnetically controlled microrobot and the target object using an image acquisition device, and using a visual algorithm to recognize the microscopic image to obtain the previous position state.

[0070] Step S220: The motion model predicts the position state based on the current position state to obtain the predicted position state at the next moment. The motion model is constructed using a neural network model.

[0071] In this embodiment of the application, in order to perform real-time closed-loop control of the target object and use the motion model to predict the predicted position state at the next moment, the predicted position state at the next moment includes: the position of the target object at the next moment, and the relative distance between the magnetically controlled microrobot and the target object at the next moment.

[0072] The phrase "motion model constructed through neural network model" refers to the fact that the model parameters of the motion model are determined by a neural network model. A neural network model is an upgraded network capable of quickly predicting unknown nonlinear functions, such as a radial basis function network (RBFN network). Constructing a motion model through a neural network model solves the problem of obtaining a motion model due to the uncertainty of dynamic parameters, as well as the problem of obtaining a motion model in complex fluid-solid coupling problems.

[0073] In an optional embodiment, the motion model is constructed using a neural network model according to steps A1 to A3:

[0074] Step A1: Determine the position state definition and control input definition in the global coordinate system. The position state definition includes: the position of the target object and the relative distance between the magnetically controlled microrobot and the target object. The control input definition includes: the projected angular velocity on the first coordinate axis of the global coordinate system and the projected angular velocity on the second coordinate axis of the global coordinate system. The magnetic field rotates in the plane formed by the first coordinate axis and the second coordinate axis.

[0075] For example, Figure 3 This is a schematic diagram of a magnetically controlled microrobot non-contactly pushing a target object, provided in an embodiment of this application. The global coordinate system refers to a three-dimensional spatial coordinate system with the ground as a reference, including a first coordinate axis (x-axis), a second coordinate axis (y-axis), and a third coordinate axis (z-axis). The magnetic field rotates within the plane formed by the first and second coordinate axes, that is, the magnetic field rotates around an axis in the xy-plane. The angle between the projection of the magnetic field onto the xy-plane and the x-axis is denoted as α. When the direction of motion of the magnetically controlled microrobot is different from the straight line connecting it to the center of the target object, the tangential fluid force F... d It will affect the target object, causing it to bend.

[0076] Therefore, this application introduces a controlled affine model to describe planar motion, constructing the motion model through a data-driven approach. The positions of the magnetically controlled microrobot and the target object are represented as x, respectively. a ∈R 2 and x u ∈R 2Since only the control of the target object is required, while maintaining a non-contact state between the magnetically controlled microrobot and the target object, the position of the target object and the relative distance between the magnetically controlled microrobot and the target object are defined as the position state, expressed as x = [x u ,x r ]∈R 4 The relative distance x r =x u -x a The control input is defined as u = [r b cosa,r b sina]∈R 2 r b cosa represents the projected angular velocity along the x-axis, r b sina represents the projected angular velocity on the y-axis, r b It represents the frequency of the magnetic field.

[0077] Step A2: Construct an initial motion model based on the defined position state and the defined control input.

[0078] In the embodiments of this application, the initial motion model is constructed under the following properties: firstly, the system is in a quasi-static state; at a low Reynolds number, viscous drag plays a dominant role, so the motion model follows a first-order model; and secondly, the motion is entirely induced by the magnetic field, making the drift vector field of the motion model zero; and thirdly, both the liquid and the magnetic field are uniformly distributed, so the control vector field is determined solely by the relative distance.

[0079] Specifically, constructing the initial motion model based on the position state definition and the control input definition means multiplying the control vector field by the derivative of the control input definition, which equals the position state definition. For example, the initial motion model can be represented as:

[0080]

[0081] in, g represents velocity, and g(·) represents the control vector field.

[0082] Step A3: Determine the motion model parameters based on the neural network model, transform the initial motion model to a local coordinate system to obtain the motion model. The local coordinate system uses the line connecting the center point of the magnetically controlled microrobot and the center point of the target object as the normal direction.

[0083] In this embodiment, an offline data-driven strategy is employed to construct the motion model, overcoming the analysis challenges posed by fluid-structure interaction. To enhance the simplicity of the motion model and reduce data sampling requirements, a local coordinate system is introduced. The line connecting the center point of the magnetically controlled microrobot and the center point of the target object is used as the normal direction reference. The control equations depend entirely on the relative distance, and the motion along the tangential and normal directions becomes decoupled. Furthermore, to further improve the generalization of the motion model and enable it to handle magnetically controlled microrobots and target objects of arbitrary scale, normalization processing is performed on the magnetically controlled microrobots and target objects.

