A micro-soft robot based on magnetic drive and a control method thereof

Abstract

This invention discloses a magnetically driven micro-soft robot and its control method. In this invention, the soft, magnetized claw of the micro-soft robot obeys a constructed bending deformation model under a uniform magnetic field. The micro-soft robot is placed within an electromagnetic coil, which is composed of a three-dimensional orthogonal assembly of an X-axis Helmholtz coil, a Y-axis Helmholtz coil, and a Z-axis Maxwell coil. The force and torque applied to the micro-soft robot by each Maxwell coil obeys the electromagnetic model. This invention controls the movement of the micro-soft robot within a container by adjusting the desired current in the electromagnetic coil, and controls the grasping and releasing of the soft, magnetized claw by adjusting the magnitude of the uniform magnetic field, thus completing cell grasping and transport operations. This invention achieves high-precision, automated cell manipulation at the microscale, reducing manual labor while improving the quality and efficiency of operations, thus promoting the efficiency of microscopic operations and scientific research.

Description

Technical Field

[0001] This invention relates to an electromagnetic drive device, specifically to a magnetically driven micro soft robot, a cell grasping and transport device, and a control method thereof, to achieve automated cell grasping and transport. Background Technology

[0002] Manipulating cells at the microscale is a crucial method for cell research, and it has been widely studied and applied in numerous fields such as life sciences, biomedical engineering, agricultural engineering, and chemical engineering. Engineering techniques for cell manipulation are mainly divided into robotic arm contact methods and remote control methods. Direct contact methods with robotic arms offer high precision and efficiency, but have limited degrees of freedom, hindering sensitive cell manipulation. Remote control methods can control multiple degrees of freedom of micro-soft robots, achieving dexterous control, but suffer from insufficient control precision and low automation levels. Furthermore, remote control-based manipulation relies heavily on the design of end-effectors; therefore, designing and controlling ingenious actuators remains a major challenge in cell manipulation. Summary of the Invention

[0003] To address the problems in the background art, this invention provides a cell grasping and transport device and its control method based on a magnetically driven micro soft robot. The entire cell manipulation process is carried out in an open container filled with liquid silicone oil. A magnetic field excited by an electric current is used as the power source to adjust the opening and closing of the magnetized claws at the end of the robot and control the position and posture of the micro soft robot, thereby achieving automated cell grasping and transport. This provides a feasible solution for high-precision and automated cell manipulation.

[0004] The technical solution adopted in this invention is as follows:

[0005] I. A magnetically driven micro soft robot

[0006] The micro soft robot includes a non-magnetic body and a soft magnetized claw; one end of the soft magnetized claw is fixedly connected to the non-magnetic body, and the other end of the soft magnetized claw serves as the grasping end.

[0007] The soft magnetized claw comprises two parallel and spaced magnetic plates. One end of each magnetic plate is fixedly connected to the non-magnetic body, and the other end of each magnetic plate serves as the gripping end. Each magnetic plate conforms to a constructed bending deformation model under a uniform magnetic field, and the formula for the bending deformation model is as follows:

[0008]

[0009] f(0)=0; f′(L)=0; s0∈[0, L]

[0010] Where s0 represents the length from the fixed end of the magnetic sheet to any position on the magnetic sheet; f(s0) represents the tangent angle of the magnetic sheet at length s0; f″(s0) represents the second derivative of the tangent angle f(s0) at length s0; A represents the cross-sectional area of ​​the magnetic sheet; B represents the magnitude of the uniform magnetic field; E represents the Young's modulus of the magnetic sheet; I represents the moment of inertia of the magnetic sheet in the bending direction; L represents the length of the magnetic sheet; f(0) represents the tangent angle value at s0 = 0 on the magnetic sheet; and f′(L) represents the value of the first derivative of f(s0) at s0 = L on the magnetic sheet.

