Magnetic micro robot dynamic path planning method based on composite artificial potential field

Abstract

The invention discloses a magnetic micro-robot dynamic path planning method based on a composite artificial potential field, and relates to the technical field of micro-nano operation, and the method comprises the steps: inputting map information to generate an initial path, optimizing path points in each path segment, and outputting a continuous and smooth global path; based on the position relation between the robot position and the global path, composite path adsorption attraction is applied; when the algorithm falls into a local minimum, constructing a disturbance sampling area based on the position of the robot, a target point and a global path, randomly generating a temporary gravitational point in the area, applying auxiliary disturbance gravitational force, and guiding the robot to jump out of a trap area; and the target point gravitation, the obstacle repulsive force, the composite path adsorption gravitation and the auxiliary disturbance gravitation are synthesized into control guiding force, and the robot is driven to move in the set target direction. The method integrates path guidance, dynamic obstacle avoidance and local optimization capabilities, and is suitable for path planning tasks of various intelligent moving bodies in a complex environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of micro-nano operation, and particularly relates to a dynamic path planning method for a magnetic micro robot based on a composite artificial potential field. BACKGROUND

[0002] In recent years, with the rapid development of Magnetically Driven Microrobots in the biomedical field, the challenges faced by path planning are further intensified. Such micro robots are usually in the order of microns to millimeters in size, can be driven by an external magnetic field, and can achieve precise control and targeted transport in narrow and tortuous complex three-dimensional environments such as blood vessels and digestive tracts. Compared with traditional mobile platforms, Magnetically Driven Microrobots need to overcome the following difficulties at the same time: (1) the space channel is extremely narrow, the environmental geometric constraints are significant, and local minima are easy to occur; (2) fluid dynamics have a significant effect, and path planning needs to consider real-time and high robustness; (3) limited by computing resources and external magnetic control feedback, the algorithm complexity needs to be kept at a low level.

[0003] The computing efficiency of the sampling type planning path method is highly dependent on the sampling efficiency, and a large number of unnecessary nodes will exist in the sampling process, thereby affecting the planning rate; at the same time, since the path is generated by connecting nodes, the path is not smooth and contains a large number of sawteeth and corners; in addition, such algorithms often fail to search for a path or detour in narrow channels and multi-obstacle environments, and are lacking in global vision. The path planning method based on the traditional Artificial Potential Field (APF) algorithm has limited adaptability in such complex micro-environments, lacks global guidance ability, and relies too much on local information, which easily causes the robot to stagnate or oscillate at the blood vessel branch, resulting in path following failure, and if there is an obstacle near the terminal point, the target will be unreachable due to the repulsive force.

[0004] Therefore, for the path planning problem of Magnetically Driven Microrobots, how to design a lightweight algorithm that has global guidance and takes into account local obstacle avoidance and real-time computing ability has become one of the key technologies to be solved in this field. This not only has important significance for improving the task success rate of Magnetically Driven Microrobots in blood vessel targeted transport, local drug delivery, minimally invasive intervention and other applications, but also provides important technical support for promoting the landing application of intelligent micro systems in precision medicine. SUMMARY

[0005] The present application relates to the technical field of micro-nano operation, and particularly relates to a dynamic path planning method for a magnetic micro robot based on a composite artificial potential field.

[0006] (1) A guided path construction mechanism is proposed: by continuously parameterizing and adjusting the node positions within the path segment, the path can be automatically generated and optimized within the feasible region. This mechanism avoids the problems of multiple bends and discontinuities caused by node splicing in traditional sampling methods, and significantly improves the smoothness, compactness and executability of the path.

[0007] (2) Integrating a static path gravity mechanism: A global path guiding force field is introduced into the traditional artificial potential field framework, so that the robot will be simultaneously subjected to composite gravity control from the target point, obstacles, and the global path direction. This mechanism can achieve stable following of the planned path under dynamic obstacle interference, thus taking into account both obstacle avoidance and trajectory maintenance capabilities.

[0008] (3) An elliptical-assisted gravity mechanism is proposed to jump out of the local minimum: When the robot is trapped in a local minimum or "target unreachable" state, an auxiliary elliptical region with the current position and the target point as the focus is constructed, and temporary gravity points are randomly sampled in the region to apply perturbation gravity, guiding the robot to jump out of the dead zone, which significantly enhances the robustness and local adaptive ability of the algorithm.