[0084] Specifically, the neural network model includes a parameter determination network; determining motion model parameters based on the neural network model, and transforming the initial motion model to a local coordinate system to obtain the motion model includes: normalizing the relative distance between the average radius of the magnetically controlled microrobot and the target object to obtain a normalized relative Euclidean distance; constructing a parameter determination network, which is trained using offline collected position state data and magnetic field parameter values ​​as training samples; obtaining a diagonal matrix generated by directional decoupling based on the normalized relative Euclidean distance and the parameter determination network; and obtaining the motion model based on the rotation matrix from the global coordinate system to the local coordinate system, the diagonal matrix generated by directional decoupling, and the control input definition.

[0085] For example, the motion model is represented as:

[0086]

[0087] in, Indicates the velocity of the target object; Represents relative velocity; R∈SO(2) is the rotation matrix from global coordinates to local coordinates; s r ∈R is the normalized relative Euclidean distance, N u N r :R→R 2,2 This represents the diagonal matrix generated by directional decoupling. The diagonal matrix generated by directional decoupling is a nonlinear function and can be determined by a parameter-determining network.

[0088] Parameter-determined networks are networks that possess rapid learning capabilities and can be updated online. Parameter-determined networks (N) k (·) can be specifically represented as:

[0089]

[0090] in, Two parameters determine the activation function vector of the network, i.e., the parameter-determined network N. u (·) and parameter determination network N rThe activation function vector of (·); W k The parameters determine the weights of the network; each parameter determines a set of weights, i.e., W. u and W r .

[0091] Furthermore, the parameter determination network uses a multi-quadric radial function as the activation function, which is expressed as:

[0092]

[0093] Where, μ i μ represents the center and width of the i-th neuron. i It is trained offline and fixed online. This activation function maintains the smoothness of the input parameters, avoids abrupt changes, and provides robustness in noisy data scenarios. The training samples for the parameter determination network are collected offline by manually manipulating the direction and frequency of the magnetic field in a real environment. During training, offline training is performed using mean squared error (MSE) loss and the Adam optimizer. The parameter determination network has a simple structure and achieves excellent fitting performance with minimal data requirements, thus facilitating manual data collection.

[0094] Step S230: With maintaining a safe distance as a constraint and minimizing the error between the ideal position state and the predicted position state as the objective, generate magnetic field control parameters. The safe distance refers to the target distance for maintaining non-operational contact operations. The magnetic field control parameters include: magnetic field direction and magnetic field frequency.

[0095] In this embodiment, the ideal position state refers to the position state corresponding to the planned path. In order to avoid the magnetically controlled microrobot from contacting the target object and to push the target object forward along the planned path, the calculation of the magnetic field control parameters is designed as a convex optimization problem to minimize the error between the ideal position state and the predicted position state, so that the target object is as close as possible to the desired position.

[0096] By introducing a safety distance constraint, the relative distance between the magnetically controlled microrobot and the target object is maintained within a safe range during the movement of the target object, enabling non-contact operation. The safety distance refers to the target distance for maintaining non-contact operation, and is the distance between a minimum and a maximum safety distance. When the relative distance between the magnetically controlled microrobot and the target object is less than or equal to the minimum safety distance, the projection of the relative velocity onto the relative distance becomes negative, resulting in an expansion of the relative distance. Conversely, when the relative distance is greater than or equal to the maximum safety distance, the projection of the relative velocity onto the relative distance becomes positive, resulting in a decrease in the relative distance.

[0097] In an optional embodiment, with maintaining a safe distance as a constraint and minimizing the error between the ideal position state and the predicted position state as the objective, magnetic field control parameters are generated, including steps B1 to B5:

[0098] Step B1: Based on the ideal position state and the predicted position state, construct a cost function, the cost function having the optimization objective of minimizing the error between the ideal position state and the predicted position state.

[0099] In this embodiment of the application, the cost function is designed to make the target object as close as possible to the ideal position. For example, the cost function is expressed as:

[0100]

[0101] Wherein, λ represents the adjustable parameter. Adjusting the value of the adjustable parameter helps to track the trajectory with better curvature. The value of the adjustable parameter is determined according to the actual control accuracy. This represents the velocity of an ideal target object; Indicates the predicted velocity of the target object; This represents the ideal relative velocity; This indicates the predicted relative speed.