[0011] In the soft magnetized claw, the magnetization of the two magnetic plates is symmetrically distributed. One of the magnetic plates is placed in a uniform magnetic field and subjected to bending deformation analysis. The magnetic field distribution of its magnetization on the X-axis satisfies the following formula:

[0012]

[0013]

[0014] Where, m(s) s represents the magnetization vector of the current magnetic sheet, the magnitude of the magnetization per unit volume, and its length position along the length direction, respectively; m x m y m z These are the components of the magnetization vector m along the x, y, and z axes, respectively.

[0015] II. A Cell Grasping and Transport Device Based on a Magnetic Drive-Driven Micro Soft Robot

[0016] The device includes a miniature soft robot, a microscope camera, electromagnetic coils, a programmable DC power supply, a light source, a container, and a microscope stage. The electromagnetic coils include an X-axis Helmholtz coil, a Y-axis Helmholtz coil, and a Z-axis Maxwell coil. The X-axis and Y-axis Helmholtz coils are orthogonally arranged on the microscope stage, while the Z-axis Maxwell coil is placed vertically on the microscope stage. The container is placed in the center of the electromagnetic coils, and the miniature soft robot and cells are placed inside the container. The microscope camera and light source are placed directly above the container. The X-axis, Y-axis, and Z-axis Maxwell coils are each connected to their respective programmable DC power supplies.

[0017] The forces and torques exerted on the micro soft robot by the X-axis Helmholtz coil, Y-axis Helmholtz coil, and Z-axis Maxwell coil follow an electromagnetic model. The formula for the electromagnetic model corresponding to the forces and torques exerted on the micro soft robot by the Z-axis Maxwell coil is as follows:

[0018]

[0019] Among them, F m This represents the magnetic force generated by the Maxwell coil along the Z-axis. I represents the magnetic field gradient produced in the z-axis direction by a unit current applied to the Maxwell coil along the z-axis. z θ represents the magnitude of the current applied to the Maxwell coil along the Z-axis, M represents the magnitude of the magnetization vector M of the micro soft robot, and θ represents the direction of the magnetization vector M of the micro soft robot in the plane.

[0020] III. A Control Method for a Magnetic Drive-Based Micro Soft Robot

[0021] Step 1: Place the micro soft robot and cells into the container of the micromanipulation stage;

[0022] Step 2: Start the programmable DC power supply and apply initial current to the electromagnetic coil. After the micro soft robot stabilizes, use a microscope camera to acquire real-time container images and input them into the deep learning target detection model. The model identifies the initial position and initial orientation of the micro soft robot as well as the position information of the cells, and records them as the initial visual recognition position information.

[0023] Step 3: Based on the cell endpoint and initial visual recognition location information, plan a reference trajectory;

[0024] Step 4: Based on the electromagnetic model and corresponding kinematic model of the electromagnetic coil, combined with real-time visual recognition position information, the desired current of the programmable DC power supply is continuously generated and applied, thereby controlling the micro soft robot to move along the reference planned trajectory to the cell.

[0025] Step 5: Based on the bending deformation model of the soft magnetized claw, drive the soft magnetized claw to bend and grasp the cell, thereby controlling the micro soft robot to move along the reference planned trajectory to the cell endpoint and release the cell at the cell endpoint;

[0026] Step 6: Drive the micro soft robot backward and leave the cell.

[0027] In step 2, the deep learning object detection model is the YOLOv7 network model.

[0028] Step 4 specifically involves:

[0029] Step 4.1: Calculate the forces and torques applied to the micro soft robot using an electromagnetic model, and then calculate the velocity and angular velocity of the micro soft robot using a kinematic model, thereby obtaining the predicted position and orientation information of the micro soft robot;

[0030] Step 4.2: Input the real-time position and orientation information of the micro soft robot, as well as the predicted position and orientation information, into the extended Kalman filter algorithm to calculate and obtain the EKF fused position and orientation information with the highest confidence.