[0009] (4) Retaining the computational efficiency advantage of the classical artificial potential field model: The computational workload is mainly concentrated on the planning of the initial global path. Real-time obstacle avoidance retains the fast response speed and simple computational structure of the artificial potential field method. It also makes up for its shortcomings such as insufficient globality and weak escape ability through guided path construction and auxiliary gravity mechanism, so that the system still has high reliability in complex obstacles and dynamic environments.

[0010] In summary, this invention integrates path guidance, dynamic obstacle avoidance, and local optimization capabilities, making it suitable for path planning tasks of various intelligent mobile entities in complex environments.

[0011] Compared with the prior art, the beneficial technical effects of the present invention are:

[0012] 1. The path is continuously adjustable, resulting in smoother execution.

[0013] Traditional path planning algorithms often use discrete node connections, resulting in jagged or zigzag paths that are difficult to execute directly. This invention introduces a guided path construction mechanism, which parameterizes the positions within path segments and uses optimization algorithms for continuous adjustment. This makes the generated path differentiable and adjustable, significantly improving path smoothness and execution stability. It is particularly suitable for scenarios with high requirements for trajectory continuity, such as differential drives and automotive-grade robots.

[0014] 2. Maintain path guidance to improve dynamic obstacle avoidance stability.

[0015] Compared to traditional artificial potential fields that rely solely on the attraction of the target point, this invention designs path adsorption and direction-following gravity by calculating the robot's projection point and tangent direction on the current path segment. When faced with dynamic obstacle interference, the robot can still flexibly adjust around the global path, achieving a balance between path following and obstacle avoidance capabilities, and preventing path deviation or loss of control.

[0016] 3. Possesses the ability to resist local minima and exhibits strong robustness.

[0017] Traditional artificial potential field methods are prone to getting trapped in local minima, making the target unreachable. This invention proposes an elliptical-assisted gravity mechanism, which constructs a perturbation sampling region based on the robot's current position and the target point position. When the robot falls into oscillation or stagnation, the auxiliary gravity is automatically activated to break the force field equilibrium and guide it out of the dead zone, greatly improving the robustness of path planning and the ability to recover from failures.

[0018] 4. Integrating multi-source gravity enables more flexible decision-making.

[0019] By fusing four types of force fields—target attraction, obstacle repulsion, composite path attraction, and auxiliary disturbance attraction—an adaptive dynamic guidance field is constructed. This field supports real-time adjustment of the robot's control direction, has excellent online response capabilities, and adapts to environmental changes more quickly, making it particularly suitable for robot navigation tasks in dynamic indoor and outdoor environments. Attached Figure Description

[0020] Figure 1 This is a simplified schematic diagram of the dynamic path planning method for magnetic microrobots based on a composite artificial potential field provided in this application.

[0021] Figure 2 This is a flowchart of the path parameter sequence optimization provided in this application.

[0022] Figure 3 This is a schematic diagram of a scenario where auxiliary perturbation gravity is applied under the elliptical-assisted gravity mechanism provided in this application.

[0023] Figure 4 This is a schematic diagram of the synthesis of control and guiding forces under a composite artificial potential field provided in this application. Detailed Implementation

[0024] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0025] One embodiment of this application provides a dynamic path planning method for a magnetic microrobot based on a composite artificial potential field, the architecture and flowchart of which are shown below. Figure 1As shown. This method addresses the autonomous navigation needs of magnetically driven microrobots in complex environments such as blood vessels. It integrates global path generation, dynamic obstacle avoidance, gravitational field control, and local escape mechanisms to achieve unified modeling and execution of high-quality path planning and real-time obstacle avoidance in complex environments. The method includes the following:

[0026] First, the environment is mapped and task parameters are set, inputting the starting point, target point, and obstacle map information. Then, the guided path construction module is entered. This module uses graph structure and heuristic search to generate an initial path, and smooths and optimizes the feasibility of the path points within each path segment, outputting a continuous and controllable global path as the reference for subsequent guidance.

[0027] A composite artificial potential field control module is designed, which includes three types of gravity source designs: first, a target point gravity field to provide global convergence; second, an obstacle repulsion field to achieve obstacle avoidance capability; and third, a path attraction gravity field. Based on the positional relationship between the current position of the magnetic microrobot and the global path, a composite path attraction gravity is applied to the magnetic microrobot to guide it to flexibly follow the global path, maintain trajectory following stability, and effectively balance obstacle avoidance freedom and path accuracy.