[0102] Step B2: Determine the first constraint condition, which is used to constrain the maximum value of the control input.

[0103] In other words, the maximum control input value under different control input operations, expressed as the first constraint, is:

[0104] ||u||2≤u max (6)

[0105] Among them, u max This indicates the maximum control input.

[0106] Step B3: Determine the second constraint condition, which is used to constrain the minimum relative distance between the magnetically controlled microrobot and the target object.

[0107] Step B4: Determine the third constraint condition, which is used to constrain the maximum relative distance between the magnetically controlled microrobot and the target object.

[0108] In this embodiment, considering that contact between the micro-magnetically controlled robot and the target object can cause solid contamination and that separation is difficult due to adhesive forces, a safety distance constraint (i.e., the second and third constraint conditions) is introduced to ensure that the micro-magnetically controlled robot and the target object are always in a non-contact state.

[0109] For example, the second constraint is expressed as:

[0110]

[0111] in, This represents the minimum relative distance (i.e., the minimum safe distance). When the relative distance between the magnetically controlled microrobot and the target object is less than or equal to the minimum relative distance, the projection of the relative velocity onto the relative distance becomes negative, resulting in an expansion of the relative distance.

[0112] The third constraint is expressed as follows:

[0113]

[0114] in, This represents the maximum relative distance (i.e., the maximum safe distance). When the relative distance between the magnetically controlled microrobot and the target object is greater than or equal to the minimum relative distance, the projection of the relative velocity onto the relative distance becomes positive, thus causing the relative distance to decrease.

[0115] Step B5: Solve the cost function using the first constraint, the third constraint, and the third constraint as constraints to obtain the magnetic field control parameters.

[0116] In this embodiment, the solution of the magnetic field control parameters is transformed into a convex optimization problem. Under the first constraint, the solved magnetic field control parameters are within the maximum allowable range. Under the second and third constraints, the solved magnetic field control parameters ensure that the micro magnetic robot and the target object are in a non-contact state, thereby achieving precise control under non-contact operation.

[0117] Step S240: Control the magnetic field according to the magnetic field control parameters to drive the magnetically controlled microrobot to push the target object along the planned path.

[0118] In this embodiment, the magnetic field control parameters are obtained based on the current position state. Therefore, the magnetic field control parameters are adjusted in real time. By controlling the magnetic field using these parameters, real-time closed-loop control of the target object is achieved, enabling the target object to move precisely along the planned path. Furthermore, by using the magnetic field frequency and direction as magnetic field control parameters, the micro-magnetically controlled robot can move in any direction by changing the magnetic field direction and frequency, thereby expanding the movement range of the micro-magnetically controlled robot and improving control flexibility.

[0119] The planned path is a path that is planned in real time based on the actual environmental conditions. Based on the real-time planned path, the magnetically controlled microrobot can achieve precise navigation in different environments and move the target object to a specific location.

[0120] In an optional embodiment, the planned path is obtained through the following steps C1 to C3:

[0121] Step C1: Obtain the starting position and ending position of the target object.

[0122] Step C2: The nearest point is determined using the turning angle distance metric. The starting position is used as the root node, and the path is iteratively generated towards the ending position by randomly sampling and adding leaf nodes to obtain the path expansion tree. The turning angle distance metric is determined based on the Euclidean distance between the two points and the turning angle.

[0123] Step C3: Extract the initial planned path from the starting position to the ending position from the path expansion tree, and smooth the initial planned path to obtain the planned path.

[0124] In this embodiment, in a cluttered micro-environment, path planning is performed based on the starting and ending positions to guide the target object from the starting position to the ending position. To make the planned trajectory smoother, a turning angle is introduced into the distance measurement during distance calculation. The turning angle between two points is expressed as:

[0125]

[0126] Where, x c x represents the position of the current node. p This represents the position of the parent node of the current node.

[0127] The turning angle distance measure is determined based on the Euclidean distance between two points and the turning angle. The turning angle distance measure is expressed as:

[0128]

[0129] Where, p r p represents a node that is randomly sampled; t Represents a node in the path expansion tree; R r It is the turning angle θ r The rotation matrix R; t It is the turning angle θ t The rotation matrix.