[0031] Step 4.3: Based on the EKF fusion of position and orientation information and the reference planned trajectory, the model predictive control algorithm combines the kinematic model and kinematic constraints of the micro soft robot to calculate the optimal control command;

[0032] Step 4.4: Based on the optimal control command, the desired magnetic force magnitude and magnetic field direction are solved according to the inverse kinematics model, and then the desired current is solved according to the inverse electromagnetic model. The programmable DC power supply is controlled to apply the desired current, thereby driving the micro soft robot to move.

[0033] Step 4.5: Repeat steps 4.1-4.4, continuously applying the desired current and controlling the micro soft robot to move along the reference planned trajectory to the cell.

[0034] The beneficial effects of this invention are:

[0035] This invention designs and builds a magnetic control platform, constructs a bending deformation model of a soft, magnetized claw, and designs a visual servoing algorithm, a position and orientation observer for the micro-soft robot, and a trajectory tracking control algorithm. This enables simultaneous control of the micro-soft robot's motion, cell grasping, and transport operations using three pairs of coils. The constructed bending deformation model of the soft, magnetized claw enables position-controlled cell grasping and release operations; and the accurate position and orientation observer design, along with a robust trajectory tracking controller, enables path tracking control of the micro-soft robot. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the micro soft robot and its container of the device of the present invention;

[0037] Figure 2 This is a schematic diagram of the structure of the device of the present invention;

[0038] Figure 3 This is a schematic diagram of the electromagnetic coil of the device of the present invention;

[0039] Figure 4 This is a control flowchart of the device of the present invention;

[0040] Figure 5 This is a diagram illustrating the circular trajectory tracking effect of the device of the present invention;

[0041] Figure 6 This is a diagram illustrating the square trajectory tracking effect of the device of the present invention;

[0042] Figure 7This is a diagram illustrating the effect of the device for grasping and transporting zebrafish embryonic cells according to the present invention.

[0043] In the diagram: 1. Magnetic sheet, 2. Soft magnetized claw, 3. Non-magnetic body, 4. Miniature soft robot, 5. Container, 6. X-axis Helmholtz coil, 7. Y-axis Helmholtz coil, 8. Z-axis Maxwell coil, 9. Computer, 10. Miniature soft robot operating table, 11. Light source, 12. Microscope camera, 13. Programmable DC power supply, 14. Electromagnetic coil, 15. Microscopic operating table. Detailed Implementation

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0045] like Figure 1 and Figure 2 As shown, the cell grasping and transport device includes a micro-soft robot 4, a microscope camera 12, an electromagnetic coil 14, a computer 9, a programmable DC power supply 13, a light source 11, a container 5, and a micromanipulation stage 15. The electromagnetic coil 14 consists of three pairs of coils, including an X-axis Helmholtz coil 6, a Y-axis Helmholtz coil 7, and a Z-axis Maxwell coil 8. The X-axis Helmholtz coil 6 and the Y-axis Helmholtz coil 7 are orthogonally arranged on the micromanipulation stage 15, while the Z-axis Maxwell coil 8 is vertically placed on the micromanipulation stage 15. The container 5 is placed in the center of the electromagnetic coil 14, and both the micro-soft robot 4 and the cells are placed inside the container 5, which contains liquid silicone oil. On the surface of the silicone oil, buoyancy and gravity cancel each other out. The microscope camera 12 and light source 11 are both placed directly above the container 5, with the lens of the microscope camera 12 facing the container 5. The illumination from the light source reduces visual interference. The microscope camera is used to observe the posture changes of the micro-soft robot 4 within the container. The microscope camera is fixed to the vertical rod of the microscope operating table and can be adjusted vertically to adjust the field of view. The microscope camera is equipped with a focusing knob to adjust the image clarity. The X-axis Helmholtz coil 6, Y-axis Helmholtz coil 7, and Z-axis Maxwell coil 8 are respectively connected to the corresponding programmable DC power supplies 13. The computer 9 is connected to the microscope camera 12 and the programmable DC power supplies 13 via data cables. The computer calculates the magnitude and direction of the output current using a constructed model and algorithm, and issues control commands to the three programmable DC power supplies according to the SCPI standard command language. After receiving the control commands, the programmable DC power supplies have an internal closed-loop controller that can quickly, accurately, and stably apply the target current to the three pairs of coils. The high-precision micro-operation device integrates the opening and closing control of the soft magnetized claw, the visual recognition of the position and orientation of the micro soft robot, and the optimal trajectory tracking control function.