[0028] To address the issue of traditional artificial potential field methods easily getting trapped in local minima in complex obstacle environments, an ellipse-assisted gravity intervention module was designed. In this module, when the robot's movement is triggered into a local minimum condition, a perturbation sampling region is constructed based on the current position of the magnetic microrobot, the target point position, and the global path. Temporary gravity points are randomly generated within this perturbation sampling region, and auxiliary perturbation gravity is applied to guide the robot out of the trap region and back into the dominant path direction, thus enhancing the system's resistance to local minima.

[0029] Finally, the system superimposes the aforementioned gravitational field vectors, combining the gravitational force of the target point on the robot, the repulsive force of the obstacle on the robot, the attraction force of the composite path, and the auxiliary disturbance gravity to form a control and guiding force, driving the robot to move towards the predetermined target direction. During continuous operation, the system refreshes obstacle information in real time and iteratively updates the composite path attraction force and the auxiliary elliptical sampling state to ensure that the system has good dynamic adaptability, path-keeping ability, and stability control.

[0030] In summary, this application's overall solution establishes a path planning method with continuous executability, dynamic obstacle avoidance robustness, and local obstacle escape capability through a three-layer linkage mechanism of path construction, gravity fusion, and obstacle escape control. This method is applicable to high-precision navigation and autonomous path planning tasks for service robots, delivery robots, inspection robots, and other autonomous mobile platforms in complex environments. The specific content of each module is described in detail below:

[0031] (1) Guided path construction module

[0032] This module is used to generate a smooth and continuous initial reference global path in advance based on a static map, serving as the main thread for the entire robot path planning. This module mainly includes a path topology construction unit, an initial path generation unit, and a parameter optimization unit. The functions of each unit are as follows:

[0033] (a) Path topology building unit:

[0034] This unit is used to find scattered passable paths on a map for subsequent path search and optimization. It primarily constructs a MAKLINK topology graph based on map information of the starting point, target point, and obstacles. Specifically, it takes a 2D map containing obstacles as input, extracts obstacle boundaries and vertex information, and uses this information to randomly generate paths between obstacles that do not pass through them (i.e., paths the robot can pass through), forming a MAKLINK feasible connected graph. Finally, using the midpoint of each line segment as a graph node and the line segment as an edge, it generates the basis for a feasible path search graph.

[0035] (b) Initial path generation unit:

[0036] This unit uses the heuristic search algorithm Dijkstra on the MAKLINK topology to find a feasible and relatively short initial path from the starting point to the target point. Specifically, the searched path is a sequence of points in the middle of a selected line segment. This path serves as the initial solution for subsequent optimization and only possesses reachability.

[0037] (c) Parameter optimization unit

[0038] Since the initial solution does not perform well in terms of continuity and smoothness, an optimization unit is added to address these issues in the original path. This unit specifically optimizes the path points within each path segment, combining... Figure 2 As shown, the process includes the following:

[0039] (i) The positions of path points within each path segment are continuously parameterized, and the path parameter sequence is initialized. Specifically, the L in the initial path... i =[W i 1 W i 2 Transform into a continuous function:

[0040] W i (γ i ) = W i 1 +γ i ·(W i 2 -W i 1),γ i ∈[0,1]

[0041] Among them, L i This represents the i-th segment in the initial path, originating from the starting point W. i 1 and the endpoint W i 2 Define a line segment, starting from W. i 1 and the endpoint W i 2 It is the coordinate value, γ i These are path parameters used to control the path points within the corresponding path segment L. i The position above.

[0042] The position of each path point is represented as a positional parameter within a segment. Therefore, the entire path can be represented as a sequence of path parameters {γ1, γ2, ... γ...} d}, where d represents the number of selected line segments in the initial path. Therefore, by optimizing {γ1, γ2, ... γ d This allows us to obtain the optimized path point locations.

[0043] (ii) Define the optimization objective L total The weighted sum of path length and path smoothness is expressed as:

[0044] L total =L+λ1·Smoo

[0045]

[0046] Here, Smoo is a quantitative indicator of path smoothness, which measures the curvature change of the path through second-order differences, while W in the formula... i+1 (γ i+1 )-2W i (γ i )+W i-1 (γ i-1 ) represents the second-order difference of three adjacent path points, reflecting the local curvature of the path. The smaller the value, the smoother the path. L is the total path length of the selected d line segments in the initial path. λ1 is the regularization coefficient, which in this example represents the weight of balancing the path length and the smoothing term. The larger the value, the more the optimization result tends to be a smooth path, and vice versa. It should be set according to actual needs.