[0130] Specifically, the process of obtaining the path expansion tree in step C2 is as follows: take the starting position as the root node, generate random sampling points, calculate the turning angle distance measure between the random sampling points and the nodes in the path expansion tree according to formula (10), find the nearest node from the path expansion tree according to the calculated turning angle distance measure, and grow from the nearest node to the random sampling point according to the generation step size, so that the path expansion tree obtains a new node; if the path expansion tree has not yet reached the end position at this time, start the next iteration growth according to the same logic until the path expansion tree reaches the end position.

[0131] After generation, an initial planned path from the starting position to the ending position is extracted from the path expansion tree, resulting in a set of paths from point {x}. des1 ,x des2 ,…,x desn The initial planned path is composed of [various parameters]. To make the path smoother, a Bézier curve is used to smooth the initial planned path, resulting in the final planned path.

[0132] Reference Figure 4 As shown, Figure 4 This is a flowchart illustrating the steps of a control method for a non-contact operation task of a magnetically controlled microrobot provided in an embodiment of this application. Figure 4 As shown, another control method for non-contact operation tasks of a magnetically controlled microrobot provided in this application embodiment may include steps S410 to S470:

[0133] Step S410: Perform path planning based on the starting and ending positions of the target object to obtain the planned path.

[0134] Specifically, the nearest point is determined using a turning angle distance metric. The starting position is used as the root node, and a path expansion tree is generated iteratively by adding leaf nodes through random sampling towards the destination position. The turning angle distance metric is determined based on the Euclidean distance between the two points and the turning angle. Then, an initial planned path from the starting position to the destination position is extracted from the path expansion tree, and the initial planned path is smoothed to obtain the planned path.

[0135] Step S420: Obtain the current position status, which includes: the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object.

[0136] Step S430: Calculate the state error value between the current position state and the corresponding predicted position state.

[0137] In this embodiment, if the motion model does not match the actual environment, the predicted position state obtained from the motion model will be inaccurate; that is, there is a state error value between the current position state (i.e., the actual position state) and the predicted position state. Therefore, in order to compensate for the error value of dynamic modeling, the motion model needs to be dynamically updated online.

[0138] Step S440: Update the model parameters of the motion model online according to the state error value to obtain the updated motion model.

[0139] In this embodiment, the motion model is constructed through a parameter determination network. Updating the model parameters of the motion model online based on the state error value means updating the weight values ​​of the parameter determination network based on the state error value to compensate for the error value of dynamic modeling, thereby enabling the motion model to adapt to changes in object and environmental conditions.

[0140] Specifically, updating the model parameters of the motion model online based on the state error value to obtain an updated motion model includes: updating the weight values ​​of the parameter determination network based on the state error value to obtain an updated parameter determination network; obtaining an updated diagonal matrix generated by directional decoupling based on the normalized relative Euclidean distance and the updated parameter determination network; and updating the model parameters of the motion model online based on the updated diagonal matrix generated by directional decoupling to obtain an updated motion model.

[0141] For example, updating the weights of the parameter-determined network based on the state error value is represented as follows:

[0142]

[0143]

[0144] in, λ represents the derivative of the parameters that determine the network weight values. k The weighting coefficients represent the relative distances, τ1 represents the weighting coefficients for state errors, τ2 represents the weighting coefficients for model errors, and e k,j The model error is represented by Δx. k,j R represents the state error value. i,j This represents the rotation matrix.

[0145] Step S450: The updated motion model predicts the position state based on the current position state to obtain the predicted position state at the next moment.

[0146] Step S460: With maintaining a safe distance as a constraint and minimizing the error between the ideal position state and the predicted position state as the objective, generate magnetic field control parameters. The safe distance refers to the target distance for maintaining non-operational contact operations. The magnetic field control parameters include: magnetic field direction and magnetic field frequency.

[0147] Step S470: Control the magnetic field according to the magnetic field control parameters to drive the magnetically controlled microrobot to push the target object along the planned path.

[0148] In this embodiment, a non-contact operation strategy for a micro-magnetically controlled robot is implemented based on motion model modeling control and path planning navigation. A decoupled affine nonlinear model (i.e., a motion module) is introduced to describe the dynamics of the fluid-force effect under external magnetic field input. A neural network model is used to effectively learn the model from data and update the weights online to improve convergence speed, solving the problem of difficulty in obtaining the motion model due to the uncertainty of dynamic parameters. The predicted position state at the next moment is obtained based on the motion model, combined with constraints to maintain a safe distance to ensure non-contact operation. Path planning is performed by using a turning angle distance metric to determine the nearest point, ensuring navigation in clutter conditions. Thus, precise control of the target object is achieved in a non-contact manner, facilitating trajectory tracking and reaching specific positions without causing any physical damage.