[0046] The microscope camera is connected to the computer via a data cable to acquire visual information in real time. The computer uses the trained deep learning object detection model YOLOv7 to identify the micro soft robot 4. A single object detection box can only determine the position of the micro soft robot 4, but not its orientation. Therefore, during training, the body and soft magnetized claws of the micro soft robot 4 are labeled simultaneously to identify the position and orientation of the micro soft robot 4 and obtain the position and orientation information of the micro soft robot 4 through visual recognition.

[0047] The X-axis Helmholtz coil 6 and Y-axis Helmholtz coil 7 generate uniform magnetic fields in any direction on the plane. By generating magnetic fields in specific directions, magnetic torque is applied to the magnet, causing the micro soft robot 4 to rotate. By generating magnetic fields of different magnitudes, the bending degree of the soft claw is changed, causing the soft magnetized claw 2 of the micro soft robot 4 to close and open. The Z-axis Maxwell coil 8 generates a gradient magnetic field in a fixed direction. By generating a gradient magnetic field in a fixed direction, a driving force consistent with the magnetization direction is applied to the micro soft robot 4, causing the micro soft robot 4 to translate.

[0048] The forces and torques applied to the micro soft robot 4 by the X-axis Helmholtz coil 6, Y-axis Helmholtz coil 7, and Z-axis Maxwell coil 8 follow an electromagnetic model. The formula for the electromagnetic model corresponding to the forces and torques applied to the micro soft robot 4 by the Z-axis Maxwell coil 8 is as follows:

[0049]

[0050] Among them, F m This represents the magnetic force generated by the Maxwell coil 8 along the Z-axis. The gradient magnetic field generated by the Maxwell coil 8 along the Z-axis is represented by M, which represents the magnetization vector of the micro soft robot 4. I represents the magnetic field gradient produced in the z-axis direction by applying a unit current to the Maxwell coil 9 along the z-axis. z The formula shows that the direction of the electromagnetic force generated by the Z-axis Maxwell coil 8 on the planar micro soft robot 4 is always consistent with the magnetization direction of the micro soft robot 4, and its magnitude is proportional to the coil current.

[0051] The velocity and angular velocity of the micro soft robot 4 applied by the electromagnetic coil 14 follow a kinematic model:

[0052] F m =k d v

[0053] Among them, F mk represents the magnetic force generated by the Maxwell coil 8 along the Z-axis. d The coefficient representing the viscous drag is the proportionality coefficient to the velocity, and v represents the velocity of the micro soft robot 4. The kinematic model shows that the magnitude of the magnetic force on the micro soft robot 4 is proportional to the velocity of the micro soft robot 4.

[0054] like Figure 3 As shown, the micro soft robot 4 includes a non-magnetic body 3 and a soft magnetized claw 2; one end of the soft magnetized claw 2 is fixedly connected to the non-magnetic body 3, and the other end of the soft magnetized claw 2 serves as the grasping end.