[0047] (iii) Based on the principle of biomimetic genetics, the path parameter sequence is iteratively updated (i.e., new strategies are generated using existing paths) to minimize the optimization objective, and heuristic guiding factors and multi-scale perturbation sampling strategies are introduced to guide parameter optimization.

[0048] The process of iteratively updating the path parameter sequence based on biomimetic genetic principles is well-known and will not be elaborated upon here. This example only provides the update formula for the path parameter sequence:

[0049]

[0050] in, These are the newly generated path parameter values. and ω is the selected parent path parameter value, ε is the random weight factor, and ε is a small random noise.

[0051] Since the algorithm records historical optimal solutions, the purpose of introducing a heuristic guiding factor is to guide the new path parameter sequence to approach the current historical optimal path parameter sequence, controlling the convergence of the search direction. Its parameter update formula is:

[0052]

[0053] in, This is the current historical optimal path parameter sequence. The generated path parameter values ​​for generation t are α, which serves as a step size factor to control the approach rate.

[0054] To make path search more exploratory and enable local fine-tuning and global jump search, a multi-scale perturbation sampling strategy is introduced when iterative optimization stalls. The parameter update formula is as follows:

[0055]

[0056] Where r represents the perturbation factor, ε is random noise following a Gaussian distribution, and σ is used to control the perturbation scale.

[0057] (iv) In each iteration, the current optimal path parameter sequence is saved to the optimal solution buffer to prevent the loss of high-quality solutions due to random perturbations. When the iteration termination condition is met, the optimal solution buffer outputs the optimal path parameter sequence and substitutes it into the continuous parameterized expression (i.e., the continuous function mentioned above), and finally a static, structurally continuous, and geometrically smooth global path can be obtained.

[0058] (2) Path attraction gravity generation module

[0059] To fully utilize existing paths, this invention, based on the target point attraction and obstacle repulsion of the traditional APF algorithm, also generates a composite path attraction force through this module, ensuring that the robot avoids obstacles around the static path. The specific implementation process of this module includes:

[0060] First, the current position p of the magnetic microrobot is acquired in real time. r(Generally refers to the position at the robot's center point), and find the distance p in the global path. r The nearest line segment [p] i ,p i+1 Next, calculate p. r The nearest projection point p on the line segment proj And calculate the line segment [p] i ,p i+1 The unit tangent vector of ] |||| indicates calculating the modulus of two coordinate points. Then along p r Point to p proj Directional generation path attraction And the path tangential guiding force generated based on the unit tangential vector Where, k p Let k be the first gravitational coefficient randomly generated within a certain range. p ~Uniform(k min ,k max λ2 is the path tangential guiding force coefficient, used to control the degree of guidance along the path direction; the larger the value, the more it tends to move forward along the line. Finally, the composite path adsorption attraction is output. The introduction of these two new gravitational forces enables the robot to flexibly follow the path, retaining a certain degree of freedom in obstacle avoidance without deviating from the global path.

[0061] (3) Ellipse-assisted gravity intervention module

[0062] Combination Figure 3 As shown, the specific implementation process of this module includes: firstly, setting the condition that the robot's movement gets stuck in a local minimum. In this embodiment, when the combined force of the attraction of the target point to the robot, the repulsive force of the obstacle to the robot, and the attraction of the combined path is detected... When the displacement is approximately zero (the robot's displacement is almost stationary) and the robot's position does not change significantly over multiple cycles (the robot is in an oscillating state), the condition for the robot to be trapped in a local minimum is met, and the intervention module is activated. After entering the intervention module, the current position p of the magnetic microrobot is used as the starting point. r The target point is the focus of the ellipse, and the distance along the global path (i.e., the distance p along the global path) is the remaining distance along the global path. r Let the major axis be a, and the minor axis be b, with the Euclidean distance from the current position to the target point as the Euclidean distance. Construct an auxiliary elliptical region as the sampling region. Randomly sample a temporary gravity point p within this auxiliary elliptical region. aux Along p r Point to p aux Direction-generating auxiliary perturbation gravity To escape the constraints of local minima. Where kaux The second gravitational coefficient k is randomly generated within a certain range. aux ~Uniform(k min ,k max ).

[0063] Optionally, the first and second gravity coefficients can be manually adjusted for different environments. For example, when the map is large and obstacles are sparse, these coefficients can be adjusted to be larger, which will result in better training time and effectiveness.