[0149] To better illustrate the control method for non-contact operation tasks of the magnetically controlled microrobot provided in this application, the following experimental analysis is conducted on the control method for non-contact operation tasks of the magnetically controlled microrobot.

[0150] First, assess the impact of different motion models and training data amounts on modeling performance. For example... Figure 5 As shown, the motion model provided in this application embodiment utilizes decoupling in the forward and tangential directions, demonstrating better data efficiency compared to the non-decoupled model. With a training set of approximately 800, the relative modeling error is approximately 20%, which is on the same order of magnitude as Brownian motion, verifying the effectiveness of the motion model provided in this application embodiment.

[0151] Then, the impact of the two activation functions and the number of hidden layers of neurons on the modeling accuracy was compared, and the comparison results are shown in Table 1. Among them, multi-quadric is the activation function used to train the parameter determination network in the embodiment of this application. It can be seen that the prediction error is the lowest for 32 neurons, ensuring the applicability of the parameter determination model (such as RBFN) to real-time control.

[0152] Table 1. Effects of activation function and number of hidden neuron layers on accuracy.

[0153]

[0154] To verify the flexibility and effectiveness of the non-contact operation of the method in this application embodiment, a trajectory involving tracking the four letters "I", "C", "R", and "A" was designed, including straight line segments and curves with different curvatures. The controller's hyperparameters were set to α = 1.0, λ = 2.0, and u... max =1.0, s=2.25, and the hyperparameters of the online updater are τ1=τ2=0.1. For example... Figure 7 As shown, the controller can successfully track all four tracks while maintaining a non-contact state. The letter "C" has the smallest average tracking error, approximately 0.4 μm, while the largest error occurs at the non-differentiable angle of the letter "A," with a measured value of 7.3 μm.

[0155] Secondly, the optimal controller (i.e., the controller provided in this application) based on the motion module provided in the embodiments of this application was compared with the model-free proportional (P-) controller (i.e., the P controller). The actual motion trajectory is as follows: Figure 6 As shown, the controller provided in this application can successfully complete this task, but the error of the P controller does not converge, which further causes the robot to come into contact with the object at [63.65μm, 83.61μm], resulting in adhesion and ultimately causing the task to fail.

[0156] Finally, the path planning method provided in this application embodiment is compared with the planner without smoothness and the random tree planning method to verify the effectiveness of navigation in chaotic scenes. The comparison results are shown in Table 2. The path planning method provided in this application embodiment generates a trajectory with the highest curvature, resulting in the minimum tracking error.

[0157] Table 2 Comparison results of different path planning methods

[0158] method Maximum curve (radians / pixel) Maximum tracking error (μm) The path planning method of this application 0.0084 2.654 Non-smooth planning methods 0.0532 5.320 Random tree planning method 0.0827 14.367

[0159] This application also provides a control system for non-contact operation tasks of a magnetically controlled microrobot, used to implement the control method for non-contact operation tasks of the magnetically controlled microrobot provided in this application embodiment, referring to... Figure 7 As shown, Figure 7 This is a schematic diagram of a control system for a non-contact operation task of a magnetically controlled microrobot provided in an embodiment of this application. The system includes:

[0160] Image acquisition equipment used to acquire microscopic images of magnetically controlled microrobots and target objects;

[0161] A computing processing device is used to identify the current position state based on the microscopic image and output magnetic field control parameters based on the current position state. The current position state includes: the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object. The magnetic field control parameters include: magnetic field direction and magnetic field frequency.

[0162] An electromagnetic device is used to generate a magnetic field according to magnetic field control parameters to drive the magnetically controlled microrobot to push the target object along a planned path.

[0163] In this embodiment, the electromagnetic device is a triaxial Helmholtz electromagnetic coil mounted on an inverted microscope (e.g., a Nikon Ti2-E microscope). An image acquisition device (e.g., a CMOS camera) is connected to the inverted microscope to acquire microscopic images of the magnetically controlled microrobot and the target object. A computational processing device uses a visual algorithm to identify the current position state of the microrobot and the target object from the microscopic images and outputs magnetic field control parameters based on the current position state. Specifically, the computational processing device outputs magnetic field control parameters based on the current position state by: predicting the position state at the next moment using a motion model constructed through a neural network model; thereby generating magnetic field control parameters with the goal of minimizing the error between the ideal position state and the predicted position state while maintaining a safe distance. The magnetic field control parameters are then transmitted to a servo amplifier to drive the electromagnetic device to generate a magnetic field.