[0055] The soft magnetized claw 2 includes two parallel and spaced magnetic plates 1. One end of each magnetic plate 1 is fixedly connected to the non-magnetic body 3, and the other end of each magnetic plate 1 serves as the gripping end. Each magnetic plate 1 obeys a constructed bending deformation model under a uniform magnetic field. The formula for the bending deformation model is as follows:

[0056]

[0057] f(0)=0; f′(L)=0; s0∈[0, L]

[0058] Wherein, s0 represents the length from the fixed end of magnetic sheet 1 to any position of magnetic sheet 1; f(s0) represents the tangent angle of magnetic sheet 1 at length s0; f″(s0) represents the second derivative of the tangent angle f(s0) of magnetic sheet 1 at length s0; A represents the cross-sectional area of ​​magnetic sheet 1; B represents the magnitude of the uniform magnetic field; E represents the Young's modulus of magnetic sheet 1; I represents the moment of inertia of magnetic sheet 1 in the bending direction; L represents the length of magnetic sheet 1; f(0) represents the tangent angle value at s0 = 0 (fixed end) of magnetic sheet; f′(L) represents the value of the first derivative of the tangent angle f(s0) of magnetic sheet 1 at length s0 at s0 = L (grabbing end) of magnetic sheet.

[0059] In specific implementation, the length L, width w, and height h of each magnetic sheet 1 are L = 10 mm, w = 0.5 mm, and h = 1 mm, respectively. The length direction of each magnetic sheet 1 is parallel to the X-axis, and the width direction of each magnetic sheet 1 is parallel to the Y-axis. That is, the plane formed by the length L and width w of each magnetic sheet 1 is located in the XY plane. Each magnetic sheet 1 is made by uniformly mixing Ecoflex-10 solution and NdFeB magnetic powder in a 1:1 mass ratio, then inverting it into a mold containing the required cuboid groove, and transferring it to an oven for heating and curing. The resulting cuboid sheet is recessed into a quarter circle along the length direction L on a cylindrical mold and magnetized in a 1.5T magnetic field. The two magnetized cuboid magnetic sheets 1 together constitute the soft magnetized claw 2 of the micro soft robot 4. In the initial position of the soft magnetized claw 2, the two magnetic sheets 1 are symmetrically installed about the X-axis.

[0060] In the soft magnetized claw 2, the magnetization of the two magnetic plates 1 is symmetrically distributed about the X-axis. One of the magnetic plates 1 (such as the magnetic plate 1 on the left) is placed in a uniform magnetic field and subjected to bending deformation analysis. The magnetic field distribution of its magnetization on the X-axis satisfies the following formula:

[0061]

[0062]

[0063] Where, m(s) s represents the magnetization vector of the current magnetic sheet 1, the magnitude of the magnetization per unit volume, and its length position along the length direction, respectively; m x m y m z These are the components of the magnetization vector m along the x, y, and z axes, respectively.

[0064] The magnetic sheet 1 is simplified into a cantilever beam model, and the break point of the magnetic sheet at s = s0 is selected. The bending moment of the remaining magnetic sheet 1 to the right of the break point is analyzed:

[0065]

[0066] in, This represents the bending moment at s = s0. A = wh represents the cross-sectional area of ​​magnetic sheet 1, and τ(s) represents the bending moment per unit volume, calculated as follows:

[0067]

[0068] Where R(s) and B represent the rotation matrix and the uniform magnetic field, respectively; B is the magnitude of the uniform magnetic field B. Assuming θ = f(s), then f(0) = 0 in the cantilever beam is obviously true. Furthermore, the internal shear force of the cantilever beam is negligible, so it follows the pure bending assumption:

[0069]

[0070] By integrating all the above formulas, a bending deformation model of the soft magnetized claw 2 can be constructed:

[0071]

[0072] f(0)=0; f′(L)=0; s0∈[0, L]

[0073] like Figure 4 As shown, the control method includes the following steps:

[0074] Step 1: Install the electromagnetic coil 14 on the micromanipulation stage 15. Adjust the position and focal length of the light source 11 and the microscope camera 12 to achieve the clearest image of the micro-soft robot 4 inside the container 5. Turn on the microscope camera 12 and publish the real-time image in the robot operating system ROS. Place the micro-soft robot 4 and the cells into the container 5 of the micromanipulation stage 15, with the positions of the micro-soft robot 4 and the cells set randomly.