[0064] Based on the same inventive concept, another embodiment of this application provides a dynamic path planning system for magnetic microrobots based on a composite artificial potential field, comprising four core functional modules: a guided path construction module, a path attraction gravity generation module, an ellipse-assisted gravity intervention module, and a composite artificial potential field control module. These modules operate collaboratively through explicit data flow and control logic, achieving a unified approach to static path optimization and dynamic obstacle avoidance control. Their functional relationships are as follows:

[0065] (1) Guided path construction module

[0066] Inputting 2D obstacle environment map information and start-point and target-point configuration parameters provided by the map building unit, a MAKLINK topology graph structure is constructed. Dijkstra's heuristic graph search is used to generate polyline paths, and each segment of the path is parameterized within the segment. A path optimization mechanism (guided path construction) is employed to optimize the continuous path parameter sequence, ultimately outputting a structurally continuous and geometrically smooth global path sequence p = {W1(γ1),...,W...}. d (γ d )}, serving as the path reference benchmark for subsequent gravity control modules.

[0067] (2) Path attraction gravity generation module

[0068] Input the current global path p from the guided path construction module, and the robot's current real-time position p. r Dynamically obtain the robot's nearest line segment on the path [p i ,p i+1 ], and calculate p r The nearest projection point p on the line segment proj and line segment tangent vector This generates path attraction pointing to the path. Directional guiding force pointing towards the target Finally, the composite path adsorption attraction is output. Used to guide robots back to a reasonable course when they deviate from the path.

[0069] (3) Ellipse-assisted gravity intervention module

[0070] Input the robot's current position p r Target point location p aoal And the remaining distance of the current global path. When the system detects... When the value is approximately zero and the robot's position does not change significantly over multiple cycles, it is determined to be trapped in a local minimum. In this case, a system is constructed with p... r and p goal The elliptical region is defined by the foci of the ellipse, the remaining path length as the major axis *a*, and the Euclidean distance from the current position to the target point as the minor axis *b*. Figure 3 As shown, temporary gravity points p are generated by random sampling within the elliptical region. aux And generate auxiliary perturbation gravity. Used for short-term disturbance control, breaking the potential field balance and guiding the robot to jump out of the trap area.

[0071] (4) Composite artificial potential field control module

[0072] like Figure 4 As shown, the input is the target gravity calculated from the target location. Obstacle repulsion force calculated by the obstacle detection module and composite path adsorption attraction and auxiliary perturbation gravity By weighting and combining the four types of gravity, the final control and guiding force vector is obtained. The robot's dynamics model is converted into control commands (such as linear velocity and angular velocity), and these commands are used to control the robot to perform path tracking and obstacle avoidance behaviors.

[0073] The above descriptions are merely preferred embodiments of this application, and the present invention is not limited to the above embodiments. It is understood that other improvements and variations directly derived or conceived by those skilled in the art without departing from the spirit and concept of the present invention should be considered to be included within the protection scope of the present invention.

Claims

1. A dynamic path planning method for a magnetic microrobot based on a composite artificial potential field, characterized in that, The method includes: Input the starting point, target point, and obstacle map information, generate an initial path using graph structure and heuristic search, optimize the path points within each path segment, and output a global path with continuous structure and geometric smoothness. Based on the positional relationship between the current position of the magnetic microrobot and the global path, a composite path adsorption force is applied to the magnetic microrobot to guide the robot to flexibly follow the global path. When the robot is triggered to fall into a local minimum condition, a perturbation sampling region is constructed based on the current position of the magnetic microrobot, the target point position, and the global path. Temporary gravity points are randomly generated within the perturbation sampling region, and auxiliary perturbation gravity is applied to guide the robot to jump out of the trap area. The robot is guided by a combination of the attraction of the target point to the robot, the repulsion of the obstacle to the robot, the attraction of the composite path, and the attraction of the auxiliary disturbance, which drives the robot to move toward the predetermined target direction.

2. The dynamic path planning method for magnetic microrobots based on composite artificial potential fields according to claim 1, characterized in that, The optimization of path points within each path segment includes: The positions of path points within each path segment are continuously parameterized, and a sequence of path parameters is initialized, wherein the path parameters are used to control the position of the path points within the corresponding path segment; Set the optimization objective as a weighted sum of path length and path smoothness; Based on the principle of biomimetic genetics, the path parameter sequence is iteratively updated to minimize the optimization objective, and a heuristic guiding factor is introduced to guide the new path parameter sequence to approach the current historical best path parameter sequence, thereby controlling the convergence of the search direction. In each iteration, the optimal path parameter sequence of the current generation is saved to the optimal solution cache. When the iteration termination condition is met, the optimal solution cache outputs the optimal path parameter sequence and substitutes it into the continuous parameterized expression to obtain the global path.