[0164] The control system for non-contact operation tasks of the magnetically controlled microrobot provided in this application utilizes a motion model to predict the next position state based on the current position state, achieving real-time closed-loop control of the target object. By adjusting the magnetic field control parameters in real time, the magnetically controlled microrobot can achieve precise operation and tracking in different environments. Furthermore, the magnetic field control parameters are generated with the constraint of maintaining a safe distance and the goal of minimizing the error between the target position state and the predicted position state, achieving non-contact operation of the target object and avoiding damage caused by direct contact. Based on these magnetic field control parameters, precise actuation of the target object can be achieved. In addition, by using the magnetic field frequency and direction as magnetic field control parameters, changing the magnetic field direction and frequency allows the micro-magnetically controlled robot to move in any direction, thereby expanding the movement range of the micro-magnetically controlled robot and improving control flexibility. Thus, precise and flexible control of the magnetically controlled microrobot for non-contact operation tasks is achieved.

[0165] This application also provides a seismic common imaging point gather processing device based on a self-supervised optical flow field estimation network, referring to... Figure 9 As shown, Figure 9This is a schematic diagram of a seismic common imaging point gather processing device based on a self-supervised optical flow field estimation network, provided in an embodiment of this application. The device includes:

[0166] The acquisition module 910 is used to acquire the current position status, which includes: the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object;

[0167] The prediction module 920 is used for the motion model to predict the position state based on the current position state, and to obtain the predicted position state at the next moment. The motion model is constructed by a neural network model.

[0168] The generation module 930 is used to generate magnetic field control parameters with the constraint of maintaining a safe distance and the goal of minimizing the error between the ideal position state and the predicted position state. The safe distance refers to the target distance for maintaining non-operational contact operations. The magnetic field control parameters include: magnetic field direction and magnetic field frequency.

[0169] The control module 940 is used to control the magnetic field according to the magnetic field control parameters, so as to drive the magnetically controlled microrobot to push the target object to move along the planned path.

[0170] In an optional embodiment, the device further includes:

[0171] The error calculation module is used to calculate the state error value between the current position state and the corresponding predicted position state;

[0172] An online update module is used to update the model parameters of the motion model online based on the state error value, so as to obtain the updated motion model;

[0173] The state prediction module is used to predict the position state of the updated motion model based on the current position state, so as to obtain the predicted position state at the next moment.

[0174] In an optional embodiment, the apparatus further includes a model building module, the model building module comprising:

[0175] The first determining module is used to determine the position state definition and control input definition in the global coordinate system. The position state definition includes: the position of the target object and the relative distance between the magnetically controlled microrobot and the target object. The control input definition includes: the projected angular velocity on the first coordinate axis and the projected angular velocity on the second coordinate axis of the global coordinate system, and the magnetic field rotates in the plane formed by the first coordinate axis and the second coordinate axis.

[0176] The first construction submodule is used to construct an initial motion model based on the position state definition and the control input definition;

[0177] The model conversion module is used to determine the motion model parameters based on the neural network model, and convert the initial motion model to a local coordinate system to obtain the motion model. The local coordinate system takes the line connecting the center point of the magnetically controlled microrobot and the center point of the target object as the normal direction.

[0178] In one optional embodiment, the neural network model includes a parameter determination network; the model conversion module includes:

[0179] The normalization module is used to normalize the relative distance between the average radius of the magnetically controlled microrobot and the target object to obtain the normalized relative Euclidean distance.

[0180] The second construction submodule is used to construct a parameter determination network, which is trained using offline collected location state data and magnetic field parameter values ​​as training samples.

[0181] The diagonal matrix module is used to determine the network based on the normalized relative Euclidean distance and the parameters, and to obtain the diagonal matrix generated by directional decoupling.

[0182] The motion model module is used to obtain the motion model based on the rotation matrix from the global coordinate system to the local coordinate system, the diagonal matrix generated by the direction decoupling, and the control input definition.

[0183] In one optional embodiment, the online update module includes:

[0184] The first update submodule is used to update the weight values ​​of the parameter determination network according to the state error value, so as to obtain the updated parameter determination network.

[0185] The second update submodule is used to determine the network based on the normalized relative Euclidean distance and the updated parameters, and to obtain the updated diagonal matrix generated by directional decoupling.

[0186] The third update submodule is used to update the model parameters of the motion model online based on the diagonal matrix generated by the updated direction decoupling, so as to obtain the updated motion model.