[0075] Step 2: Start the programmable DC power supply 13 and apply initial current to the electromagnetic coil 14, and publish the voltage and current information in ROS in real time. After the micro soft robot 4 stabilizes, use the microscope camera 12 to acquire real-time container images and input them into the trained deep learning object detection model. The model identifies the initial position and initial orientation of the micro soft robot 4 as well as the position information of the cells, and records them as the initial visual recognition position information, and publishes them in ROS in real time. The deep learning object detection model is the YOLOv7 network model.

[0076] Step 3: Based on the set cell endpoint and initial visual recognition position information, use the Fast Extended Random Tree (RRT) algorithm and the Cubic Spline interpolation algorithm to quickly plan a smooth and feasible reference trajectory with the endpoint being the target cell position.

[0077] Step 4: Based on the electromagnetic model and corresponding kinematic model of electromagnetic coil 14, combined with real-time visual recognition position information, the desired current of programmable DC power supply 13 is continuously generated and applied, thereby controlling the micro soft robot 4 to move to the cell along the reference planned trajectory.

[0078] Step 4 is as follows:

[0079] Step 4.1: Collect the voltage and current information of the programmable DC power supply 13 in ROS, calculate the force and torque applied to the micro soft robot 4 through the constructed electromagnetic model, and then calculate the velocity and angular velocity of the micro soft robot 4 through the constructed kinematic model, thereby obtaining the predicted position and orientation information of the micro soft robot 4.

[0080] Step 4.2: Input the real-time position and orientation information (i.e., real-time visual recognition position information) of the micro soft robot 4 and the predicted position and orientation information into the extended Kalman filter algorithm EKF, calculate the EKF fused position and orientation information with the highest confidence, and publish it in ROS in real time;

[0081] Step 4.3: Based on the EKF fusion of position and orientation information and reference planned trajectory, the Model Predictive Control (MPC) algorithm, combined with the kinematic model and kinematic constraints of the micro soft robot 4, calculates the optimal control commands (i.e., optimal velocity and angular velocity) and publishes them in ROS.

[0082] Step 4.4: Receive the optimal control command calculated by MPC in ROS, and based on the optimal control command, solve the desired magnetic force magnitude and magnetic field direction according to the inverse kinematics model, and then solve the desired current according to the inverse electromagnetic model. Control the programmable DC power supply 13 to apply the desired current, thereby driving the micro soft robot 4 to move, and publish the voltage and current information in ROS in real time.

[0083] Step 4.5: Repeat steps 4.1-4.4, continuously applying the desired current and controlling the micro soft robot 4 to move along the reference planned trajectory to the cell.

[0084] Step 5: When the soft magnetized claw 2 of the micro soft robot 4 arrives at the position for grasping the target cell, according to the bending deformation model of the soft magnetized claw 2, the magnitude of the uniform magnetic field is controlled to drive the soft magnetized claw 2 to bend and grasp the cell. Then, the desired current is applied to the electromagnetic coil 14 by the programmable DC power supply 13 to control the micro soft robot 4 to move along the reference planned trajectory to the cell endpoint. At the cell endpoint, according to the bending deformation model of the soft magnetized claw 2, the magnitude of the uniform magnetic field is controlled to make the soft magnetized claw 2 release the cell.

[0085] Step 6: The programmable DC power supply 13 applies the desired current to the electromagnetic coil 14, driving the micro soft robot 4 to retreat and leave the cell.

[0086] The control method of this invention achieves automated cell grasping and transportation by controlling the electromagnetic coil 14 and the micro soft robot 4. This method simultaneously controls the grasping, releasing and transportation actions of the micro soft robot 4. The constructed bending deformation model is used for the opening and closing operation of the soft magnetized claw 2 to realize cell grasping and release. The EKF observer used obtains accurate position and orientation information of the micro soft robot 4 to realize information feedback. The MPC controller used obtains the optimal control command to realize the optimal trajectory tracking control function.