3. The dynamic path planning method for magnetic microrobots based on composite artificial potential fields according to claim 2, characterized in that, Within each path segment Path point location W i (γ i The formula for continuous parameterization is: W i (c i )=W i 1 +g i ·(W i 2 -W i 1 ),c i ∈[0,1] Among them, L i This represents the i-th segment in the initial path, which originates from the starting point W. i 1 and the endpoint W i 2 A line segment is defined, wherein the starting point W i 1 and the endpoint W i 2 It is the coordinate value, γ i The path parameters are as described.

4. The dynamic path planning method for magnetic microrobots based on composite artificial potential fields according to claim 2, characterized in that, Optimization objective L total Represented as: L total =L+λ1·Smoo Where Smoo is the quantitative index of path smoothness, L is the total path length of the selected d line segments in the initial path, and W i (γ i ) is the parameterized representation of the path point position in the i-th segment, and λ1 is the regularization coefficient.

5. The dynamic path planning method for magnetic microrobots based on composite artificial potential fields according to claim 2, characterized in that, Based on the principle of biomimetic genetics, the formula for iteratively updating the path parameter sequence is as follows: in, These are the newly generated path parameter values. and ω is the selected parent path parameter value, ε is the random weight factor, and ε is the small random noise. The parameter update formula for introducing the heuristic guiding factor is as follows: in, The current historical optimal path parameter sequence, The generated path parameter values ​​for generation t are α, which serves as a step size factor to control the approach rate.

6. The dynamic path planning method for magnetic microrobots based on composite artificial potential fields according to any one of claims 1-5, characterized in that, The optimization of path points within each path segment also includes: When iterative optimization stalls, the parameter update formula for introducing a multi-scale perturbation sampling strategy is: in, Let r represent the generated path parameter values ​​for generation t, r represent the perturbation factor, ε represent random noise following a Gaussian distribution, and σ be used to control the perturbation scale.

7. The dynamic path planning method for magnetic microrobots based on composite artificial potential fields according to claim 1, characterized in that, Based on the positional relationship between the current position of the magnetic microrobot and the global path, applying a composite path adsorption force to the magnetic microrobot includes: Real-time acquisition of the current position p of the magnetic microrobot r And find the distance from p in the global path. r The nearest line segment; Calculate p r The nearest projection point p on the line segment proj And calculate the unit tangent vector of the line segment. Along the p r Pointing to p proj Directional generation path attraction And the path tangential guiding force generated based on the unit tangential vector. Output the composite path adsorption attraction Where, k p λ1 is the first gravitational coefficient randomly generated within a certain range, and λ2 is the path tangential guiding force coefficient, used to control the degree of guidance along the path direction.

8. The dynamic path planning method for magnetic microrobots based on composite artificial potential fields according to claim 1, characterized in that, Based on the current position of the magnetic microrobot, the target point position, and the global path, a perturbation sampling region is constructed. Within this perturbation sampling region, temporary gravity points are randomly generated, and auxiliary perturbation gravity is applied, including: With the current position p of the magnetic microrobot r Using the target point as the foci of an ellipse, with the remaining global path distance as the major axis and the Euclidean distance from the current position to the target point as the minor axis, an auxiliary elliptical region is constructed as the perturbation sampling region; A temporary gravitational point p is randomly sampled within the auxiliary elliptical region. aux Along the p r Pointing to p aux The direction generates the auxiliary perturbation gravity. Where, k aux The second gravitational coefficient is randomly generated within a certain range.

9. The dynamic path planning method for magnetic microrobots based on composite artificial potential fields according to claim 1, characterized in that, The method of generating an initial path using graph structure and heuristic search includes: Based on the starting point, target point, and obstacle map information, construct the MAKLINK topology map structure; On the MAKLINK topology, a feasible and relatively short initial path is searched from the starting point to the target point using the heuristic search algorithm Dijkstra.

10. The dynamic path planning method for magnetic microrobots based on composite artificial potential fields according to claim 1, characterized in that, The condition under which the robot's movement gets trapped in a local minimum is: When the combined force of the attraction of the target point to the robot, the repulsion of the obstacle to the robot, and the attraction of the composite path is approximately zero, and the robot's position does not change significantly over multiple cycles, it is determined that the robot's movement is trapped in a local minimum condition.

Images