[0187] In one optional embodiment, the generation module includes:

[0188] The third construction submodule is used to construct a cost function based on the ideal position state and the predicted position state, wherein the cost function has the optimization objective of minimizing the error between the ideal position state and the predicted position state.

[0189] The second determining module is used to determine the first constraint condition, which is used to constrain the maximum value of the control input;

[0190] The third determining module is used to determine the second constraint condition, which is used to constrain the minimum relative distance between the magnetically controlled microrobot and the target object.

[0191] The fourth determining module is used to determine the third constraint condition, which is used to constrain the maximum relative distance between the magnetically controlled microrobot and the target object;

[0192] The solution module is used to solve the cost function using the first constraint, the third constraint, and the third constraint as constraints to obtain the magnetic field control parameters.

[0193] In an optional embodiment, the apparatus further includes a path planning module, the path planning module comprising:

[0194] The position determination module is used to obtain the starting position and the ending position of the target object;

[0195] The iterative growth module is used to determine the nearest point by using the turning angle distance metric. Taking the starting position as the root node, the module iteratively generates the path expansion tree by adding leaf nodes through random sampling towards the ending position. The turning angle distance metric is determined based on the Euclidean distance between the two points and the turning angle.

[0196] The path extraction module is used to extract the initial planned path from the starting position to the ending position from the path expansion tree, and to filter the initial planned path to obtain the planned path.

[0197] This application also provides an electronic device, see embodiments thereof. Figure 10 , Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. For example... Figure 10 As shown, the electronic device 1000 includes a memory 1010 and a processor 1020. The memory 1010 and the processor 1020 are connected via a bus for communication. The memory 1010 stores a computer program, which can run on the processor 1020 to implement the steps of the control method for the non-contact operation task of the magnetically controlled microrobot described in the embodiments of this application.

[0198] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the control method for non-contact operation tasks of the magnetically controlled microrobot described in this application.

[0199] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0200] This application describes embodiments of methods, systems, apparatuses, and devices according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0201] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0202] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0203] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0204] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0205] The control method and system for non-contact operation tasks of a magnetically controlled microrobot provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A control method for non-contact operation tasks of a magnetically controlled microrobot, characterized in that, The method includes: Obtain the current position status, which includes: the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object; The motion model predicts the position state based on the current position state to obtain the predicted position state at the next moment. The motion model is constructed using a neural network model. With the constraint of maintaining a safe distance and the objective of minimizing the error between the ideal position state and the predicted position state, magnetic field control parameters are generated. The safe distance refers to the target distance for maintaining non-operational contact operations. The magnetic field control parameters include: magnetic field direction and magnetic field frequency. The magnetic field is controlled according to the magnetic field control parameters to drive the magnetically controlled microrobot to push the target object along the planned path; The motion model is constructed using a neural network model according to the following steps: The position state definition and control input definition in the global coordinate system are determined. The position state definition includes: the position of the target object and the relative distance between the magnetically controlled microrobot and the target object. The control input definition includes: the projected angular velocity on the first coordinate axis and the projected angular velocity on the second coordinate axis of the global coordinate system. The magnetic field rotates in the plane formed by the first coordinate axis and the second coordinate axis. Based on the defined position state and the defined control input, an initial motion model is constructed; The motion model parameters are determined based on the neural network model, and the initial motion model is transformed to a local coordinate system to obtain the motion model. This includes: normalizing the relative distance between the average radii of the magnetically controlled microrobot and the target object to obtain a normalized relative Euclidean distance; constructing a parameter determination network, which is trained using offline collected position state data and magnetic field parameter values ​​as training samples; obtaining a diagonal matrix generated by directional decoupling based on the normalized relative Euclidean distance and the parameter determination network; and obtaining the motion model based on the rotation matrix from the global coordinate system to the local coordinate system, the diagonal matrix generated by directional decoupling, and the defined control input. The local coordinate system uses the line connecting the center point of the magnetically controlled microrobot and the center point of the target object as its normal direction.

2. The method according to claim 1, characterized in that, The method further includes: Calculate the state error value between the current position state and the corresponding predicted position state; The model parameters of the motion model are updated online based on the state error value to obtain the updated motion model; The motion model predicts the position state based on the current position state to obtain the predicted position state at the next moment, including: The updated motion model predicts the position state based on the current position state to obtain the predicted position state at the next moment.