[0087] In the trajectory tracking experiment, the micro soft robot 4, under the control of PID and EKF-MPC methods (the control method of this invention), tracked a circular trajectory, and the trajectory tracking errors were as follows: Figure 5 (a) and Figure 5 As shown in (b);

[0088] The micro soft robot 4, under the control of PID and EKF-MPC methods (the control method of this invention), respectively tracks the trajectory of a square, and the trajectory tracking errors are as follows: Figure 6 (a) and Figure 6 As shown in (b);

[0089] like Figure 7 As shown in the specific embodiment, the micro soft robot 4, driven by the constructed model and controller, realizes the grasping and transporting of zebrafish embryonic cells with a diameter of 800μm. Figure 7 (a) to Figure 7 (b) is a schematic diagram of the micro soft robot 4 approaching the cell. Figure 7 (b) to Figure 7 (c) represents the soft magnetized claw 2 grasping the cell. Figure 7 (c) to Figure 7 (d) represents the transport of cells to the target location by the micro soft robot 4. Figure 7 (d) to Figure 7 (e) is the soft magnetized claw 2 releasing the cell, finally Figure 7 (e) to Figure 7 (f) represents the micro soft robot 4 retracting and leaving the cell.

[0090] Finally, it should be noted that the above embodiments and descriptions are only used to illustrate the technical solutions of the present invention and not to limit it. Those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the disclosure of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the protection scope of the claims of the present invention.

Claims

1. A magnetically driven micro soft robot, characterized in that, The micro soft robot (4) includes a non-magnetic body (3) and a soft magnetized claw (2); one end of the soft magnetized claw (2) is fixedly connected to the non-magnetic body (3), and the other end of the soft magnetized claw (2) serves as the grasping end; The soft magnetized claw (2) includes two parallel and spaced magnetic plates (1). One end of each magnetic plate (1) is fixedly connected to the non-magnetic body (3), and the other end of each magnetic plate (1) serves as the gripping end. Each magnetic plate (1) obeys a constructed bending deformation model under a uniform magnetic field. The formula of the bending deformation model is as follows: in, This represents the length from the fixed end of the magnetic sheet (1) to any position on the magnetic sheet (1); Indicates length The tangent angle of the magnetic thin sheet (1); Indicates length The angle of the tangent of the magnetic thin sheet (1) The second derivative function; This represents the cross-sectional area of ​​the magnetic sheet (1); Indicates the magnitude of a uniform magnetic field; The Young's modulus of the magnetic thin sheet (1) is represented; The moment of inertia of the magnetic sheet (1) in the bending direction is represented by the magnetic sheet (1). Indicates the length of the magnetic sheet (1); This indicates that in a magnetic thin film (1) The tangent angle value at that point; express The first derivative of the magnetic thin film (1) The value at; In the soft magnetized claw (2), the magnetization intensity of the two magnetic sheets (1) is symmetrically distributed. One of the magnetic sheets (1) is placed in a uniform magnetic field for bending deformation analysis. The magnetic field distribution of its magnetization intensity on the X-axis satisfies the following formula: in, These represent the magnetization vector of the current magnetic sheet (1), the magnitude of the magnetization per unit volume, and the length position in the length direction, respectively. , , These are the magnetization vectors. Components along the x, y, and z axes.

2. The cell grasping and transport device for a magnetically driven micro soft robot according to any one of claims 1, characterized in that, The system includes a micro soft robot (4), a microscope camera (12), an electromagnetic coil (14), a programmable DC power supply (13), a light source (11), a container (5), and a microscope stage (15). The electromagnetic coil (14) includes an X-axis Helmholtz coil (6), a Y-axis Helmholtz coil (7), and a Z-axis Maxwell coil (8). The X-axis Helmholtz coil (6) and the Y-axis Helmholtz coil (7) are orthogonally arranged on the microscope stage (15), and the Z-axis Maxwell coil (8) is placed vertically on the microscope stage (15). The container (5) is placed in the center of the electromagnetic coil (14). The micro soft robot (4) and the cells are placed inside the container (5). The microscope camera (12) and the light source (11) are placed directly above the container (5). The X-axis Helmholtz coil (6), the Y-axis Helmholtz coil (7), and the Z-axis Maxwell coil (8) are connected to the corresponding programmable DC power supply (13).

3. The cell grasping and transport device for a magnetically driven micro soft robot according to claim 2, characterized in that, The forces and torques exerted on the micro soft robot (4) by the X-axis Helmholtz coil (6), Y-axis Helmholtz coil (7), and Z-axis Maxwell coil (8) follow an electromagnetic model. The formula for the electromagnetic model corresponding to the forces and torques exerted on the micro soft robot (4) by the Z-axis Maxwell coil (8) is as follows: in, This represents the magnetic force generated by the Maxwell coil (8) along the Z-axis. This represents the magnetic field gradient produced in the z-axis direction by applying a unit current to the Maxwell coil (8) along the Z-axis. This indicates the magnitude of the current applied to the Maxwell coil (8) along the Z-axis. Represents the magnetization vector of the micro soft robot (4) Size, Represents the magnetization vector of the micro soft robot (4) Direction on a plane.

4. A control method for a magnetically driven micro soft robot according to any one of claims 1, characterized in that, Includes the following steps: Step 1: Place the micro soft robot (4) and the cell into the container (5) of the micromanipulation stage (15); Step 2: Start the programmable DC power supply (13) and apply initial current to the electromagnetic coil (14). After the micro soft robot (4) stabilizes, use the microscope camera (12) to collect real-time container images and input them into the deep learning target detection model. The model identifies the initial position and initial orientation of the micro soft robot (4) and the position information of the cells, and records them as the initial visual recognition position information. Step 3: Based on the cell endpoint and initial visual recognition location information, plan a reference trajectory; Step 4: Based on the electromagnetic model of the electromagnetic coil (14) and the corresponding kinematic model, combined with real-time visual recognition position information, the desired current of the programmable DC power supply (13) is continuously generated and applied, thereby controlling the micro soft robot (4) to move to the cell along the reference planned trajectory; Step 5: Based on the bending deformation model of the soft magnetized claw (2), drive the soft magnetized claw (2) to bend and grasp the cell, thereby controlling the micro soft robot (4) to move along the reference planned trajectory to the cell endpoint and release the cell at the cell endpoint; Step 6: Drive the micro soft robot (4) backward and leave the cell.

5. The control method for a magnetically driven micro soft robot according to claim 4, characterized in that, In step 2, the deep learning object detection model is the YOLOv7 network model.

6. The control method for a magnetically driven micro soft robot according to claim 4, characterized in that, Step 4 specifically involves: Step 4.1: Calculate the force and torque applied to the micro soft robot (4) through the electromagnetic model, and then calculate the velocity and angular velocity of the micro soft robot (4) through the kinematic model, so as to obtain the predicted position and orientation information of the micro soft robot (4); Step 4.2: Input the real-time position and orientation information of the micro soft robot (4) and the predicted position and orientation information into the extended Kalman filter algorithm to calculate the EKF fused position and orientation information with the highest confidence. Step 4.3: Based on the EKF fusion position and orientation information and the reference planned trajectory, the model predictive control algorithm combines the kinematic model and kinematic constraints of the micro soft robot (4) to calculate the optimal control command; Step 4.4: Based on the optimal control command, the desired magnetic force magnitude and magnetic field direction are solved according to the inverse kinematics model, and then the desired current is solved according to the inverse electromagnetic model. The programmable DC power supply (13) is controlled to apply the desired current, thereby driving the micro soft robot (4) to move. Step 4.5: Repeat steps 4.1-4.4, continuously apply the desired current and control the micro soft robot (4) to move along the reference planned trajectory to the cell.

Images