3. The method according to claim 2, characterized in that, The motion model parameters are updated online based on the state error value to obtain the updated motion model, including: The weight values ​​of the parameter determination network are updated based on the state error value to obtain the updated parameter determination network; The network is determined based on the normalized relative Euclidean distance and the updated parameters, and the updated diagonal matrix generated by directional decoupling is obtained. Based on the diagonal matrix generated by the updated directional decoupling, the model parameters of the motion model are updated online to obtain the updated motion model.

4. The method according to claim 1, characterized in that, With maintaining a safe distance as a constraint and minimizing the error between the ideal position state and the predicted position state as the objective, magnetic field control parameters are generated, including: Based on the ideal position state and the predicted position state, a cost function is constructed, the cost function having the optimization objective of minimizing the error between the ideal position state and the predicted position state; Determine the first constraint condition, which is used to constrain the maximum value of the control input; A second constraint condition is determined, which is used to constrain the minimum relative distance between the magnetically controlled microrobot and the target object; A third constraint condition is determined, which is used to constrain the maximum relative distance between the magnetically controlled microrobot and the target object; The cost function is solved using the first constraint, the third constraint, and the third constraint as constraints to obtain the magnetic field control parameters.

5. The method according to claim 1, characterized in that, The planned path is obtained through the following steps: Obtain the starting position and ending position of the target object; The nearest point is determined by using a turning angle distance metric. The starting position is used as the root node, and the path is iteratively generated towards the ending position by randomly sampling and adding leaf nodes to obtain a path expansion tree. The turning angle distance metric is determined based on the Euclidean distance between the two points and the turning angle. The initial planned path from the starting position to the ending position is extracted from the path expansion tree, and the initial planned path is smoothed to obtain the planned path.

6. A control system for non-contact operation tasks of a magnetically controlled microrobot, characterized in that, A control method for performing non-contact operation tasks of the magnetically controlled microrobot according to any one of claims 1-5, the system comprising: Image acquisition equipment used to acquire microscopic images of magnetically controlled microrobots and target objects; A computing processing device is used to identify the current position state based on the microscopic image and output magnetic field control parameters based on the current position state. The current position state includes: the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object. The magnetic field control parameters include: magnetic field direction and magnetic field frequency. An electromagnetic device is used to generate a magnetic field according to magnetic field control parameters to drive the magnetically controlled microrobot to move the target object along a planned path.

7. A control device for non-contact operation tasks of a magnetically controlled microrobot, characterized in that, The device includes: The acquisition module is used to acquire the current position status, which includes: the current position of the target object and the relative distance between the magnetically controlled microrobot and the target object; The prediction module is used by the motion model to predict the position state of the next moment based on the current position state. The motion model is constructed by a neural network model. The generation module is used to generate magnetic field control parameters with the constraint of maintaining a safe distance and the goal of minimizing the error between the ideal position state and the predicted position state. The safe distance refers to the target distance for maintaining non-operational contact operations. The magnetic field control parameters include: magnetic field direction and magnetic field frequency. The control module is used to control the magnetic field according to the magnetic field control parameters, so as to drive the magnetically controlled microrobot to push the target object to move along the planned path; The motion model is constructed using a neural network model according to the following steps: The position state definition and control input definition in the global coordinate system are determined. The position state definition includes: the position of the target object and the relative distance between the magnetically controlled microrobot and the target object. The control input definition includes: the projected angular velocity on the first coordinate axis and the projected angular velocity on the second coordinate axis of the global coordinate system. The magnetic field rotates in the plane formed by the first coordinate axis and the second coordinate axis. Based on the defined position state and the defined control input, an initial motion model is constructed; The motion model parameters are determined based on the neural network model, and the initial motion model is transformed to a local coordinate system to obtain the motion model. This includes: normalizing the relative distance between the average radii of the magnetically controlled microrobot and the target object to obtain a normalized relative Euclidean distance; constructing a parameter determination network, which is trained using offline collected position state data and magnetic field parameter values ​​as training samples; obtaining a diagonal matrix generated by directional decoupling based on the normalized relative Euclidean distance and the parameter determination network; and obtaining the motion model based on the rotation matrix from the global coordinate system to the local coordinate system, the diagonal matrix generated by directional decoupling, and the defined control input. The local coordinate system uses the line connecting the center point of the magnetically controlled microrobot and the center point of the target object as its normal direction.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the control method for non-contact operation tasks of the magnetically controlled microrobot according to any one of claims 1-5.

Images